lavfi: switch to AVFrame.

Deprecate AVFilterBuffer/AVFilterBufferRef and everything related to it
and use AVFrame instead.

59 files changed, 1061 insertions, 942 deletions

diff --git a/doc/filters.texi b/doc/filters.texi
index a7d6a56d41..c7e2d9d184 100644
--- a/doc/filters.texi
+++ b/doc/filters.texi
@@ -622,9 +622,6 @@ same as @var{out_w} and @var{out_h}
 @item n
 the number of input frame, starting from 0
 
-@item pos
-the position in the file of the input frame, NAN if unknown
-
 @item t
 timestamp expressed in seconds, NAN if the input timestamp is unknown
 
@@ -1760,9 +1757,6 @@ the frame is bottom-field-first
 @item key
 1 if the filtered frame is a key-frame, 0 otherwise
 
-@item pos
-the position in the file of the filtered frame, -1 if the information
-is not available (e.g. for synthetic video)
 @end table
 
 The default value of the select expression is "1".
@@ -1854,10 +1848,6 @@ the PTS of the first video frame
 @item INTERLACED
 tell if the current frame is interlaced
 
-@item POS
-original position in the file of the frame, or undefined if undefined
-for the current frame
-
 @item PREV_INPTS
 previous input PTS
 
diff --git a/libavfilter/af_amix.c b/libavfilter/af_amix.c
index c2fb158031..5a4ec2e60e 100644
--- a/libavfilter/af_amix.c
+++ b/libavfilter/af_amix.c
@@ -275,18 +275,18 @@ static int output_frame(AVFilterLink *outlink, int nb_samples)
 {
     AVFilterContext *ctx = outlink->src;
     MixContext      *s = ctx->priv;
-    AVFilterBufferRef *out_buf, *in_buf;
+    AVFrame *out_buf, *in_buf;
     int i;
 
     calculate_scales(s, nb_samples);
 
-    out_buf = ff_get_audio_buffer(outlink, AV_PERM_WRITE, nb_samples);
+    out_buf = ff_get_audio_buffer(outlink, nb_samples);
     if (!out_buf)
         return AVERROR(ENOMEM);
 
-    in_buf = ff_get_audio_buffer(outlink, AV_PERM_WRITE, nb_samples);
+    in_buf = ff_get_audio_buffer(outlink, nb_samples);
     if (!in_buf) {
-        avfilter_unref_buffer(out_buf);
+        av_frame_free(&out_buf);
         return AVERROR(ENOMEM);
     }
 
@@ -308,7 +308,7 @@ static int output_frame(AVFilterLink *outlink, int nb_samples)
             }
         }
     }
-    avfilter_unref_buffer(in_buf);
+    av_frame_free(&in_buf);
 
     out_buf->pts = s->next_pts;
     if (s->next_pts != AV_NOPTS_VALUE)
@@ -455,7 +455,7 @@ static int request_frame(AVFilterLink *outlink)
     return output_frame(outlink, available_samples);
 }
 
-static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *buf)
+static int filter_frame(AVFilterLink *inlink, AVFrame *buf)
 {
     AVFilterContext  *ctx = inlink->dst;
     MixContext       *s = ctx->priv;
@@ -474,16 +474,16 @@ static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *buf)
     if (i == 0) {
         int64_t pts = av_rescale_q(buf->pts, inlink->time_base,
                                    outlink->time_base);
-        ret = frame_list_add_frame(s->frame_list, buf->audio->nb_samples, pts);
+        ret = frame_list_add_frame(s->frame_list, buf->nb_samples, pts);
         if (ret < 0)
             goto fail;
     }
 
     ret = av_audio_fifo_write(s->fifos[i], (void **)buf->extended_data,
-                              buf->audio->nb_samples);
+                              buf->nb_samples);
 
 fail:
-    avfilter_unref_buffer(buf);
+    av_frame_free(&buf);
 
     return ret;
 }
diff --git a/libavfilter/af_ashowinfo.c b/libavfilter/af_ashowinfo.c
index c8e830eee7..0855440529 100644
--- a/libavfilter/af_ashowinfo.c
+++ b/libavfilter/af_ashowinfo.c
@@ -65,16 +65,16 @@ static void uninit(AVFilterContext *ctx)
     av_freep(&s->plane_checksums);
 }
 
-static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *buf)
+static int filter_frame(AVFilterLink *inlink, AVFrame *buf)
 {
     AVFilterContext *ctx = inlink->dst;
     AShowInfoContext *s  = ctx->priv;
     char chlayout_str[128];
     uint32_t checksum = 0;
-    int channels    = av_get_channel_layout_nb_channels(buf->audio->channel_layout);
+    int channels    = av_get_channel_layout_nb_channels(buf->channel_layout);
     int planar      = av_sample_fmt_is_planar(buf->format);
     int block_align = av_get_bytes_per_sample(buf->format) * (planar ? 1 : channels);
-    int data_size   = buf->audio->nb_samples * block_align;
+    int data_size   = buf->nb_samples * block_align;
     int planes      = planar ? channels : 1;
     int i;
 
@@ -87,7 +87,7 @@ static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *buf)
     }
 
     av_get_channel_layout_string(chlayout_str, sizeof(chlayout_str), -1,
-                                 buf->audio->channel_layout);
+                                 buf->channel_layout);
 
     av_log(ctx, AV_LOG_INFO,
            "n:%"PRIu64" pts:%"PRId64" pts_time:%f "
@@ -95,7 +95,7 @@ static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *buf)
            "checksum:%08X ",
            s->frame, buf->pts, buf->pts * av_q2d(inlink->time_base),
            av_get_sample_fmt_name(buf->format), chlayout_str,
-           buf->audio->sample_rate, buf->audio->nb_samples,
+           buf->sample_rate, buf->nb_samples,
            checksum);
 
     av_log(ctx, AV_LOG_INFO, "plane_checksums: [ ");
@@ -114,7 +114,6 @@ static const AVFilterPad inputs[] = {
         .get_audio_buffer = ff_null_get_audio_buffer,
         .config_props     = config_input,
         .filter_frame     = filter_frame,
-        .min_perms        = AV_PERM_READ,
     },
     { NULL },
 };
diff --git a/libavfilter/af_asyncts.c b/libavfilter/af_asyncts.c
index 40680c8559..2e651ffca7 100644
--- a/libavfilter/af_asyncts.c
+++ b/libavfilter/af_asyncts.c
@@ -158,14 +158,13 @@ static int request_frame(AVFilterLink *link)
             handle_trimming(ctx);
 
         if (nb_samples = get_delay(s)) {
-            AVFilterBufferRef *buf = ff_get_audio_buffer(link, AV_PERM_WRITE,
-                                                         nb_samples);
+            AVFrame *buf = ff_get_audio_buffer(link, nb_samples);
             if (!buf)
                 return AVERROR(ENOMEM);
             ret = avresample_convert(s->avr, buf->extended_data,
                                      buf->linesize[0], nb_samples, NULL, 0, 0);
             if (ret <= 0) {
-                avfilter_unref_bufferp(&buf);
+                av_frame_free(&buf);
                 return (ret < 0) ? ret : AVERROR_EOF;
             }
 
@@ -177,20 +176,20 @@ static int request_frame(AVFilterLink *link)
     return ret;
 }
 
-static int write_to_fifo(ASyncContext *s, AVFilterBufferRef *buf)
+static int write_to_fifo(ASyncContext *s, AVFrame *buf)
 {
     int ret = avresample_convert(s->avr, NULL, 0, 0, buf->extended_data,
-                                 buf->linesize[0], buf->audio->nb_samples);
-    avfilter_unref_buffer(buf);
+                                 buf->linesize[0], buf->nb_samples);
+    av_frame_free(&buf);
     return ret;
 }
 
-static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *buf)
+static int filter_frame(AVFilterLink *inlink, AVFrame *buf)
 {
     AVFilterContext  *ctx = inlink->dst;
     ASyncContext       *s = ctx->priv;
     AVFilterLink *outlink = ctx->outputs[0];
-    int nb_channels = av_get_channel_layout_nb_channels(buf->audio->channel_layout);
+    int nb_channels = av_get_channel_layout_nb_channels(buf->channel_layout);
     int64_t pts = (buf->pts == AV_NOPTS_VALUE) ? buf->pts :
                   av_rescale_q(buf->pts, inlink->time_base, outlink->time_base);
     int out_size, ret;
@@ -229,8 +228,7 @@ static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *buf)
     }
 
     if (out_size > 0) {
-        AVFilterBufferRef *buf_out = ff_get_audio_buffer(outlink, AV_PERM_WRITE,
-                                                         out_size);
+        AVFrame *buf_out = ff_get_audio_buffer(outlink, out_size);
         if (!buf_out) {
             ret = AVERROR(ENOMEM);
             goto fail;
@@ -272,11 +270,11 @@ static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *buf)
 
     s->pts = pts - avresample_get_delay(s->avr);
     ret = avresample_convert(s->avr, NULL, 0, 0, buf->extended_data,
-                             buf->linesize[0], buf->audio->nb_samples);
+                             buf->linesize[0], buf->nb_samples);
 
     s->first_frame = 0;
 fail:
-    avfilter_unref_buffer(buf);
+    av_frame_free(&buf);
 
     return ret;
 }
diff --git a/libavfilter/af_channelmap.c b/libavfilter/af_channelmap.c
index f741881f41..a7e73f0ccb 100644
--- a/libavfilter/af_channelmap.c
+++ b/libavfilter/af_channelmap.c
@@ -313,7 +313,7 @@ static int channelmap_query_formats(AVFilterContext *ctx)
     return 0;
 }
 
-static int channelmap_filter_frame(AVFilterLink *inlink, AVFilterBufferRef *buf)
+static int channelmap_filter_frame(AVFilterLink *inlink, AVFrame *buf)
 {
     AVFilterContext  *ctx = inlink->dst;
     AVFilterLink *outlink = ctx->outputs[0];
@@ -331,7 +331,7 @@ static int channelmap_filter_frame(AVFilterLink *inlink, AVFilterBufferRef *buf)
             uint8_t **new_extended_data =
                 av_mallocz(nch_out * sizeof(*buf->extended_data));
             if (!new_extended_data) {
-                avfilter_unref_buffer(buf);
+                av_frame_free(&buf);
                 return AVERROR(ENOMEM);
             }
             if (buf->extended_data == buf->data) {
diff --git a/libavfilter/af_channelsplit.c b/libavfilter/af_channelsplit.c
index cc379f3126..d6110ef909 100644
--- a/libavfilter/af_channelsplit.c
+++ b/libavfilter/af_channelsplit.c
@@ -111,13 +111,13 @@ static int query_formats(AVFilterContext *ctx)
     return 0;
 }
 
-static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *buf)
+static int filter_frame(AVFilterLink *inlink, AVFrame *buf)
 {
     AVFilterContext *ctx = inlink->dst;
     int i, ret = 0;
 
     for (i = 0; i < ctx->nb_outputs; i++) {
-        AVFilterBufferRef *buf_out = avfilter_ref_buffer(buf, ~AV_PERM_WRITE);
+        AVFrame *buf_out = av_frame_clone(buf);
 
         if (!buf_out) {
             ret = AVERROR(ENOMEM);
@@ -125,14 +125,14 @@ static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *buf)
         }
 
         buf_out->data[0] = buf_out->extended_data[0] = buf_out->extended_data[i];
-        buf_out->audio->channel_layout =
-            av_channel_layout_extract_channel(buf->audio->channel_layout, i);
+        buf_out->channel_layout =
+            av_channel_layout_extract_channel(buf->channel_layout, i);
 
         ret = ff_filter_frame(ctx->outputs[i], buf_out);
         if (ret < 0)
             break;
     }
-    avfilter_unref_buffer(buf);
+    av_frame_free(&buf);
     return ret;
 }
 
diff --git a/libavfilter/af_join.c b/libavfilter/af_join.c
index 2b715d55ee..14742de0e4 100644
--- a/libavfilter/af_join.c
+++ b/libavfilter/af_join.c
@@ -56,24 +56,14 @@ typedef struct JoinContext {
     /**
      * Temporary storage for input frames, until we get one on each input.
      */
-    AVFilterBufferRef **input_frames;
+    AVFrame **input_frames;
 
     /**
-     *  Temporary storage for data pointers, for assembling the output buffer.
+     *  Temporary storage for buffer references, for assembling the output frame.
      */
-    uint8_t **data;
+    AVBufferRef **buffers;
 } JoinContext;
 
-/**
- * To avoid copying the data from input buffers, this filter creates
- * a custom output buffer that stores references to all inputs and
- * unrefs them on free.
- */
-typedef struct JoinBufferPriv {
-    AVFilterBufferRef **in_buffers;
-    int              nb_in_buffers;
-} JoinBufferPriv;
-
 #define OFFSET(x) offsetof(JoinContext, x)
 #define A AV_OPT_FLAG_AUDIO_PARAM
 static const AVOption join_options[] = {
@@ -93,7 +83,7 @@ static const AVClass join_class = {
     .version    = LIBAVUTIL_VERSION_INT,
 };
 
-static int filter_frame(AVFilterLink *link, AVFilterBufferRef *buf)
+static int filter_frame(AVFilterLink *link, AVFrame *frame)
 {
     AVFilterContext *ctx = link->dst;
     JoinContext       *s = ctx->priv;
@@ -104,7 +94,7 @@ static int filter_frame(AVFilterLink *link, AVFilterBufferRef *buf)
             break;
     av_assert0(i < ctx->nb_inputs);
     av_assert0(!s->input_frames[i]);
-    s->input_frames[i] = buf;
+    s->input_frames[i] = frame;
 
     return 0;
 }
@@ -208,9 +198,9 @@ static int join_init(AVFilterContext *ctx, const char *args)
 
     s->nb_channels  = av_get_channel_layout_nb_channels(s->channel_layout);
     s->channels     = av_mallocz(sizeof(*s->channels) * s->nb_channels);
-    s->data         = av_mallocz(sizeof(*s->data)     * s->nb_channels);
+    s->buffers      = av_mallocz(sizeof(*s->buffers)  * s->nb_channels);
     s->input_frames = av_mallocz(sizeof(*s->input_frames) * s->inputs);
-    if (!s->channels || !s->data || !s->input_frames) {
+    if (!s->channels || !s->buffers|| !s->input_frames) {
         ret = AVERROR(ENOMEM);
         goto fail;
     }
@@ -249,11 +239,11 @@ static void join_uninit(AVFilterContext *ctx)
 
     for (i = 0; i < ctx->nb_inputs; i++) {
         av_freep(&ctx->input_pads[i].name);
-        avfilter_unref_bufferp(&s->input_frames[i]);
+        av_frame_free(&s->input_frames[i]);
     }
 
     av_freep(&s->channels);
-    av_freep(&s->data);
+    av_freep(&s->buffers);
     av_freep(&s->input_frames);
 }
 
@@ -395,34 +385,14 @@ fail:
     return ret;
 }
 
-static void join_free_buffer(AVFilterBuffer *buf)
-{
-    JoinBufferPriv *priv = buf->priv;
-
-    if (priv) {
-        int i;
-
-        for (i = 0; i < priv->nb_in_buffers; i++)
-            avfilter_unref_bufferp(&priv->in_buffers[i]);
-
-        av_freep(&priv->in_buffers);
-        av_freep(&buf->priv);
-    }
-
-    if (buf->extended_data != buf->data)
-        av_freep(&buf->extended_data);
-    av_freep(&buf);
-}
-
 static int join_request_frame(AVFilterLink *outlink)
 {
     AVFilterContext *ctx = outlink->src;
     JoinContext *s       = ctx->priv;
-    AVFilterBufferRef *buf;
-    JoinBufferPriv *priv;
+    AVFrame *frame;
     int linesize   = INT_MAX;
-    int perms      = ~0;
     int nb_samples = 0;
+    int nb_buffers = 0;
     int i, j, ret;
 
     /* get a frame on each input */
@@ -435,54 +405,95 @@ static int join_request_frame(AVFilterLink *outlink)
 
         /* request the same number of samples on all inputs */
         if (i == 0) {
-            nb_samples = s->input_frames[0]->audio->nb_samples;
+            nb_samples = s->input_frames[0]->nb_samples;
 
             for (j = 1; !i && j < ctx->nb_inputs; j++)
                 ctx->inputs[j]->request_samples = nb_samples;
         }
     }
 
+    /* setup the output frame */
+    frame = av_frame_alloc();
+    if (!frame)
+        return AVERROR(ENOMEM);
+    if (s->nb_channels > FF_ARRAY_ELEMS(frame->data)) {
+        frame->extended_data = av_mallocz(s->nb_channels *
+                                          sizeof(*frame->extended_data));
+        if (!frame->extended_data) {
+            ret = AVERROR(ENOMEM);
+            goto fail;
+        }
+    }
+
+    /* copy the data pointers */
     for (i = 0; i < s->nb_channels; i++) {
         ChannelMap *ch = &s->channels[i];
-        AVFilterBufferRef *cur_buf = s->input_frames[ch->input];
-
-        s->data[i] = cur_buf->extended_data[ch->in_channel_idx];
-        linesize   = FFMIN(linesize, cur_buf->linesize[0]);
-        perms     &= cur_buf->perms;
-    }
+        AVFrame *cur   = s->input_frames[ch->input];
+        AVBufferRef *buf;
 
-    av_assert0(nb_samples > 0);
-    buf = avfilter_get_audio_buffer_ref_from_arrays(s->data, linesize, perms,
-                                                    nb_samples, outlink->format,
-                                                    outlink->channel_layout);
-    if (!buf)
-        return AVERROR(ENOMEM);
+        frame->extended_data[i] = cur->extended_data[ch->in_channel_idx];
+        linesize = FFMIN(linesize, cur->linesize[0]);
 
-    buf->buf->free = join_free_buffer;
-    buf->pts       = s->input_frames[0]->pts;
+        /* add the buffer where this plan is stored to the list if it's
+         * not already there */
+        buf = av_frame_get_plane_buffer(cur, ch->in_channel_idx);
+        if (!buf) {
+            ret = AVERROR(EINVAL);
+            goto fail;
+        }
+        for (j = 0; j < nb_buffers; j++)
+            if (s->buffers[j]->buffer == buf->buffer)
+                break;
+        if (j == i)
+            s->buffers[nb_buffers++] = buf;
+    }
 
-    if (!(priv = av_mallocz(sizeof(*priv))))
-        goto fail;
-    if (!(priv->in_buffers = av_mallocz(sizeof(*priv->in_buffers) * ctx->nb_inputs)))
-        goto fail;
+    /* create references to the buffers we copied to output */
+    if (nb_buffers > FF_ARRAY_ELEMS(frame->buf)) {
+        frame->nb_extended_buf = nb_buffers - FF_ARRAY_ELEMS(frame->buf);
+        frame->extended_buf = av_mallocz(sizeof(*frame->extended_buf) *
+                                         frame->nb_extended_buf);
+        if (!frame->extended_buf) {
+            frame->nb_extended_buf = 0;
+            ret = AVERROR(ENOMEM);
+            goto fail;
+        }
+    }
+    for (i = 0; i < FFMIN(FF_ARRAY_ELEMS(frame->buf), nb_buffers); i++) {
+        frame->buf[i] = av_buffer_ref(s->buffers[i]);
+        if (!frame->buf[i]) {
+            ret = AVERROR(ENOMEM);
+            goto fail;
+        }
+    }
+    for (i = 0; i < frame->nb_extended_buf; i++) {
+        frame->extended_buf[i] = av_buffer_ref(s->buffers[i +
+                                               FF_ARRAY_ELEMS(frame->buf)]);
+        if (!frame->extended_buf[i]) {
+            ret = AVERROR(ENOMEM);
+            goto fail;
+        }
+    }
 
-    for (i = 0; i < ctx->nb_inputs; i++)
-        priv->in_buffers[i] = s->input_frames[i];
-    priv->nb_in_buffers = ctx->nb_inputs;
-    buf->buf->priv      = priv;
+    frame->nb_samples     = nb_samples;
+    frame->channel_layout = outlink->channel_layout;
+    frame->sample_rate    = outlink->sample_rate;
+    frame->pts            = s->input_frames[0]->pts;
+    frame->linesize[0]    = linesize;
+    if (frame->data != frame->extended_data) {
+        memcpy(frame->data, frame->extended_data, sizeof(*frame->data) *
+               FFMIN(FF_ARRAY_ELEMS(frame->data), s->nb_channels));
+    }
 
-    ret = ff_filter_frame(outlink, buf);
+    ret = ff_filter_frame(outlink, frame);
 
     memset(s->input_frames, 0, sizeof(*s->input_frames) * ctx->nb_inputs);
 
     return ret;
 
 fail:
-    avfilter_unref_buffer(buf);
-    if (priv)
-        av_freep(&priv->in_buffers);
-    av_freep(&priv);
-    return AVERROR(ENOMEM);
+    av_frame_free(&frame);
+    return ret;
 }
 
 static const AVFilterPad avfilter_af_join_outputs[] = {
diff --git a/libavfilter/af_resample.c b/libavfilter/af_resample.c
index 84ca8f5501..f82a970bb3 100644
--- a/libavfilter/af_resample.c
+++ b/libavfilter/af_resample.c
@@ -174,7 +174,7 @@ static int request_frame(AVFilterLink *outlink)
 
     /* flush the lavr delay buffer */
     if (ret == AVERROR_EOF && s->avr) {
-        AVFilterBufferRef *buf;
+        AVFrame *frame;
         int nb_samples = av_rescale_rnd(avresample_get_delay(s->avr),
                                         outlink->sample_rate,
                                         ctx->inputs[0]->sample_rate,
@@ -183,25 +183,25 @@ static int request_frame(AVFilterLink *outlink)
         if (!nb_samples)
             return ret;
 
-        buf = ff_get_audio_buffer(outlink, AV_PERM_WRITE, nb_samples);
-        if (!buf)
+        frame = ff_get_audio_buffer(outlink, nb_samples);
+        if (!frame)
             return AVERROR(ENOMEM);
 
-        ret = avresample_convert(s->avr, buf->extended_data,
-                                 buf->linesize[0], nb_samples,
+        ret = avresample_convert(s->avr, frame->extended_data,
+                                 frame->linesize[0], nb_samples,
                                  NULL, 0, 0);
         if (ret <= 0) {
-            avfilter_unref_buffer(buf);
+            av_frame_free(&frame);
             return (ret == 0) ? AVERROR_EOF : ret;
         }
 
-        buf->pts = s->next_pts;
-        return ff_filter_frame(outlink, buf);
+        frame->pts = s->next_pts;
+        return ff_filter_frame(outlink, frame);
     }
     return ret;
 }
 
-static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *buf)
+static int filter_frame(AVFilterLink *inlink, AVFrame *in)
 {
     AVFilterContext  *ctx = inlink->dst;
     ResampleContext    *s = ctx->priv;
@@ -209,27 +209,26 @@ static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *buf)
     int ret;
 
     if (s->avr) {
-        AVFilterBufferRef *buf_out;
+        AVFrame *out;
         int delay, nb_samples;
 
         /* maximum possible samples lavr can output */
         delay      = avresample_get_delay(s->avr);
-        nb_samples = av_rescale_rnd(buf->audio->nb_samples + delay,
+        nb_samples = av_rescale_rnd(in->nb_samples + delay,
                                     outlink->sample_rate, inlink->sample_rate,
                                     AV_ROUND_UP);
 
-        buf_out = ff_get_audio_buffer(outlink, AV_PERM_WRITE, nb_samples);
-        if (!buf_out) {
+        out = ff_get_audio_buffer(outlink, nb_samples);
+        if (!out) {
             ret = AVERROR(ENOMEM);
             goto fail;
         }
 
-        ret     = avresample_convert(s->avr, buf_out->extended_data,
-                                     buf_out->linesize[0], nb_samples,
-                                     buf->extended_data, buf->linesize[0],
-                                     buf->audio->nb_samples);
+        ret = avresample_convert(s->avr, out->extended_data, out->linesize[0],
+                                 nb_samples, in->extended_data, in->linesize[0],
+                                 in->nb_samples);
         if (ret <= 0) {
-            avfilter_unref_buffer(buf_out);
+            av_frame_free(&out);
             if (ret < 0)
                 goto fail;
         }
@@ -237,36 +236,36 @@ static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *buf)
         av_assert0(!avresample_available(s->avr));
 
         if (s->next_pts == AV_NOPTS_VALUE) {
-            if (buf->pts == AV_NOPTS_VALUE) {
+            if (in->pts == AV_NOPTS_VALUE) {
                 av_log(ctx, AV_LOG_WARNING, "First timestamp is missing, "
                        "assuming 0.\n");
                 s->next_pts = 0;
             } else
-                s->next_pts = av_rescale_q(buf->pts, inlink->time_base,
+                s->next_pts = av_rescale_q(in->pts, inlink->time_base,
                                            outlink->time_base);
         }
 
         if (ret > 0) {
-            buf_out->audio->nb_samples = ret;
-            if (buf->pts != AV_NOPTS_VALUE) {
-                buf_out->pts = av_rescale_q(buf->pts, inlink->time_base,
+            out->nb_samples = ret;
+            if (in->pts != AV_NOPTS_VALUE) {
+                out->pts = av_rescale_q(in->pts, inlink->time_base,
                                             outlink->time_base) -
                                av_rescale(delay, outlink->sample_rate,
                                           inlink->sample_rate);
             } else
-                buf_out->pts = s->next_pts;
+                out->pts = s->next_pts;
 
-            s->next_pts = buf_out->pts + buf_out->audio->nb_samples;
+            s->next_pts = out->pts + out->nb_samples;
 
-            ret = ff_filter_frame(outlink, buf_out);
+            ret = ff_filter_frame(outlink, out);
             s->got_output = 1;
         }
 
 fail:
-        avfilter_unref_buffer(buf);
+        av_frame_free(&in);
     } else {
-        buf->format = outlink->format;
-        ret = ff_filter_frame(outlink, buf);
+        in->format = outlink->format;
+        ret = ff_filter_frame(outlink, in);
         s->got_output = 1;
     }
 
@@ -278,7 +277,6 @@ static const AVFilterPad avfilter_af_resample_inputs[] = {
         .name           = "default",
         .type           = AVMEDIA_TYPE_AUDIO,
         .filter_frame   = filter_frame,
-        .min_perms      = AV_PERM_READ
     },
     { NULL }
 };
diff --git a/libavfilter/af_volume.c b/libavfilter/af_volume.c
index 3f3ad47258..44181af7d8 100644
--- a/libavfilter/af_volume.c
+++ b/libavfilter/af_volume.c
@@ -233,21 +233,21 @@ static int config_output(AVFilterLink *outlink)
     return 0;
 }
 
-static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *buf)
+static int filter_frame(AVFilterLink *inlink, AVFrame *buf)
 {
     VolumeContext *vol    = inlink->dst->priv;
     AVFilterLink *outlink = inlink->dst->outputs[0];
-    int nb_samples        = buf->audio->nb_samples;
-    AVFilterBufferRef *out_buf;
+    int nb_samples        = buf->nb_samples;
+    AVFrame *out_buf;
 
     if (vol->volume == 1.0 || vol->volume_i == 256)
         return ff_filter_frame(outlink, buf);
 
     /* do volume scaling in-place if input buffer is writable */
-    if (buf->perms & AV_PERM_WRITE) {
+    if (av_frame_is_writable(buf)) {
         out_buf = buf;
     } else {
-        out_buf = ff_get_audio_buffer(inlink, AV_PERM_WRITE, nb_samples);
+        out_buf = ff_get_audio_buffer(inlink, nb_samples);
         if (!out_buf)
             return AVERROR(ENOMEM);
         out_buf->pts = buf->pts;
@@ -283,7 +283,7 @@ static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *buf)
     }
 
     if (buf != out_buf)
-        avfilter_unref_buffer(buf);
+        av_frame_free(&buf);
 
     return ff_filter_frame(outlink, out_buf);
 }
diff --git a/libavfilter/asink_anullsink.c b/libavfilter/asink_anullsink.c
index ede54c0065..756dff08d0 100644
--- a/libavfilter/asink_anullsink.c
+++ b/libavfilter/asink_anullsink.c
@@ -20,9 +20,9 @@
 #include "avfilter.h"
 #include "internal.h"
 
-static int null_filter_frame(AVFilterLink *link, AVFilterBufferRef *samplesref)
+static int null_filter_frame(AVFilterLink *link, AVFrame *frame)
 {
-    avfilter_unref_bufferp(&samplesref);
+    av_frame_free(&frame);
     return 0;
 }
 
diff --git a/libavfilter/audio.c b/libavfilter/audio.c
index bbe12b211b..5295423e44 100644
--- a/libavfilter/audio.c
+++ b/libavfilter/audio.c
@@ -23,60 +23,65 @@
 #include "avfilter.h"
 #include "internal.h"
 
-AVFilterBufferRef *ff_null_get_audio_buffer(AVFilterLink *link, int perms,
-                                            int nb_samples)
+AVFrame *ff_null_get_audio_buffer(AVFilterLink *link, int nb_samples)
 {
-    return ff_get_audio_buffer(link->dst->outputs[0], perms, nb_samples);
+    return ff_get_audio_buffer(link->dst->outputs[0], nb_samples);
 }
 
-AVFilterBufferRef *ff_default_get_audio_buffer(AVFilterLink *link, int perms,
-                                               int nb_samples)
+AVFrame *ff_default_get_audio_buffer(AVFilterLink *link, int nb_samples)
 {
-    AVFilterBufferRef *samplesref = NULL;
-    uint8_t **data;
-    int planar      = av_sample_fmt_is_planar(link->format);
-    int nb_channels = av_get_channel_layout_nb_channels(link->channel_layout);
-    int planes      = planar ? nb_channels : 1;
-    int linesize;
-
-    if (!(data = av_mallocz(sizeof(*data) * planes)))
+    AVFrame *frame = av_frame_alloc();
+    int channels = av_get_channel_layout_nb_channels(link->channel_layout);
+    int buf_size, ret;
+
+    if (!frame)
+        return NULL;
+
+    buf_size = av_samples_get_buffer_size(NULL, channels, nb_samples,
+                                          link->format, 0);
+    if (buf_size < 0)
         goto fail;
 
-    if (av_samples_alloc(data, &linesize, nb_channels, nb_samples, link->format, 0) < 0)
+    frame->buf[0] = av_buffer_alloc(buf_size);
+    if (!frame->buf[0])
         goto fail;
 
-    samplesref = avfilter_get_audio_buffer_ref_from_arrays(data, linesize, perms,
-                                                           nb_samples, link->format,
-                                                           link->channel_layout);
-    if (!samplesref)
+    frame->nb_samples = nb_samples;
+    ret = avcodec_fill_audio_frame(frame, channels, link->format,
+                                   frame->buf[0]->data, buf_size, 0);
+    if (ret < 0)
         goto fail;
 
-    av_freep(&data);
+    av_samples_set_silence(frame->extended_data, 0, nb_samples, channels,
+                           link->format);
+
+    frame->nb_samples     = nb_samples;
+    frame->format         = link->format;
+    frame->channel_layout = link->channel_layout;
+    frame->sample_rate    = link->sample_rate;
+
+    return frame;
 
 fail:
-    if (data)
-        av_freep(&data[0]);
-    av_freep(&data);
-    return samplesref;
+    av_buffer_unref(&frame->buf[0]);
+    av_frame_free(&frame);
+    return NULL;
 }
 
-AVFilterBufferRef *ff_get_audio_buffer(AVFilterLink *link, int perms,
-                                       int nb_samples)
+AVFrame *ff_get_audio_buffer(AVFilterLink *link, int nb_samples)
 {
-    AVFilterBufferRef *ret = NULL;
+    AVFrame *ret = NULL;
 
     if (link->dstpad->get_audio_buffer)
-        ret = link->dstpad->get_audio_buffer(link, perms, nb_samples);
+        ret = link->dstpad->get_audio_buffer(link, nb_samples);
 
     if (!ret)
-        ret = ff_default_get_audio_buffer(link, perms, nb_samples);
-
-    if (ret)
-        ret->type = AVMEDIA_TYPE_AUDIO;
+        ret = ff_default_get_audio_buffer(link, nb_samples);
 
     return ret;
 }
 
+#if FF_API_AVFILTERBUFFER
 AVFilterBufferRef* avfilter_get_audio_buffer_ref_from_arrays(uint8_t **data,
                                                              int linesize,int perms,
                                                              int nb_samples,
@@ -146,3 +151,4 @@ fail:
     av_freep(&samples);
     return NULL;
 }
+#endif
diff --git a/libavfilter/audio.h b/libavfilter/audio.h
index a3775037e9..4684b6ce60 100644
--- a/libavfilter/audio.h
+++ b/libavfilter/audio.h
@@ -22,24 +22,20 @@
 #include "avfilter.h"
 
 /** default handler for get_audio_buffer() for audio inputs */
-AVFilterBufferRef *ff_default_get_audio_buffer(AVFilterLink *link, int perms,
-                                                     int nb_samples);
+AVFrame *ff_default_get_audio_buffer(AVFilterLink *link, int nb_samples);
 
 /** get_audio_buffer() handler for filters which simply pass audio along */
-AVFilterBufferRef *ff_null_get_audio_buffer(AVFilterLink *link, int perms,
-                                                  int nb_samples);
+AVFrame *ff_null_get_audio_buffer(AVFilterLink *link, int nb_samples);
 
 /**
  * Request an audio samples buffer with a specific set of permissions.
  *
  * @param link           the output link to the filter from which the buffer will
  *                       be requested
- * @param perms          the required access permissions
  * @param nb_samples     the number of samples per channel
  * @return               A reference to the samples. This must be unreferenced with
  *                       avfilter_unref_buffer when you are finished with it.
  */
-AVFilterBufferRef *ff_get_audio_buffer(AVFilterLink *link, int perms,
-                                             int nb_samples);
+AVFrame *ff_get_audio_buffer(AVFilterLink *link, int nb_samples);
 
 #endif /* AVFILTER_AUDIO_H */
diff --git a/libavfilter/avfilter.c b/libavfilter/avfilter.c
index 93302ccd23..af9d9eeb42 100644
--- a/libavfilter/avfilter.c
+++ b/libavfilter/avfilter.c
@@ -451,17 +451,16 @@ enum AVMediaType avfilter_pad_get_type(AVFilterPad *pads, int pad_idx)
     return pads[pad_idx].type;
 }
 
-static int default_filter_frame(AVFilterLink *link, AVFilterBufferRef *frame)
+static int default_filter_frame(AVFilterLink *link, AVFrame *frame)
 {
     return ff_filter_frame(link->dst->outputs[0], frame);
 }
 
-int ff_filter_frame(AVFilterLink *link, AVFilterBufferRef *frame)
+int ff_filter_frame(AVFilterLink *link, AVFrame *frame)
 {
-    int (*filter_frame)(AVFilterLink *, AVFilterBufferRef *);
+    int (*filter_frame)(AVFilterLink *, AVFrame *);
     AVFilterPad *dst = link->dstpad;
-    AVFilterBufferRef *out;
-    int perms = frame->perms;
+    AVFrame *out;
 
     FF_DPRINTF_START(NULL, filter_frame);
     ff_dlog_link(NULL, link, 1);
@@ -469,47 +468,40 @@ int ff_filter_frame(AVFilterLink *link, AVFilterBufferRef *frame)
     if (!(filter_frame = dst->filter_frame))
         filter_frame = default_filter_frame;
 
-    if (frame->linesize[0] < 0)
-        perms |= AV_PERM_NEG_LINESIZES;
-    /* prepare to copy the frame if the buffer has insufficient permissions */
-    if ((dst->min_perms & perms) != dst->min_perms ||
-        dst->rej_perms & perms) {
-        av_log(link->dst, AV_LOG_DEBUG,
-               "Copying data in avfilter (have perms %x, need %x, reject %x)\n",
-               perms, link->dstpad->min_perms, link->dstpad->rej_perms);
+    /* copy the frame if needed */
+    if (dst->needs_writable && !av_frame_is_writable(frame)) {
+        av_log(link->dst, AV_LOG_DEBUG, "Copying data in avfilter.\n");
 
         switch (link->type) {
         case AVMEDIA_TYPE_VIDEO:
-            out = ff_get_video_buffer(link, dst->min_perms,
-                                      link->w, link->h);
+            out = ff_get_video_buffer(link, link->w, link->h);
             break;
         case AVMEDIA_TYPE_AUDIO:
-            out = ff_get_audio_buffer(link, dst->min_perms,
-                                      frame->audio->nb_samples);
+            out = ff_get_audio_buffer(link, frame->nb_samples);
             break;
         default: return AVERROR(EINVAL);
         }
         if (!out) {
-            avfilter_unref_buffer(frame);
+            av_frame_free(&frame);
             return AVERROR(ENOMEM);
         }
-        avfilter_copy_buffer_ref_props(out, frame);
+        av_frame_copy_props(out, frame);
 
         switch (link->type) {
         case AVMEDIA_TYPE_VIDEO:
             av_image_copy(out->data, out->linesize, frame->data, frame->linesize,
-                          frame->format, frame->video->w, frame->video->h);
+                          frame->format, frame->width, frame->height);
             break;
         case AVMEDIA_TYPE_AUDIO:
             av_samples_copy(out->extended_data, frame->extended_data,
-                            0, 0, frame->audio->nb_samples,
-                            av_get_channel_layout_nb_channels(frame->audio->channel_layout),
+                            0, 0, frame->nb_samples,
+                            av_get_channel_layout_nb_channels(frame->channel_layout),
                             frame->format);
             break;
         default: return AVERROR(EINVAL);
         }
 
-        avfilter_unref_buffer(frame);
+        av_frame_free(&frame);
     } else
         out = frame;
 
diff --git a/libavfilter/avfilter.h b/libavfilter/avfilter.h
index c5f8d56297..b3522f87d3 100644
--- a/libavfilter/avfilter.h
+++ b/libavfilter/avfilter.h
@@ -23,6 +23,7 @@
 #define AVFILTER_AVFILTER_H
 
 #include "libavutil/avutil.h"
+#include "libavutil/frame.h"
 #include "libavutil/log.h"
 #include "libavutil/samplefmt.h"
 #include "libavutil/pixfmt.h"
@@ -54,6 +55,7 @@ typedef struct AVFilterLink    AVFilterLink;
 typedef struct AVFilterPad     AVFilterPad;
 typedef struct AVFilterFormats AVFilterFormats;
 
+#if FF_API_AVFILTERBUFFER
 /**
  * A reference-counted buffer data type used by the filter system. Filters
  * should not store pointers to this structure directly, but instead use the
@@ -177,6 +179,7 @@ typedef struct AVFilterBufferRef {
 /**
  * Copy properties of src to dst, without copying the actual data
  */
+attribute_deprecated
 void avfilter_copy_buffer_ref_props(AVFilterBufferRef *dst, AVFilterBufferRef *src);
 
 /**
@@ -188,6 +191,7 @@ void avfilter_copy_buffer_ref_props(AVFilterBufferRef *dst, AVFilterBufferRef *s
  * @return      a new reference to the buffer with the same properties as the
  *              old, excluding any permissions denied by pmask
  */
+attribute_deprecated
 AVFilterBufferRef *avfilter_ref_buffer(AVFilterBufferRef *ref, int pmask);
 
 /**
@@ -199,6 +203,7 @@ AVFilterBufferRef *avfilter_ref_buffer(AVFilterBufferRef *ref, int pmask);
  * @note it is recommended to use avfilter_unref_bufferp() instead of this
  * function
  */
+attribute_deprecated
 void avfilter_unref_buffer(AVFilterBufferRef *ref);
 
 /**
@@ -208,7 +213,9 @@ void avfilter_unref_buffer(AVFilterBufferRef *ref);
  *
  * @param ref pointer to the buffer reference
  */
+attribute_deprecated
 void avfilter_unref_bufferp(AVFilterBufferRef **ref);
+#endif
 
 #if FF_API_AVFILTERPAD_PUBLIC
 /**
@@ -239,7 +246,7 @@ struct AVFilterPad {
      *
      * Input pads only.
      */
-    int min_perms;
+    attribute_deprecated int min_perms;
 
     /**
      * Permissions which are not accepted on incoming buffers. Any buffer
@@ -250,7 +257,7 @@ struct AVFilterPad {
      *
      * Input pads only.
      */
-    int rej_perms;
+    attribute_deprecated int rej_perms;
 
     /**
      * @deprecated unused
@@ -263,7 +270,7 @@ struct AVFilterPad {
      *
      * Input video pads only.
      */
-    AVFilterBufferRef *(*get_video_buffer)(AVFilterLink *link, int perms, int w, int h);
+    AVFrame *(*get_video_buffer)(AVFilterLink *link, int w, int h);
 
     /**
      * Callback function to get an audio buffer. If NULL, the filter system will
@@ -271,8 +278,7 @@ struct AVFilterPad {
      *
      * Input audio pads only.
      */
-    AVFilterBufferRef *(*get_audio_buffer)(AVFilterLink *link, int perms,
-                                           int nb_samples);
+    AVFrame *(*get_audio_buffer)(AVFilterLink *link, int nb_samples);
 
     /**
      * @deprecated unused
@@ -294,7 +300,7 @@ struct AVFilterPad {
      * must ensure that samplesref is properly unreferenced on error if it
      * hasn't been passed on to another filter.
      */
-    int (*filter_frame)(AVFilterLink *link, AVFilterBufferRef *frame);
+    int (*filter_frame)(AVFilterLink *link, AVFrame *frame);
 
     /**
      * Frame poll callback. This returns the number of immediately available
@@ -339,6 +345,8 @@ struct AVFilterPad {
      * input pads only.
      */
     int needs_fifo;
+
+    int needs_writable;
 };
 #endif
 
@@ -535,6 +543,7 @@ int avfilter_link(AVFilterContext *src, unsigned srcpad,
  */
 int avfilter_config_links(AVFilterContext *filter);
 
+#if FF_API_AVFILTERBUFFER
 /**
  * Create a buffer reference wrapped around an already allocated image
  * buffer.
@@ -546,6 +555,7 @@ int avfilter_config_links(AVFilterContext *filter);
  * @param h the height of the image specified by the data and linesize arrays
  * @param format the pixel format of the image specified by the data and linesize arrays
  */
+attribute_deprecated
 AVFilterBufferRef *
 avfilter_get_video_buffer_ref_from_arrays(uint8_t *data[4], int linesize[4], int perms,
                                           int w, int h, enum AVPixelFormat format);
@@ -561,12 +571,14 @@ avfilter_get_video_buffer_ref_from_arrays(uint8_t *data[4], int linesize[4], int
  * @param sample_fmt     the format of each sample in the buffer to allocate
  * @param channel_layout the channel layout of the buffer
  */
+attribute_deprecated
 AVFilterBufferRef *avfilter_get_audio_buffer_ref_from_arrays(uint8_t **data,
                                                              int linesize,
                                                              int perms,
                                                              int nb_samples,
                                                              enum AVSampleFormat sample_fmt,
                                                              uint64_t channel_layout);
+#endif
 
 /** Initialize the filter system. Register all builtin filters. */
 void avfilter_register_all(void);
@@ -645,12 +657,14 @@ void avfilter_free(AVFilterContext *filter);
 int avfilter_insert_filter(AVFilterLink *link, AVFilterContext *filt,
                            unsigned filt_srcpad_idx, unsigned filt_dstpad_idx);
 
+#if FF_API_AVFILTERBUFFER
 /**
  * Copy the frame properties of src to dst, without copying the actual
  * image data.
  *
  * @return 0 on success, a negative number on error.
  */
+attribute_deprecated
 int avfilter_copy_frame_props(AVFilterBufferRef *dst, const AVFrame *src);
 
 /**
@@ -659,6 +673,8 @@ int avfilter_copy_frame_props(AVFilterBufferRef *dst, const AVFrame *src);
  *
  * @return 0 on success, a negative number on error.
  */
+attribute_deprecated
 int avfilter_copy_buf_props(AVFrame *dst, const AVFilterBufferRef *src);
+#endif
 
 #endif /* AVFILTER_AVFILTER_H */
diff --git a/libavfilter/buffersink.c b/libavfilter/buffersink.c
index a315cb394f..6f7529172c 100644
--- a/libavfilter/buffersink.c
+++ b/libavfilter/buffersink.c
@@ -35,7 +35,7 @@
 #include "internal.h"
 
 typedef struct {
-    AVFilterBufferRef *cur_buf;  ///< last buffer delivered on the sink
+    AVFrame *cur_frame;          ///< last frame delivered on the sink
     AVAudioFifo  *audio_fifo;    ///< FIFO for audio samples
     int64_t next_pts;            ///< interpolating audio pts
 } BufferSinkContext;
@@ -48,59 +48,57 @@ static av_cold void uninit(AVFilterContext *ctx)
         av_audio_fifo_free(sink->audio_fifo);
 }
 
-static int filter_frame(AVFilterLink *link, AVFilterBufferRef *buf)
+static int filter_frame(AVFilterLink *link, AVFrame *frame)
 {
     BufferSinkContext *s = link->dst->priv;
 
-    av_assert0(!s->cur_buf);
-    s->cur_buf    = buf;
+    av_assert0(!s->cur_frame);
+    s->cur_frame = frame;
 
     return 0;
 }
 
-int av_buffersink_read(AVFilterContext *ctx, AVFilterBufferRef **buf)
+int av_buffersink_get_frame(AVFilterContext *ctx, AVFrame *frame)
 {
     BufferSinkContext *s    = ctx->priv;
     AVFilterLink      *link = ctx->inputs[0];
     int ret;
 
-    if (!buf)
-        return ff_poll_frame(ctx->inputs[0]);
-
     if ((ret = ff_request_frame(link)) < 0)
         return ret;
 
-    if (!s->cur_buf)
+    if (!s->cur_frame)
         return AVERROR(EINVAL);
 
-    *buf       = s->cur_buf;
-    s->cur_buf = NULL;
+    av_frame_move_ref(frame, s->cur_frame);
+    av_frame_free(&s->cur_frame);
 
     return 0;
 }
 
-static int read_from_fifo(AVFilterContext *ctx, AVFilterBufferRef **pbuf,
+static int read_from_fifo(AVFilterContext *ctx, AVFrame *frame,
                           int nb_samples)
 {
     BufferSinkContext *s = ctx->priv;
     AVFilterLink   *link = ctx->inputs[0];
-    AVFilterBufferRef *buf;
+    AVFrame *tmp;
 
-    if (!(buf = ff_get_audio_buffer(link, AV_PERM_WRITE, nb_samples)))
+    if (!(tmp = ff_get_audio_buffer(link, nb_samples)))
         return AVERROR(ENOMEM);
-    av_audio_fifo_read(s->audio_fifo, (void**)buf->extended_data, nb_samples);
+    av_audio_fifo_read(s->audio_fifo, (void**)tmp->extended_data, nb_samples);
 
-    buf->pts = s->next_pts;
+    tmp->pts = s->next_pts;
     s->next_pts += av_rescale_q(nb_samples, (AVRational){1, link->sample_rate},
                                 link->time_base);
 
-    *pbuf = buf;
+    av_frame_move_ref(frame, tmp);
+    av_frame_free(&tmp);
+
     return 0;
 
 }
 
-int av_buffersink_read_samples(AVFilterContext *ctx, AVFilterBufferRef **pbuf,
-                               int nb_samples)
+int av_buffersink_get_samples(AVFilterContext *ctx, AVFrame *frame, int nb_samples)
 {
     BufferSinkContext *s = ctx->priv;
     AVFilterLink   *link = ctx->inputs[0];
@@ -113,38 +111,107 @@ int av_buffersink_read_samples(AVFilterContext *ctx, AVFilterBufferRef **pbuf,
     }
 
     while (ret >= 0) {
-        AVFilterBufferRef *buf;
-
         if (av_audio_fifo_size(s->audio_fifo) >= nb_samples)
-            return read_from_fifo(ctx, pbuf, nb_samples);
+            return read_from_fifo(ctx, frame, nb_samples);
 
-        ret = av_buffersink_read(ctx, &buf);
+        ret = ff_request_frame(link);
         if (ret == AVERROR_EOF && av_audio_fifo_size(s->audio_fifo))
-            return read_from_fifo(ctx, pbuf, av_audio_fifo_size(s->audio_fifo));
+            return read_from_fifo(ctx, frame, av_audio_fifo_size(s->audio_fifo));
         else if (ret < 0)
             return ret;
 
-        if (buf->pts != AV_NOPTS_VALUE) {
-            s->next_pts = buf->pts -
+        if (s->cur_frame->pts != AV_NOPTS_VALUE) {
+            s->next_pts = s->cur_frame->pts -
                           av_rescale_q(av_audio_fifo_size(s->audio_fifo),
                                        (AVRational){ 1, link->sample_rate },
                                        link->time_base);
         }
 
-        ret = av_audio_fifo_write(s->audio_fifo, (void**)buf->extended_data,
-                                  buf->audio->nb_samples);
-        avfilter_unref_buffer(buf);
+        ret = av_audio_fifo_write(s->audio_fifo, (void**)s->cur_frame->extended_data,
+                                  s->cur_frame->nb_samples);
+        av_frame_free(&s->cur_frame);
     }
 
     return ret;
+
+}
+
+#if FF_API_AVFILTERBUFFER
+static void compat_free_buffer(AVFilterBuffer *buf)
+{
+    AVFrame *frame = buf->priv;
+    av_frame_free(&frame);
+    av_free(buf);
+}
+
+static int compat_read(AVFilterContext *ctx, AVFilterBufferRef **pbuf, int nb_samples)
+{
+    AVFilterBufferRef *buf;
+    AVFrame *frame;
+    int ret;
+
+    if (!pbuf)
+        return ff_poll_frame(ctx->inputs[0]);
+
+    frame = av_frame_alloc();
+    if (!frame)
+        return AVERROR(ENOMEM);
+
+    if (!nb_samples)
+        ret = av_buffersink_get_frame(ctx, frame);
+    else
+        ret = av_buffersink_get_samples(ctx, frame, nb_samples);
+
+    if (ret < 0)
+        goto fail;
+
+    if (ctx->inputs[0]->type == AVMEDIA_TYPE_VIDEO) {
+        buf = avfilter_get_video_buffer_ref_from_arrays(frame->data, frame->linesize,
+                                                        AV_PERM_READ,
+                                                        frame->width, frame->height,
+                                                        frame->format);
+    } else {
+        buf = avfilter_get_audio_buffer_ref_from_arrays(frame->extended_data,
+                                                        frame->linesize[0], AV_PERM_READ,
+                                                        frame->nb_samples,
+                                                        frame->format,
+                                                        frame->channel_layout);
+    }
+    if (!buf) {
+        ret = AVERROR(ENOMEM);
+        goto fail;
+    }
+
+    avfilter_copy_frame_props(buf, frame);
+
+    buf->buf->priv = frame;
+    buf->buf->free = compat_free_buffer;
+
+    *pbuf = buf;
+
+    return 0;
+fail:
+    av_frame_free(&frame);
+    return ret;
+}
+
+int av_buffersink_read(AVFilterContext *ctx, AVFilterBufferRef **buf)
+{
+    return compat_read(ctx, buf, 0);
+}
+
+int av_buffersink_read_samples(AVFilterContext *ctx, AVFilterBufferRef **buf,
+                               int nb_samples)
+{
+    return compat_read(ctx, buf, nb_samples);
 }
+#endif
 
 static const AVFilterPad avfilter_vsink_buffer_inputs[] = {
     {
         .name        = "default",
         .type        = AVMEDIA_TYPE_VIDEO,
         .filter_frame = filter_frame,
-        .min_perms   = AV_PERM_READ,
         .needs_fifo  = 1
     },
     { NULL }
@@ -165,7 +232,6 @@ static const AVFilterPad avfilter_asink_abuffer_inputs[] = {
         .name           = "default",
         .type           = AVMEDIA_TYPE_AUDIO,
         .filter_frame   = filter_frame,
-        .min_perms      = AV_PERM_READ,
         .needs_fifo     = 1
     },
     { NULL }
diff --git a/libavfilter/buffersink.h b/libavfilter/buffersink.h
index e358ac3412..75cea35b77 100644
--- a/libavfilter/buffersink.h
+++ b/libavfilter/buffersink.h
@@ -26,6 +26,7 @@
 
 #include "avfilter.h"
 
+#if FF_API_AVFILTERBUFFER
 /**
  * Get a buffer with filtered data from sink and put it in buf.
  *
@@ -38,6 +39,7 @@
  * @return >= 0 in case of success, a negative AVERROR code in case of
  *         failure.
  */
+attribute_deprecated
 int av_buffersink_read(AVFilterContext *ctx, AVFilterBufferRef **buf);
 
 /**
@@ -56,7 +58,37 @@ int av_buffersink_read(AVFilterContext *ctx, AVFilterBufferRef **buf);
  * @warning do not mix this function with av_buffersink_read(). Use only one or
  * the other with a single sink, not both.
  */
+attribute_deprecated
 int av_buffersink_read_samples(AVFilterContext *ctx, AVFilterBufferRef **buf,
                                int nb_samples);
+#endif
+
+/**
+ * Get a frame with filtered data from sink and put it in frame.
+ *
+ * @param ctx pointer to a context of a buffersink or abuffersink AVFilter.
+ * @param frame pointer to an allocated frame that will be filled with data.
+ *              The data must be freed using av_frame_unref() / av_frame_free()
+ *
+ * @return >= 0 in case of success, a negative AVERROR code in case of
+ *         failure.
+ */
+int av_buffersink_get_frame(AVFilterContext *ctx, AVFrame *frame);
+
+/**
+ * Same as av_buffersink_get_frame(), but with the ability to specify the number
+ * of samples read. This function is less efficient than
+ * av_buffersink_get_frame(), because it copies the data around.
+ *
+ * @param ctx pointer to a context of the abuffersink AVFilter.
+ * @param frame pointer to an allocated frame that will be filled with data.
+ *              The data must be freed using av_frame_unref() / av_frame_free()
+ *              frame will contain exactly nb_samples audio samples, except at
+ *              the end of stream, when it can contain less than nb_samples.
+ *
+ * @warning do not mix this function with av_buffersink_get_frame(). Use only one or
+ * the other with a single sink, not both.
+ */
+int av_buffersink_get_samples(AVFilterContext *ctx, AVFrame *frame, int nb_samples);
 
 #endif /* AVFILTER_BUFFERSINK_H */
diff --git a/libavfilter/buffersrc.c b/libavfilter/buffersrc.c
index 3cee68da43..d35b302a9e 100644
--- a/libavfilter/buffersrc.c
+++ b/libavfilter/buffersrc.c
@@ -26,6 +26,7 @@
 #include "libavutil/channel_layout.h"
 #include "libavutil/common.h"
 #include "libavutil/fifo.h"
+#include "libavutil/frame.h"
 #include "libavutil/imgutils.h"
 #include "libavutil/opt.h"
 #include "libavutil/samplefmt.h"
@@ -69,95 +70,168 @@ typedef struct {
         return AVERROR(EINVAL);\
     }
 
-int av_buffersrc_write_frame(AVFilterContext *buffer_filter, const AVFrame *frame)
+int av_buffersrc_write_frame(AVFilterContext *ctx, const AVFrame *frame)
 {
-    BufferSourceContext *c = buffer_filter->priv;
-    AVFilterBufferRef *buf;
+    AVFrame *copy;
+    int ret = 0;
+
+    if (!(copy = av_frame_alloc()))
+        return AVERROR(ENOMEM);
+    ret = av_frame_ref(copy, frame);
+    if (ret >= 0)
+        ret = av_buffersrc_add_frame(ctx, copy);
+
+    av_frame_free(&copy);
+    return ret;
+}
+
+int av_buffersrc_add_frame(AVFilterContext *ctx, AVFrame *frame)
+{
+    BufferSourceContext *s = ctx->priv;
+    AVFrame *copy;
     int ret;
 
     if (!frame) {
-        c->eof = 1;
+        s->eof = 1;
         return 0;
-    } else if (c->eof)
+    } else if (s->eof)
         return AVERROR(EINVAL);
 
-    if (!av_fifo_space(c->fifo) &&
-        (ret = av_fifo_realloc2(c->fifo, av_fifo_size(c->fifo) +
-                                         sizeof(buf))) < 0)
-        return ret;
-
-    switch (buffer_filter->outputs[0]->type) {
+    switch (ctx->outputs[0]->type) {
     case AVMEDIA_TYPE_VIDEO:
-        CHECK_VIDEO_PARAM_CHANGE(buffer_filter, c, frame->width, frame->height,
+        CHECK_VIDEO_PARAM_CHANGE(ctx, s, frame->width, frame->height,
                                  frame->format);
-        buf = ff_get_video_buffer(buffer_filter->outputs[0], AV_PERM_WRITE,
-                                  c->w, c->h);
-        if (!buf)
-            return AVERROR(ENOMEM);
-
-        av_image_copy(buf->data, buf->linesize, frame->data, frame->linesize,
-                      c->pix_fmt, c->w, c->h);
         break;
     case AVMEDIA_TYPE_AUDIO:
-        CHECK_AUDIO_PARAM_CHANGE(buffer_filter, c, frame->sample_rate, frame->channel_layout,
+        CHECK_AUDIO_PARAM_CHANGE(ctx, s, frame->sample_rate, frame->channel_layout,
                                  frame->format);
-        buf = ff_get_audio_buffer(buffer_filter->outputs[0], AV_PERM_WRITE,
-                                  frame->nb_samples);
-        if (!buf)
-            return AVERROR(ENOMEM);
-
-        av_samples_copy(buf->extended_data, frame->extended_data,
-                        0, 0, frame->nb_samples,
-                        av_get_channel_layout_nb_channels(frame->channel_layout),
-                        frame->format);
         break;
     default:
         return AVERROR(EINVAL);
     }
 
-    avfilter_copy_frame_props(buf, frame);
+    if (!av_fifo_space(s->fifo) &&
+        (ret = av_fifo_realloc2(s->fifo, av_fifo_size(s->fifo) +
+                                         sizeof(copy))) < 0)
+        return ret;
+
+    if (!(copy = av_frame_alloc()))
+        return AVERROR(ENOMEM);
+    av_frame_move_ref(copy, frame);
 
-    if ((ret = av_fifo_generic_write(c->fifo, &buf, sizeof(buf), NULL)) < 0) {
-        avfilter_unref_buffer(buf);
+    if ((ret = av_fifo_generic_write(s->fifo, &copy, sizeof(copy), NULL)) < 0) {
+        av_frame_move_ref(frame, copy);
+        av_frame_free(&copy);
         return ret;
     }
 
     return 0;
 }
 
-int av_buffersrc_buffer(AVFilterContext *s, AVFilterBufferRef *buf)
+#if FF_API_AVFILTERBUFFER
+static void compat_free_buffer(void *opaque, uint8_t *data)
 {
-    BufferSourceContext *c = s->priv;
-    int ret;
+    AVFilterBufferRef *buf = opaque;
+    avfilter_unref_buffer(buf);
+}
+
+static void compat_unref_buffer(void *opaque, uint8_t *data)
+{
+    AVBufferRef *buf = opaque;
+    av_buffer_unref(&buf);
+}
+
+int av_buffersrc_buffer(AVFilterContext *ctx, AVFilterBufferRef *buf)
+{
+    BufferSourceContext *s = ctx->priv;
+    AVFrame *frame = NULL;
+    AVBufferRef *dummy_buf = NULL;
+    int ret = 0, planes, i;
 
     if (!buf) {
-        c->eof = 1;
+        s->eof = 1;
         return 0;
-    } else if (c->eof)
+    } else if (s->eof)
         return AVERROR(EINVAL);
 
-    if (!av_fifo_space(c->fifo) &&
-        (ret = av_fifo_realloc2(c->fifo, av_fifo_size(c->fifo) +
-                                         sizeof(buf))) < 0)
-        return ret;
+    frame = av_frame_alloc();
+    if (!frame)
+        return AVERROR(ENOMEM);
 
-    switch (s->outputs[0]->type) {
-    case AVMEDIA_TYPE_VIDEO:
-        CHECK_VIDEO_PARAM_CHANGE(s, c, buf->video->w, buf->video->h, buf->format);
-        break;
-    case AVMEDIA_TYPE_AUDIO:
-        CHECK_AUDIO_PARAM_CHANGE(s, c, buf->audio->sample_rate, buf->audio->channel_layout,
-                                 buf->format);
-        break;
-    default:
-        return AVERROR(EINVAL);
+    dummy_buf = av_buffer_create(NULL, 0, compat_free_buffer, buf, 0);
+    if (!dummy_buf) {
+        ret = AVERROR(ENOMEM);
+        goto fail;
     }
 
-    if ((ret = av_fifo_generic_write(c->fifo, &buf, sizeof(buf), NULL)) < 0)
-        return ret;
+    if ((ret = avfilter_copy_buf_props(frame, buf)) < 0)
+        goto fail;
 
-    return 0;
+#define WRAP_PLANE(ref_out, data, data_size)                            \
+do {                                                                    \
+    AVBufferRef *dummy_ref = av_buffer_ref(dummy_buf);                  \
+    if (!dummy_ref) {                                                   \
+        ret = AVERROR(ENOMEM);                                          \
+        goto fail;                                                      \
+    }                                                                   \
+    ref_out = av_buffer_create(data, data_size, compat_unref_buffer,    \
+                               dummy_ref, 0);                           \
+    if (!ref_out) {                                                     \
+        av_frame_unref(frame);                                          \
+        ret = AVERROR(ENOMEM);                                          \
+        goto fail;                                                      \
+    }                                                                   \
+} while (0)
+
+    if (ctx->outputs[0]->type  == AVMEDIA_TYPE_VIDEO) {
+        const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(frame->format);
+
+        if (!desc) {
+            ret = AVERROR(EINVAL);
+            goto fail;
+        }
+        planes = (desc->flags & PIX_FMT_PLANAR) ? desc->nb_components : 1;
+
+        for (i = 0; i < planes; i++) {
+            int h_shift    = (i == 1 || i == 2) ? desc->log2_chroma_h : 0;
+            int plane_size = (frame->width >> h_shift) * frame->linesize[i];
+
+            WRAP_PLANE(frame->buf[i], frame->data[i], plane_size);
+        }
+    } else {
+        int planar = av_sample_fmt_is_planar(frame->format);
+        int channels = av_get_channel_layout_nb_channels(frame->channel_layout);
+
+        planes = planar ? channels : 1;
+
+        if (planes > FF_ARRAY_ELEMS(frame->buf)) {
+            frame->nb_extended_buf = planes - FF_ARRAY_ELEMS(frame->buf);
+            frame->extended_buf = av_mallocz(sizeof(*frame->extended_buf) *
+                                             frame->nb_extended_buf);
+            if (!frame->extended_buf) {
+                ret = AVERROR(ENOMEM);
+                goto fail;
+            }
+        }
+
+        for (i = 0; i < FFMIN(planes, FF_ARRAY_ELEMS(frame->buf)); i++)
+            WRAP_PLANE(frame->buf[i], frame->extended_data[i], frame->linesize[0]);
+
+        for (i = 0; i < planes - FF_ARRAY_ELEMS(frame->buf); i++)
+            WRAP_PLANE(frame->extended_buf[i],
+                       frame->extended_data[i + FF_ARRAY_ELEMS(frame->buf)],
+                       frame->linesize[0]);
+    }
+
+    ret = av_buffersrc_add_frame(ctx, frame);
+
+fail:
+    av_buffer_unref(&dummy_buf);
+    av_frame_free(&frame);
+
+    return ret;
 }
+#endif
 
 static av_cold int init_video(AVFilterContext *ctx, const char *args)
 {
@@ -181,7 +255,7 @@ static av_cold int init_video(AVFilterContext *ctx, const char *args)
         }
     }
 
-    if (!(c->fifo = av_fifo_alloc(sizeof(AVFilterBufferRef*))))
+    if (!(c->fifo = av_fifo_alloc(sizeof(AVFrame*))))
         return AVERROR(ENOMEM);
 
     av_log(ctx, AV_LOG_VERBOSE, "w:%d h:%d pixfmt:%s\n", c->w, c->h, av_get_pix_fmt_name(c->pix_fmt));
@@ -234,7 +308,7 @@ static av_cold int init_audio(AVFilterContext *ctx, const char *args)
         goto fail;
     }
 
-    if (!(s->fifo = av_fifo_alloc(sizeof(AVFilterBufferRef*)))) {
+    if (!(s->fifo = av_fifo_alloc(sizeof(AVFrame*)))) {
         ret = AVERROR(ENOMEM);
         goto fail;
     }
@@ -255,9 +329,9 @@ static av_cold void uninit(AVFilterContext *ctx)
 {
     BufferSourceContext *s = ctx->priv;
     while (s->fifo && av_fifo_size(s->fifo)) {
-        AVFilterBufferRef *buf;
-        av_fifo_generic_read(s->fifo, &buf, sizeof(buf), NULL);
-        avfilter_unref_buffer(buf);
+        AVFrame *frame;
+        av_fifo_generic_read(s->fifo, &frame, sizeof(frame), NULL);
+        av_frame_free(&frame);
     }
     av_fifo_free(s->fifo);
     s->fifo = NULL;
@@ -317,7 +391,7 @@ static int config_props(AVFilterLink *link)
 static int request_frame(AVFilterLink *link)
 {
     BufferSourceContext *c = link->src->priv;
-    AVFilterBufferRef *buf;
+    AVFrame *frame;
     int ret = 0;
 
     if (!av_fifo_size(c->fifo)) {
@@ -325,9 +399,9 @@ static int request_frame(AVFilterLink *link)
             return AVERROR_EOF;
         return AVERROR(EAGAIN);
     }
-    av_fifo_generic_read(c->fifo, &buf, sizeof(buf), NULL);
+    av_fifo_generic_read(c->fifo, &frame, sizeof(frame), NULL);
 
-    ff_filter_frame(link, buf);
+    ff_filter_frame(link, frame);
 
     return ret;
 }
@@ -338,7 +412,7 @@ static int poll_frame(AVFilterLink *link)
     int size = av_fifo_size(c->fifo);
     if (!size && c->eof)
         return AVERROR_EOF;
-    return size/sizeof(AVFilterBufferRef*);
+    return size/sizeof(AVFrame*);
 }
 
 static const AVFilterPad avfilter_vsrc_buffer_outputs[] = {
diff --git a/libavfilter/buffersrc.h b/libavfilter/buffersrc.h
index 452c691b09..c6f7ff964b 100644
--- a/libavfilter/buffersrc.h
+++ b/libavfilter/buffersrc.h
@@ -27,24 +27,47 @@
 
 #include "avfilter.h"
 
+#if FF_API_AVFILTERBUFFER
 /**
  * Add a buffer to the filtergraph s.
  *
  * @param buf buffer containing frame data to be passed down the filtergraph.
  * This function will take ownership of buf, the user must not free it.
  * A NULL buf signals EOF -- i.e. no more frames will be sent to this filter.
+ *
+ * @deprecated use av_buffersrc_write_frame() or av_buffersrc_add_frame()
  */
+attribute_deprecated
 int av_buffersrc_buffer(AVFilterContext *s, AVFilterBufferRef *buf);
+#endif
 
 /**
  * Add a frame to the buffer source.
  *
  * @param s an instance of the buffersrc filter.
- * @param frame frame to be added.
+ * @param frame frame to be added. If the frame is reference counted, this
+ * function will make a new reference to it. Otherwise the frame data will be
+ * copied.
  *
- * @warning frame data will be memcpy()ed, which may be a big performance
- *          hit. Use av_buffersrc_buffer() to avoid copying the data.
+ * @return 0 on success, a negative AVERROR on error
  */
 int av_buffersrc_write_frame(AVFilterContext *s, const AVFrame *frame);
 
+/**
+ * Add a frame to the buffer source.
+ *
+ * @param s an instance of the buffersrc filter.
+ * @param frame frame to be added. If the frame is reference counted, this
+ * function will take ownership of the reference(s) and reset the frame.
+ * Otherwise the frame data will be copied. If this function returns an error,
+ * the input frame is not touched.
+ *
+ * @return 0 on success, a negative AVERROR on error.
+ *
+ * @note the difference between this function and av_buffersrc_write_frame() is
+ * that av_buffersrc_write_frame() creates a new reference to the input frame,
+ * while this function takes ownership of the reference passed to it.
+ */
+int av_buffersrc_add_frame(AVFilterContext *ctx, AVFrame *frame);
+
 #endif /* AVFILTER_BUFFERSRC_H */
diff --git a/libavfilter/fifo.c b/libavfilter/fifo.c
index 88c44fe3b9..b34f33b27c 100644
--- a/libavfilter/fifo.c
+++ b/libavfilter/fifo.c
@@ -35,7 +35,7 @@
 #include "video.h"
 
 typedef struct Buf {
-    AVFilterBufferRef *buf;
+    AVFrame *frame;
     struct Buf        *next;
 } Buf;
 
@@ -47,8 +47,8 @@ typedef struct {
      * When a specific number of output samples is requested, the partial
      * buffer is stored here
      */
-    AVFilterBufferRef *buf_out;
-    int allocated_samples;      ///< number of samples buf_out was allocated for
+    AVFrame *out;
+    int allocated_samples;      ///< number of samples out was allocated for
 } FifoContext;
 
 static av_cold int init(AVFilterContext *ctx, const char *args)
@@ -66,25 +66,25 @@ static av_cold void uninit(AVFilterContext *ctx)
 
     for (buf = fifo->root.next; buf; buf = tmp) {
         tmp = buf->next;
-        avfilter_unref_bufferp(&buf->buf);
+        av_frame_free(&buf->frame);
         av_free(buf);
     }
 
-    avfilter_unref_bufferp(&fifo->buf_out);
+    av_frame_free(&fifo->out);
 }
 
-static int add_to_queue(AVFilterLink *inlink, AVFilterBufferRef *buf)
+static int add_to_queue(AVFilterLink *inlink, AVFrame *frame)
 {
     FifoContext *fifo = inlink->dst->priv;
 
     fifo->last->next = av_mallocz(sizeof(Buf));
     if (!fifo->last->next) {
-        avfilter_unref_buffer(buf);
+        av_frame_free(&frame);
         return AVERROR(ENOMEM);
     }
 
     fifo->last = fifo->last->next;
-    fifo->last->buf = buf;
+    fifo->last->frame = frame;
 
     return 0;
 }
@@ -101,7 +101,7 @@ static void queue_pop(FifoContext *s)
 /**
  * Move data pointers and pts offset samples forward.
  */
-static void buffer_offset(AVFilterLink *link, AVFilterBufferRef *buf,
+static void buffer_offset(AVFilterLink *link, AVFrame *frame,
                           int offset)
 {
     int nb_channels = av_get_channel_layout_nb_channels(link->channel_layout);
@@ -110,32 +110,32 @@ static void buffer_offset(AVFilterLink *link, AVFilterBufferRef *buf,
     int block_align = av_get_bytes_per_sample(link->format) * (planar ? 1 : nb_channels);
     int i;
 
-    av_assert0(buf->audio->nb_samples > offset);
+    av_assert0(frame->nb_samples > offset);
 
     for (i = 0; i < planes; i++)
-        buf->extended_data[i] += block_align*offset;
-    if (buf->data != buf->extended_data)
-        memcpy(buf->data, buf->extended_data,
-               FFMIN(planes, FF_ARRAY_ELEMS(buf->data)) * sizeof(*buf->data));
-    buf->linesize[0] -= block_align*offset;
-    buf->audio->nb_samples -= offset;
-
-    if (buf->pts != AV_NOPTS_VALUE) {
-        buf->pts += av_rescale_q(offset, (AVRational){1, link->sample_rate},
-                                 link->time_base);
+        frame->extended_data[i] += block_align * offset;
+    if (frame->data != frame->extended_data)
+        memcpy(frame->data, frame->extended_data,
+               FFMIN(planes, FF_ARRAY_ELEMS(frame->data)) * sizeof(*frame->data));
+    frame->linesize[0] -= block_align*offset;
+    frame->nb_samples -= offset;
+
+    if (frame->pts != AV_NOPTS_VALUE) {
+        frame->pts += av_rescale_q(offset, (AVRational){1, link->sample_rate},
+                                   link->time_base);
     }
 }
 
-static int calc_ptr_alignment(AVFilterBufferRef *buf)
+static int calc_ptr_alignment(AVFrame *frame)
 {
-    int planes = av_sample_fmt_is_planar(buf->format) ?
-                 av_get_channel_layout_nb_channels(buf->audio->channel_layout) : 1;
+    int planes = av_sample_fmt_is_planar(frame->format) ?
+                 av_get_channel_layout_nb_channels(frame->channel_layout) : 1;
     int min_align = 128;
     int p;
 
     for (p = 0; p < planes; p++) {
         int cur_align = 128;
-        while ((intptr_t)buf->extended_data[p] % cur_align)
+        while ((intptr_t)frame->extended_data[p] % cur_align)
             cur_align >>= 1;
         if (cur_align < min_align)
             min_align = cur_align;
@@ -147,35 +147,34 @@ static int return_audio_frame(AVFilterContext *ctx)
 {
     AVFilterLink *link = ctx->outputs[0];
     FifoContext *s = ctx->priv;
-    AVFilterBufferRef *head = s->root.next->buf;
-    AVFilterBufferRef *buf_out;
+    AVFrame *head = s->root.next->frame;
+    AVFrame *out;
     int ret;
 
-    if (!s->buf_out &&
-        head->audio->nb_samples >= link->request_samples &&
+    if (!s->out &&
+        head->nb_samples >= link->request_samples &&
         calc_ptr_alignment(head) >= 32) {
-        if (head->audio->nb_samples == link->request_samples) {
-            buf_out = head;
+        if (head->nb_samples == link->request_samples) {
+            out = head;
             queue_pop(s);
         } else {
-            buf_out = avfilter_ref_buffer(head, AV_PERM_READ);
-            if (!buf_out)
+            out = av_frame_clone(head);
+            if (!out)
                 return AVERROR(ENOMEM);
 
-            buf_out->audio->nb_samples = link->request_samples;
+            out->nb_samples = link->request_samples;
             buffer_offset(link, head, link->request_samples);
         }
     } else {
         int nb_channels = av_get_channel_layout_nb_channels(link->channel_layout);
 
-        if (!s->buf_out) {
-            s->buf_out = ff_get_audio_buffer(link, AV_PERM_WRITE,
-                                             link->request_samples);
-            if (!s->buf_out)
+        if (!s->out) {
+            s->out = ff_get_audio_buffer(link, link->request_samples);
+            if (!s->out)
                 return AVERROR(ENOMEM);
 
-            s->buf_out->audio->nb_samples = 0;
-            s->buf_out->pts               = head->pts;
+            s->out->nb_samples = 0;
+            s->out->pts                   = head->pts;
             s->allocated_samples          = link->request_samples;
         } else if (link->request_samples != s->allocated_samples) {
             av_log(ctx, AV_LOG_ERROR, "request_samples changed before the "
@@ -183,41 +182,41 @@ static int return_audio_frame(AVFilterContext *ctx)
             return AVERROR(EINVAL);
         }
 
-        while (s->buf_out->audio->nb_samples < s->allocated_samples) {
-            int len = FFMIN(s->allocated_samples - s->buf_out->audio->nb_samples,
-                            head->audio->nb_samples);
+        while (s->out->nb_samples < s->allocated_samples) {
+            int len = FFMIN(s->allocated_samples - s->out->nb_samples,
+                            head->nb_samples);
 
-            av_samples_copy(s->buf_out->extended_data, head->extended_data,
-                            s->buf_out->audio->nb_samples, 0, len, nb_channels,
+            av_samples_copy(s->out->extended_data, head->extended_data,
+                            s->out->nb_samples, 0, len, nb_channels,
                             link->format);
-            s->buf_out->audio->nb_samples += len;
+            s->out->nb_samples += len;
 
-            if (len == head->audio->nb_samples) {
-                avfilter_unref_buffer(head);
+            if (len == head->nb_samples) {
+                av_frame_free(&head);
                 queue_pop(s);
 
                 if (!s->root.next &&
                     (ret = ff_request_frame(ctx->inputs[0])) < 0) {
                     if (ret == AVERROR_EOF) {
-                        av_samples_set_silence(s->buf_out->extended_data,
-                                               s->buf_out->audio->nb_samples,
+                        av_samples_set_silence(s->out->extended_data,
+                                               s->out->nb_samples,
                                                s->allocated_samples -
-                                               s->buf_out->audio->nb_samples,
+                                               s->out->nb_samples,
                                                nb_channels, link->format);
-                        s->buf_out->audio->nb_samples = s->allocated_samples;
+                        s->out->nb_samples = s->allocated_samples;
                         break;
                     }
                     return ret;
                 }
-                head = s->root.next->buf;
+                head = s->root.next->frame;
             } else {
                 buffer_offset(link, head, len);
             }
         }
-        buf_out = s->buf_out;
-        s->buf_out = NULL;
+        out = s->out;
+        s->out = NULL;
     }
-    return ff_filter_frame(link, buf_out);
+    return ff_filter_frame(link, out);
 }
 
 static int request_frame(AVFilterLink *outlink)
@@ -233,7 +232,7 @@ static int request_frame(AVFilterLink *outlink)
     if (outlink->request_samples) {
         return return_audio_frame(outlink->src);
     } else {
-        ret = ff_filter_frame(outlink, fifo->root.next->buf);
+        ret = ff_filter_frame(outlink, fifo->root.next->frame);
         queue_pop(fifo);
     }
 
@@ -246,7 +245,6 @@ static const AVFilterPad avfilter_vf_fifo_inputs[] = {
         .type             = AVMEDIA_TYPE_VIDEO,
         .get_video_buffer = ff_null_get_video_buffer,
         .filter_frame     = add_to_queue,
-        .rej_perms        = AV_PERM_REUSE2,
     },
     { NULL }
 };
@@ -279,7 +277,6 @@ static const AVFilterPad avfilter_af_afifo_inputs[] = {
         .type             = AVMEDIA_TYPE_AUDIO,
         .get_audio_buffer = ff_null_get_audio_buffer,
         .filter_frame     = add_to_queue,
-        .rej_perms        = AV_PERM_REUSE2,
     },
     { NULL }
 };
diff --git a/libavfilter/internal.h b/libavfilter/internal.h
index 216a3554fc..62eff729af 100644
--- a/libavfilter/internal.h
+++ b/libavfilter/internal.h
@@ -44,32 +44,12 @@ struct AVFilterPad {
     enum AVMediaType type;
 
     /**
-     * Minimum required permissions on incoming buffers. Any buffer with
-     * insufficient permissions will be automatically copied by the filter
-     * system to a new buffer which provides the needed access permissions.
-     *
-     * Input pads only.
-     */
-    int min_perms;
-
-    /**
-     * Permissions which are not accepted on incoming buffers. Any buffer
-     * which has any of these permissions set will be automatically copied
-     * by the filter system to a new buffer which does not have those
-     * permissions. This can be used to easily disallow buffers with
-     * AV_PERM_REUSE.
-     *
-     * Input pads only.
-     */
-    int rej_perms;
-
-    /**
      * Callback function to get a video buffer. If NULL, the filter system will
      * use avfilter_default_get_video_buffer().
      *
      * Input video pads only.
      */
-    AVFilterBufferRef *(*get_video_buffer)(AVFilterLink *link, int perms, int w, int h);
+    AVFrame *(*get_video_buffer)(AVFilterLink *link, int w, int h);
 
     /**
      * Callback function to get an audio buffer. If NULL, the filter system will
@@ -77,8 +57,7 @@ struct AVFilterPad {
      *
      * Input audio pads only.
      */
-    AVFilterBufferRef *(*get_audio_buffer)(AVFilterLink *link, int perms,
-                                           int nb_samples);
+    AVFrame *(*get_audio_buffer)(AVFilterLink *link, int nb_samples);
 
     /**
      * Filtering callback. This is where a filter receives a frame with
@@ -90,7 +69,7 @@ struct AVFilterPad {
      * must ensure that samplesref is properly unreferenced on error if it
      * hasn't been passed on to another filter.
      */
-    int (*filter_frame)(AVFilterLink *link, AVFilterBufferRef *frame);
+    int (*filter_frame)(AVFilterLink *link, AVFrame *frame);
 
     /**
      * Frame poll callback. This returns the number of immediately available
@@ -215,6 +194,6 @@ int ff_request_frame(AVFilterLink *link);
  * @return >= 0 on success, a negative AVERROR on error. The receiving filter
  * is responsible for unreferencing frame in case of error.
  */
-int ff_filter_frame(AVFilterLink *link, AVFilterBufferRef *frame);
+int ff_filter_frame(AVFilterLink *link, AVFrame *frame);
 
 #endif /* AVFILTER_INTERNAL_H */
diff --git a/libavfilter/split.c b/libavfilter/split.c
index c1e1669bd8..404cbcedbe 100644
--- a/libavfilter/split.c
+++ b/libavfilter/split.c
@@ -67,13 +67,13 @@ static void split_uninit(AVFilterContext *ctx)
         av_freep(&ctx->output_pads[i].name);
 }
 
-static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *frame)
+static int filter_frame(AVFilterLink *inlink, AVFrame *frame)
 {
     AVFilterContext *ctx = inlink->dst;
     int i, ret = 0;
 
     for (i = 0; i < ctx->nb_outputs; i++) {
-        AVFilterBufferRef *buf_out = avfilter_ref_buffer(frame, ~AV_PERM_WRITE);
+        AVFrame *buf_out = av_frame_clone(frame);
         if (!buf_out) {
             ret = AVERROR(ENOMEM);
             break;
@@ -83,7 +83,7 @@ static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *frame)
         if (ret < 0)
             break;
     }
-    avfilter_unref_bufferp(&frame);
+    av_frame_free(&frame);
     return ret;
 }
 
diff --git a/libavfilter/version.h b/libavfilter/version.h
index b62f25c1d9..adcfa1e003 100644
--- a/libavfilter/version.h
+++ b/libavfilter/version.h
@@ -52,5 +52,8 @@
 #ifndef FF_API_FOO_COUNT
 #define FF_API_FOO_COUNT                    (LIBAVFILTER_VERSION_MAJOR < 4)
 #endif
+#ifndef FF_API_AVFILTERBUFFER
+#define FF_API_AVFILTERBUFFER               (LIBAVFILTER_VERSION_MAJOR < 4)
+#endif
 
 #endif /* AVFILTER_VERSION_H */
diff --git a/libavfilter/vf_aspect.c b/libavfilter/vf_aspect.c
index d7e851c928..71fb4c0fce 100644
--- a/libavfilter/vf_aspect.c
+++ b/libavfilter/vf_aspect.c
@@ -65,11 +65,11 @@ static av_cold int init(AVFilterContext *ctx, const char *args)
     return 0;
 }
 
-static int filter_frame(AVFilterLink *link, AVFilterBufferRef *frame)
+static int filter_frame(AVFilterLink *link, AVFrame *frame)
 {
     AspectContext *aspect = link->dst->priv;
 
-    frame->video->pixel_aspect = aspect->aspect;
+    frame->sample_aspect_ratio = aspect->aspect;
     return ff_filter_frame(link->dst->outputs[0], frame);
 }
 
diff --git a/libavfilter/vf_blackframe.c b/libavfilter/vf_blackframe.c
index 275ebb2070..5181392b1b 100644
--- a/libavfilter/vf_blackframe.c
+++ b/libavfilter/vf_blackframe.c
@@ -78,7 +78,7 @@ static av_cold int init(AVFilterContext *ctx, const char *args)
     return 0;
 }
 
-static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *frame)
+static int filter_frame(AVFilterLink *inlink, AVFrame *frame)
 {
     AVFilterContext *ctx = inlink->dst;
     BlackFrameContext *blackframe = ctx->priv;
@@ -86,7 +86,7 @@ static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *frame)
     int pblack = 0;
     uint8_t *p = frame->data[0];
 
-    for (i = 0; i < frame->video->h; i++) {
+    for (i = 0; i < frame->height; i++) {
         for (x = 0; x < inlink->w; x++)
             blackframe->nblack += p[x] < blackframe->bthresh;
         p += frame->linesize[0];
@@ -94,8 +94,8 @@ static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *frame)
 
     pblack = blackframe->nblack * 100 / (inlink->w * inlink->h);
     if (pblack >= blackframe->bamount)
-        av_log(ctx, AV_LOG_INFO, "frame:%u pblack:%u pos:%"PRId64" pts:%"PRId64" t:%f\n",
-               blackframe->frame, pblack, frame->pos, frame->pts,
+        av_log(ctx, AV_LOG_INFO, "frame:%u pblack:%u pts:%"PRId64" t:%f\n",
+               blackframe->frame, pblack, frame->pts,
                frame->pts == AV_NOPTS_VALUE ? -1 : frame->pts * av_q2d(inlink->time_base));
 
     blackframe->frame++;
diff --git a/libavfilter/vf_boxblur.c b/libavfilter/vf_boxblur.c
index d72c602d06..e839c12060 100644
--- a/libavfilter/vf_boxblur.c
+++ b/libavfilter/vf_boxblur.c
@@ -307,23 +307,23 @@ static void vblur(uint8_t *dst, int dst_linesize, const uint8_t *src, int src_li
                    h, radius, power, temp);
 }
 
-static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *in)
+static int filter_frame(AVFilterLink *inlink, AVFrame *in)
 {
     AVFilterContext *ctx = inlink->dst;
     BoxBlurContext *boxblur = ctx->priv;
     AVFilterLink *outlink = inlink->dst->outputs[0];
-    AVFilterBufferRef *out;
+    AVFrame *out;
     int plane;
-    int cw = inlink->w >> boxblur->hsub, ch = in->video->h >> boxblur->vsub;
+    int cw = inlink->w >> boxblur->hsub, ch = in->height >> boxblur->vsub;
     int w[4] = { inlink->w, cw, cw, inlink->w };
-    int h[4] = { in->video->h, ch, ch, in->video->h };
+    int h[4] = { in->height, ch, ch, in->height };
 
-    out = ff_get_video_buffer(outlink, AV_PERM_WRITE, outlink->w, outlink->h);
+    out = ff_get_video_buffer(outlink, outlink->w, outlink->h);
     if (!out) {
-        avfilter_unref_bufferp(&in);
+        av_frame_free(&in);
         return AVERROR(ENOMEM);
     }
-    avfilter_copy_buffer_ref_props(out, in);
+    av_frame_copy_props(out, in);
 
     for (plane = 0; in->data[plane] && plane < 4; plane++)
         hblur(out->data[plane], out->linesize[plane],
@@ -337,7 +337,7 @@ static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *in)
               w[plane], h[plane], boxblur->radius[plane], boxblur->power[plane],
               boxblur->temp);
 
-    avfilter_unref_bufferp(&in);
+    av_frame_free(&in);
 
     return ff_filter_frame(outlink, out);
 }
@@ -348,7 +348,6 @@ static const AVFilterPad avfilter_vf_boxblur_inputs[] = {
         .type         = AVMEDIA_TYPE_VIDEO,
         .config_props = config_input,
         .filter_frame = filter_frame,
-        .min_perms    = AV_PERM_READ
     },
     { NULL }
 };
diff --git a/libavfilter/vf_copy.c b/libavfilter/vf_copy.c
index 8ece5cf8c5..f53395932b 100644
--- a/libavfilter/vf_copy.c
+++ b/libavfilter/vf_copy.c
@@ -21,17 +21,35 @@
  * copy video filter
  */
 
+#include "libavutil/imgutils.h"
 #include "libavutil/internal.h"
 #include "avfilter.h"
 #include "internal.h"
 #include "video.h"
 
+static int filter_frame(AVFilterLink *inlink, AVFrame *in)
+{
+    AVFilterLink *outlink = inlink->dst->outputs[0];
+    AVFrame *out = ff_get_video_buffer(outlink, in->width, in->height);
+
+    if (!out) {
+        av_frame_free(&in);
+        return AVERROR(ENOMEM);
+    }
+    av_frame_copy_props(out, in);
+    av_image_copy(out->data, out->linesize, in->data, in->linesize,
+                  in->format, in->width, in->height);
+
+    av_frame_free(&in);
+    return ff_filter_frame(outlink, out);
+}
+
 static const AVFilterPad avfilter_vf_copy_inputs[] = {
     {
         .name             = "default",
         .type             = AVMEDIA_TYPE_VIDEO,
         .get_video_buffer = ff_null_get_video_buffer,
-        .rej_perms        = ~0
+        .filter_frame     = filter_frame,
     },
     { NULL }
 };
diff --git a/libavfilter/vf_crop.c b/libavfilter/vf_crop.c
index 981dfd6cb0..2b571c4f8a 100644
--- a/libavfilter/vf_crop.c
+++ b/libavfilter/vf_crop.c
@@ -65,7 +65,6 @@ enum var_name {
     VAR_X,
     VAR_Y,
     VAR_N,
-    VAR_POS,
     VAR_T,
     VAR_VARS_NB
 };
@@ -174,7 +173,6 @@ static int config_input(AVFilterLink *link)
     crop->var_values[VAR_OUT_H] = crop->var_values[VAR_OH] = NAN;
     crop->var_values[VAR_N]     = 0;
     crop->var_values[VAR_T]     = NAN;
-    crop->var_values[VAR_POS]   = NAN;
 
     av_image_fill_max_pixsteps(crop->max_step, NULL, pix_desc);
     crop->hsub = pix_desc->log2_chroma_w;
@@ -243,19 +241,18 @@ static int config_output(AVFilterLink *link)
     return 0;
 }
 
-static int filter_frame(AVFilterLink *link, AVFilterBufferRef *frame)
+static int filter_frame(AVFilterLink *link, AVFrame *frame)
 {
     AVFilterContext *ctx = link->dst;
     CropContext *crop = ctx->priv;
     const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(link->format);
     int i;
 
-    frame->video->w = crop->w;
-    frame->video->h = crop->h;
+    frame->width  = crop->w;
+    frame->height = crop->h;
 
     crop->var_values[VAR_T] = frame->pts == AV_NOPTS_VALUE ?
         NAN : frame->pts * av_q2d(link->time_base);
-    crop->var_values[VAR_POS] = frame->pos == -1 ? NAN : frame->pos;
     crop->var_values[VAR_X] = av_expr_eval(crop->x_pexpr, crop->var_values, NULL);
     crop->var_values[VAR_Y] = av_expr_eval(crop->y_pexpr, crop->var_values, NULL);
     crop->var_values[VAR_X] = av_expr_eval(crop->x_pexpr, crop->var_values, NULL);
diff --git a/libavfilter/vf_cropdetect.c b/libavfilter/vf_cropdetect.c
index eebd8bc144..a1d957cbaa 100644
--- a/libavfilter/vf_cropdetect.c
+++ b/libavfilter/vf_cropdetect.c
@@ -117,7 +117,7 @@ static int config_input(AVFilterLink *inlink)
     return 0;
 }
 
-static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *frame)
+static int filter_frame(AVFilterLink *inlink, AVFrame *frame)
 {
     AVFilterContext *ctx = inlink->dst;
     CropDetectContext *cd = ctx->priv;
@@ -128,36 +128,36 @@ static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *frame)
     if (++cd->frame_nb > 0) {
         // Reset the crop area every reset_count frames, if reset_count is > 0
         if (cd->reset_count > 0 && cd->frame_nb > cd->reset_count) {
-            cd->x1 = frame->video->w-1;
-            cd->y1 = frame->video->h-1;
+            cd->x1 = frame->width  - 1;
+            cd->y1 = frame->height - 1;
             cd->x2 = 0;
             cd->y2 = 0;
             cd->frame_nb = 1;
         }
 
         for (y = 0; y < cd->y1; y++) {
-            if (checkline(ctx, frame->data[0] + frame->linesize[0] * y, bpp, frame->video->w, bpp) > cd->limit) {
+            if (checkline(ctx, frame->data[0] + frame->linesize[0] * y, bpp, frame->width, bpp) > cd->limit) {
                 cd->y1 = y;
                 break;
             }
         }
 
-        for (y = frame->video->h-1; y > cd->y2; y--) {
-            if (checkline(ctx, frame->data[0] + frame->linesize[0] * y, bpp, frame->video->w, bpp) > cd->limit) {
+        for (y = frame->height - 1; y > cd->y2; y--) {
+            if (checkline(ctx, frame->data[0] + frame->linesize[0] * y, bpp, frame->width, bpp) > cd->limit) {
                 cd->y2 = y;
                 break;
             }
         }
 
         for (y = 0; y < cd->x1; y++) {
-            if (checkline(ctx, frame->data[0] + bpp*y, frame->linesize[0], frame->video->h, bpp) > cd->limit) {
+            if (checkline(ctx, frame->data[0] + bpp*y, frame->linesize[0], frame->height, bpp) > cd->limit) {
                 cd->x1 = y;
                 break;
             }
         }
 
-        for (y = frame->video->w-1; y > cd->x2; y--) {
-            if (checkline(ctx, frame->data[0] + bpp*y, frame->linesize[0], frame->video->h, bpp) > cd->limit) {
+        for (y = frame->width - 1; y > cd->x2; y--) {
+            if (checkline(ctx, frame->data[0] + bpp*y, frame->linesize[0], frame->height, bpp) > cd->limit) {
                 cd->x2 = y;
                 break;
             }
@@ -187,8 +187,8 @@ static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *frame)
         y += (shrink_by/2 + 1) & ~1;
 
         av_log(ctx, AV_LOG_INFO,
-               "x1:%d x2:%d y1:%d y2:%d w:%d h:%d x:%d y:%d pos:%"PRId64" pts:%"PRId64" t:%f crop=%d:%d:%d:%d\n",
-               cd->x1, cd->x2, cd->y1, cd->y2, w, h, x, y, frame->pos, frame->pts,
+               "x1:%d x2:%d y1:%d y2:%d w:%d h:%d x:%d y:%d pts:%"PRId64" t:%f crop=%d:%d:%d:%d\n",
+               cd->x1, cd->x2, cd->y1, cd->y2, w, h, x, y, frame->pts,
                frame->pts == AV_NOPTS_VALUE ? -1 : frame->pts * av_q2d(inlink->time_base),
                w, h, x, y);
     }
diff --git a/libavfilter/vf_delogo.c b/libavfilter/vf_delogo.c
index af479c2b12..d4135f1b9b 100644
--- a/libavfilter/vf_delogo.c
+++ b/libavfilter/vf_delogo.c
@@ -215,30 +215,30 @@ static av_cold int init(AVFilterContext *ctx, const char *args)
     return 0;
 }
 
-static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *in)
+static int filter_frame(AVFilterLink *inlink, AVFrame *in)
 {
     DelogoContext *delogo = inlink->dst->priv;
     AVFilterLink *outlink = inlink->dst->outputs[0];
     const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(inlink->format);
-    AVFilterBufferRef *out;
+    AVFrame *out;
     int hsub0 = desc->log2_chroma_w;
     int vsub0 = desc->log2_chroma_h;
     int direct = 0;
     int plane;
 
-    if ((in->perms & AV_PERM_WRITE) && !(in->perms & AV_PERM_PRESERVE)) {
+    if (av_frame_is_writable(in)) {
         direct = 1;
         out = in;
     } else {
-        out = ff_get_video_buffer(outlink, AV_PERM_WRITE, outlink->w, outlink->h);
+        out = ff_get_video_buffer(outlink, outlink->w, outlink->h);
         if (!out) {
-            avfilter_unref_bufferp(&in);
+            av_frame_free(&in);
             return AVERROR(ENOMEM);
         }
 
-        avfilter_copy_buffer_ref_props(out, in);
-        out->video->w = outlink->w;
-        out->video->h = outlink->h;
+        av_frame_copy_props(out, in);
+        out->width  = outlink->w;
+        out->height = outlink->h;
     }
 
     for (plane = 0; plane < 4 && in->data[plane]; plane++) {
@@ -255,7 +255,7 @@ static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *in)
     }
 
     if (!direct)
-        avfilter_unref_bufferp(&in);
+        av_frame_free(&in);
 
     return ff_filter_frame(outlink, out);
 }
@@ -266,8 +266,6 @@ static const AVFilterPad avfilter_vf_delogo_inputs[] = {
         .type             = AVMEDIA_TYPE_VIDEO,
         .get_video_buffer = ff_null_get_video_buffer,
         .filter_frame     = filter_frame,
-        .min_perms        = AV_PERM_WRITE | AV_PERM_READ,
-        .rej_perms        = AV_PERM_PRESERVE
     },
     { NULL }
 };
diff --git a/libavfilter/vf_drawbox.c b/libavfilter/vf_drawbox.c
index c47422ef1b..cd523c0cf2 100644
--- a/libavfilter/vf_drawbox.c
+++ b/libavfilter/vf_drawbox.c
@@ -96,20 +96,20 @@ static int config_input(AVFilterLink *inlink)
     return 0;
 }
 
-static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *frame)
+static int filter_frame(AVFilterLink *inlink, AVFrame *frame)
 {
     DrawBoxContext *drawbox = inlink->dst->priv;
     int plane, x, y, xb = drawbox->x, yb = drawbox->y;
     unsigned char *row[4];
 
-    for (y = FFMAX(yb, 0); y < frame->video->h && y < (yb + drawbox->h); y++) {
+    for (y = FFMAX(yb, 0); y < frame->height && y < (yb + drawbox->h); y++) {
         row[0] = frame->data[0] + y * frame->linesize[0];
 
         for (plane = 1; plane < 3; plane++)
             row[plane] = frame->data[plane] +
                  frame->linesize[plane] * (y >> drawbox->vsub);
 
-        for (x = FFMAX(xb, 0); x < (xb + drawbox->w) && x < frame->video->w; x++) {
+        for (x = FFMAX(xb, 0); x < (xb + drawbox->w) && x < frame->width; x++) {
             double alpha = (double)drawbox->yuv_color[A] / 255;
 
             if ((y - yb < 3) || (yb + drawbox->h - y < 4) ||
@@ -131,8 +131,7 @@ static const AVFilterPad avfilter_vf_drawbox_inputs[] = {
         .config_props     = config_input,
         .get_video_buffer = ff_null_get_video_buffer,
         .filter_frame     = filter_frame,
-        .min_perms        = AV_PERM_WRITE | AV_PERM_READ,
-        .rej_perms        = AV_PERM_PRESERVE
+        .needs_writable   = 1,
     },
     { NULL }
 };
diff --git a/libavfilter/vf_drawtext.c b/libavfilter/vf_drawtext.c
index f8800d2820..8e8ac8d554 100644
--- a/libavfilter/vf_drawtext.c
+++ b/libavfilter/vf_drawtext.c
@@ -627,19 +627,19 @@ static int config_input(AVFilterLink *inlink)
         (bitmap->buffer[(r) * bitmap->pitch + ((c)>>3)] & (0x80 >> ((c)&7))) * 255 : \
          bitmap->buffer[(r) * bitmap->pitch +  (c)]
 
-#define SET_PIXEL_YUV(picref, yuva_color, val, x, y, hsub, vsub) {           \
-    luma_pos    = ((x)          ) + ((y)          ) * picref->linesize[0]; \
+#define SET_PIXEL_YUV(frame, yuva_color, val, x, y, hsub, vsub) {           \
+    luma_pos    = ((x)          ) + ((y)          ) * frame->linesize[0]; \
     alpha = yuva_color[3] * (val) * 129;                               \
-    picref->data[0][luma_pos]    = (alpha * yuva_color[0] + (255*255*129 - alpha) * picref->data[0][luma_pos]   ) >> 23; \
+    frame->data[0][luma_pos]    = (alpha * yuva_color[0] + (255*255*129 - alpha) * frame->data[0][luma_pos]   ) >> 23; \
     if (((x) & ((1<<(hsub)) - 1)) == 0 && ((y) & ((1<<(vsub)) - 1)) == 0) {\
-        chroma_pos1 = ((x) >> (hsub)) + ((y) >> (vsub)) * picref->linesize[1]; \
-        chroma_pos2 = ((x) >> (hsub)) + ((y) >> (vsub)) * picref->linesize[2]; \
-        picref->data[1][chroma_pos1] = (alpha * yuva_color[1] + (255*255*129 - alpha) * picref->data[1][chroma_pos1]) >> 23; \
-        picref->data[2][chroma_pos2] = (alpha * yuva_color[2] + (255*255*129 - alpha) * picref->data[2][chroma_pos2]) >> 23; \
+        chroma_pos1 = ((x) >> (hsub)) + ((y) >> (vsub)) * frame->linesize[1]; \
+        chroma_pos2 = ((x) >> (hsub)) + ((y) >> (vsub)) * frame->linesize[2]; \
+        frame->data[1][chroma_pos1] = (alpha * yuva_color[1] + (255*255*129 - alpha) * frame->data[1][chroma_pos1]) >> 23; \
+        frame->data[2][chroma_pos2] = (alpha * yuva_color[2] + (255*255*129 - alpha) * frame->data[2][chroma_pos2]) >> 23; \
     }\
 }
 
-static inline int draw_glyph_yuv(AVFilterBufferRef *picref, FT_Bitmap *bitmap, unsigned int x,
+static inline int draw_glyph_yuv(AVFrame *frame, FT_Bitmap *bitmap, unsigned int x,
                                  unsigned int y, unsigned int width, unsigned int height,
                                  const uint8_t yuva_color[4], int hsub, int vsub)
 {
@@ -654,22 +654,22 @@ static inline int draw_glyph_yuv(AVFilterBufferRef *picref, FT_Bitmap *bitmap, u
             if (!src_val)
                 continue;
 
-            SET_PIXEL_YUV(picref, yuva_color, src_val, c+x, y+r, hsub, vsub);
+            SET_PIXEL_YUV(frame, yuva_color, src_val, c+x, y+r, hsub, vsub);
         }
     }
 
     return 0;
 }
 
-#define SET_PIXEL_RGB(picref, rgba_color, val, x, y, pixel_step, r_off, g_off, b_off, a_off) { \
-    p   = picref->data[0] + (x) * pixel_step + ((y) * picref->linesize[0]); \
+#define SET_PIXEL_RGB(frame, rgba_color, val, x, y, pixel_step, r_off, g_off, b_off, a_off) { \
+    p   = frame->data[0] + (x) * pixel_step + ((y) * frame->linesize[0]); \
     alpha = rgba_color[3] * (val) * 129;                              \
     *(p+r_off) = (alpha * rgba_color[0] + (255*255*129 - alpha) * *(p+r_off)) >> 23; \
     *(p+g_off) = (alpha * rgba_color[1] + (255*255*129 - alpha) * *(p+g_off)) >> 23; \
     *(p+b_off) = (alpha * rgba_color[2] + (255*255*129 - alpha) * *(p+b_off)) >> 23; \
 }
 
-static inline int draw_glyph_rgb(AVFilterBufferRef *picref, FT_Bitmap *bitmap,
+static inline int draw_glyph_rgb(AVFrame *frame, FT_Bitmap *bitmap,
                                  unsigned int x, unsigned int y,
                                  unsigned int width, unsigned int height, int pixel_step,
                                  const uint8_t rgba_color[4], const uint8_t rgba_map[4])
@@ -685,7 +685,7 @@ static inline int draw_glyph_rgb(AVFilterBufferRef *picref, FT_Bitmap *bitmap,
             if (!src_val)
                 continue;
 
-            SET_PIXEL_RGB(picref, rgba_color, src_val, c+x, y+r, pixel_step,
+            SET_PIXEL_RGB(frame, rgba_color, src_val, c+x, y+r, pixel_step,
                           rgba_map[0], rgba_map[1], rgba_map[2], rgba_map[3]);
         }
     }
@@ -693,7 +693,7 @@ static inline int draw_glyph_rgb(AVFilterBufferRef *picref, FT_Bitmap *bitmap,
     return 0;
 }
 
-static inline void drawbox(AVFilterBufferRef *picref, unsigned int x, unsigned int y,
+static inline void drawbox(AVFrame *frame, unsigned int x, unsigned int y,
                            unsigned int width, unsigned int height,
                            uint8_t *line[4], int pixel_step[4], uint8_t color[4],
                            int hsub, int vsub, int is_rgba_packed, uint8_t rgba_map[4])
@@ -705,22 +705,22 @@ static inline void drawbox(AVFilterBufferRef *picref, unsigned int x, unsigned i
             uint8_t *p;
             for (j = 0; j < height; j++)
                 for (i = 0; i < width; i++)
-                    SET_PIXEL_RGB(picref, color, 255, i+x, y+j, pixel_step[0],
+                    SET_PIXEL_RGB(frame, color, 255, i+x, y+j, pixel_step[0],
                                   rgba_map[0], rgba_map[1], rgba_map[2], rgba_map[3]);
         } else {
             unsigned int luma_pos, chroma_pos1, chroma_pos2;
             for (j = 0; j < height; j++)
                 for (i = 0; i < width; i++)
-                    SET_PIXEL_YUV(picref, color, 255, i+x, y+j, hsub, vsub);
+                    SET_PIXEL_YUV(frame, color, 255, i+x, y+j, hsub, vsub);
         }
     } else {
-        ff_draw_rectangle(picref->data, picref->linesize,
+        ff_draw_rectangle(frame->data, frame->linesize,
                           line, pixel_step, hsub, vsub,
                           x, y, width, height);
     }
 }
 
-static int draw_glyphs(DrawTextContext *dtext, AVFilterBufferRef *picref,
+static int draw_glyphs(DrawTextContext *dtext, AVFrame *frame,
                        int width, int height, const uint8_t rgbcolor[4], const uint8_t yuvcolor[4], int x, int y)
 {
     char *text = HAVE_LOCALTIME_R ? dtext->expanded_text : dtext->text;
@@ -745,11 +745,11 @@ static int draw_glyphs(DrawTextContext *dtext, AVFilterBufferRef *picref,
             return AVERROR(EINVAL);
 
         if (dtext->is_packed_rgb) {
-            draw_glyph_rgb(picref, &glyph->bitmap,
+            draw_glyph_rgb(frame, &glyph->bitmap,
                            dtext->positions[i].x+x, dtext->positions[i].y+y, width, height,
                            dtext->pixel_step[0], rgbcolor, dtext->rgba_map);
         } else {
-            draw_glyph_yuv(picref, &glyph->bitmap,
+            draw_glyph_yuv(frame, &glyph->bitmap,
                            dtext->positions[i].x+x, dtext->positions[i].y+y, width, height,
                            yuvcolor, dtext->hsub, dtext->vsub);
         }
@@ -758,7 +758,7 @@ static int draw_glyphs(DrawTextContext *dtext, AVFilterBufferRef *picref,
     return 0;
 }
 
-static int draw_text(AVFilterContext *ctx, AVFilterBufferRef *picref,
+static int draw_text(AVFilterContext *ctx, AVFrame *frame,
                      int width, int height)
 {
     DrawTextContext *dtext = ctx->priv;
@@ -766,13 +766,13 @@ static int draw_text(AVFilterContext *ctx, AVFilterBufferRef *picref,
 
     /* draw box */
     if (dtext->draw_box)
-        drawbox(picref, dtext->x, dtext->y, dtext->w, dtext->h,
+        drawbox(frame, dtext->x, dtext->y, dtext->w, dtext->h,
                 dtext->box_line, dtext->pixel_step, dtext->boxcolor,
                 dtext->hsub, dtext->vsub, dtext->is_packed_rgb,
                 dtext->rgba_map);
 
     if (dtext->shadowx || dtext->shadowy) {
-        if ((ret = draw_glyphs(dtext, picref, width, height,
+        if ((ret = draw_glyphs(dtext, frame, width, height,
                                dtext->shadowcolor_rgba,
                                dtext->shadowcolor,
                                dtext->x + dtext->shadowx,
@@ -780,7 +780,7 @@ static int draw_text(AVFilterContext *ctx, AVFilterBufferRef *picref,
             return ret;
     }
 
-    if ((ret = draw_glyphs(dtext, picref, width, height,
+    if ((ret = draw_glyphs(dtext, frame, width, height,
                            dtext->fontcolor_rgba,
                            dtext->fontcolor,
                            dtext->x,
@@ -805,7 +805,7 @@ static inline int normalize_double(int *n, double d)
     return ret;
 }
 
-static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *frame)
+static int filter_frame(AVFilterLink *inlink, AVFrame *frame)
 {
     AVFilterContext *ctx = inlink->dst;
     DrawTextContext *dtext = ctx->priv;
@@ -813,7 +813,7 @@ static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *frame)
 
     if ((ret = dtext_prepare_text(ctx)) < 0) {
         av_log(ctx, AV_LOG_ERROR, "Can't draw text\n");
-        avfilter_unref_bufferp(&frame);
+        av_frame_free(&frame);
         return ret;
     }
 
@@ -848,7 +848,7 @@ static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *frame)
             dtext->x, dtext->y, dtext->x+dtext->w, dtext->y+dtext->h);
 
     if (dtext->draw)
-        draw_text(inlink->dst, frame, frame->video->w, frame->video->h);
+        draw_text(inlink->dst, frame, frame->width, frame->height);
 
     dtext->var_values[VAR_N] += 1.0;
 
@@ -862,9 +862,7 @@ static const AVFilterPad avfilter_vf_drawtext_inputs[] = {
         .get_video_buffer = ff_null_get_video_buffer,
         .filter_frame     = filter_frame,
         .config_props     = config_input,
-        .min_perms        = AV_PERM_WRITE |
-                            AV_PERM_READ,
-        .rej_perms = AV_PERM_PRESERVE
+        .needs_writable   = 1,
     },
     { NULL }
 };
diff --git a/libavfilter/vf_fade.c b/libavfilter/vf_fade.c
index f609db1248..f537404ef8 100644
--- a/libavfilter/vf_fade.c
+++ b/libavfilter/vf_fade.c
@@ -98,7 +98,7 @@ static int config_props(AVFilterLink *inlink)
     return 0;
 }
 
-static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *frame)
+static int filter_frame(AVFilterLink *inlink, AVFrame *frame)
 {
     FadeContext *fade = inlink->dst->priv;
     uint8_t *p;
@@ -106,7 +106,7 @@ static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *frame)
 
     if (fade->factor < UINT16_MAX) {
         /* luma or rgb plane */
-        for (i = 0; i < frame->video->h; i++) {
+        for (i = 0; i < frame->height; i++) {
             p = frame->data[0] + i * frame->linesize[0];
             for (j = 0; j < inlink->w * fade->bpp; j++) {
                 /* fade->factor is using 16 lower-order bits for decimal
@@ -120,7 +120,7 @@ static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *frame)
         if (frame->data[1] && frame->data[2]) {
             /* chroma planes */
             for (plane = 1; plane < 3; plane++) {
-                for (i = 0; i < frame->video->h; i++) {
+                for (i = 0; i < frame->height; i++) {
                     p = frame->data[plane] + (i >> fade->vsub) * frame->linesize[plane];
                     for (j = 0; j < inlink->w >> fade->hsub; j++) {
                         /* 8421367 = ((128 << 1) + 1) << 15. It is an integer
@@ -150,8 +150,7 @@ static const AVFilterPad avfilter_vf_fade_inputs[] = {
         .config_props     = config_props,
         .get_video_buffer = ff_null_get_video_buffer,
         .filter_frame     = filter_frame,
-        .min_perms        = AV_PERM_READ | AV_PERM_WRITE,
-        .rej_perms        = AV_PERM_PRESERVE,
+        .needs_writable   = 1,
     },
     { NULL }
 };
diff --git a/libavfilter/vf_fieldorder.c b/libavfilter/vf_fieldorder.c
index 5f0cc3be33..e7745e1690 100644
--- a/libavfilter/vf_fieldorder.c
+++ b/libavfilter/vf_fieldorder.c
@@ -113,15 +113,15 @@ static int config_input(AVFilterLink *inlink)
     return 0;
 }
 
-static AVFilterBufferRef *get_video_buffer(AVFilterLink *inlink, int perms, int w, int h)
+static AVFrame *get_video_buffer(AVFilterLink *inlink, int w, int h)
 {
     AVFilterContext   *ctx        = inlink->dst;
     AVFilterLink      *outlink    = ctx->outputs[0];
 
-    return ff_get_video_buffer(outlink, perms, w, h);
+    return ff_get_video_buffer(outlink, w, h);
 }
 
-static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *frame)
+static int filter_frame(AVFilterLink *inlink, AVFrame *frame)
 {
     AVFilterContext   *ctx     = inlink->dst;
     FieldOrderContext *s       = ctx->priv;
@@ -129,14 +129,14 @@ static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *frame)
     int h, plane, line_step, line_size, line;
     uint8_t *data;
 
-    if (!frame->video->interlaced ||
-        frame->video->top_field_first == s->dst_tff)
+    if (!frame->interlaced_frame ||
+        frame->top_field_first == s->dst_tff)
         return ff_filter_frame(outlink, frame);
 
     av_dlog(ctx,
             "picture will move %s one line\n",
             s->dst_tff ? "up" : "down");
-    h = frame->video->h;
+    h = frame->height;
     for (plane = 0; plane < 4 && frame->data[plane]; plane++) {
         line_step = frame->linesize[plane];
         line_size = s->line_size[plane];
@@ -148,7 +148,7 @@ static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *frame)
              *  The new last line is created as a copy of the
              *  penultimate line from that field. */
             for (line = 0; line < h; line++) {
-                if (1 + line < frame->video->h) {
+                if (1 + line < frame->height) {
                     memcpy(data, data + line_step, line_size);
                 } else {
                     memcpy(data, data - line_step - line_step, line_size);
@@ -172,7 +172,7 @@ static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *frame)
             }
         }
     }
-    frame->video->top_field_first = s->dst_tff;
+    frame->top_field_first = s->dst_tff;
 
     return ff_filter_frame(outlink, frame);
 }
@@ -184,8 +184,7 @@ static const AVFilterPad avfilter_vf_fieldorder_inputs[] = {
         .config_props     = config_input,
         .get_video_buffer = get_video_buffer,
         .filter_frame     = filter_frame,
-        .min_perms        = AV_PERM_READ | AV_PERM_WRITE,
-        .rej_perms        = AV_PERM_REUSE2 | AV_PERM_PRESERVE,
+        .needs_writable   = 1,
     },
     { NULL }
 };
diff --git a/libavfilter/vf_fps.c b/libavfilter/vf_fps.c
index 8fd51bd42e..74e4d248a4 100644
--- a/libavfilter/vf_fps.c
+++ b/libavfilter/vf_fps.c
@@ -84,7 +84,7 @@ static av_cold int init(AVFilterContext *ctx, const char *args)
     }
     av_opt_free(s);
 
-    if (!(s->fifo = av_fifo_alloc(2*sizeof(AVFilterBufferRef*))))
+    if (!(s->fifo = av_fifo_alloc(2*sizeof(AVFrame*))))
         return AVERROR(ENOMEM);
 
     av_log(ctx, AV_LOG_VERBOSE, "fps=%d/%d\n", s->framerate.num, s->framerate.den);
@@ -94,9 +94,9 @@ static av_cold int init(AVFilterContext *ctx, const char *args)
 static void flush_fifo(AVFifoBuffer *fifo)
 {
     while (av_fifo_size(fifo)) {
-        AVFilterBufferRef *tmp;
+        AVFrame *tmp;
         av_fifo_generic_read(fifo, &tmp, sizeof(tmp), NULL);
-        avfilter_unref_buffer(tmp);
+        av_frame_free(&tmp);
     }
 }
 
@@ -139,7 +139,7 @@ static int request_frame(AVFilterLink *outlink)
     if (ret == AVERROR_EOF && av_fifo_size(s->fifo)) {
         int i;
         for (i = 0; av_fifo_size(s->fifo); i++) {
-            AVFilterBufferRef *buf;
+            AVFrame *buf;
 
             av_fifo_generic_read(s->fifo, &buf, sizeof(buf), NULL);
             buf->pts = av_rescale_q(s->first_pts, ctx->inputs[0]->time_base,
@@ -156,13 +156,13 @@ static int request_frame(AVFilterLink *outlink)
     return ret;
 }
 
-static int write_to_fifo(AVFifoBuffer *fifo, AVFilterBufferRef *buf)
+static int write_to_fifo(AVFifoBuffer *fifo, AVFrame *buf)
 {
     int ret;
 
     if (!av_fifo_space(fifo) &&
         (ret = av_fifo_realloc2(fifo, 2*av_fifo_size(fifo)))) {
-        avfilter_unref_bufferp(&buf);
+        av_frame_free(&buf);
         return ret;
     }
 
@@ -170,7 +170,7 @@ static int write_to_fifo(AVFifoBuffer *fifo, AVFilterBufferRef *buf)
     return 0;
 }
 
-static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *buf)
+static int filter_frame(AVFilterLink *inlink, AVFrame *buf)
 {
     AVFilterContext    *ctx = inlink->dst;
     FPSContext           *s = ctx->priv;
@@ -190,7 +190,7 @@ static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *buf)
         } else {
             av_log(ctx, AV_LOG_WARNING, "Discarding initial frame(s) with no "
                    "timestamp.\n");
-            avfilter_unref_buffer(buf);
+            av_frame_free(&buf);
             s->drop++;
         }
         return 0;
@@ -207,8 +207,8 @@ static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *buf)
 
     if (delta < 1) {
         /* drop the frame and everything buffered except the first */
-        AVFilterBufferRef *tmp;
-        int drop = av_fifo_size(s->fifo)/sizeof(AVFilterBufferRef*);
+        AVFrame *tmp;
+        int drop = av_fifo_size(s->fifo)/sizeof(AVFrame*);
 
         av_log(ctx, AV_LOG_DEBUG, "Dropping %d frame(s).\n", drop);
         s->drop += drop;
@@ -217,18 +217,18 @@ static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *buf)
         flush_fifo(s->fifo);
         ret = write_to_fifo(s->fifo, tmp);
 
-        avfilter_unref_buffer(buf);
+        av_frame_free(&buf);
         return ret;
     }
 
     /* can output >= 1 frames */
     for (i = 0; i < delta; i++) {
-        AVFilterBufferRef *buf_out;
+        AVFrame *buf_out;
         av_fifo_generic_read(s->fifo, &buf_out, sizeof(buf_out), NULL);
 
         /* duplicate the frame if needed */
         if (!av_fifo_size(s->fifo) && i < delta - 1) {
-            AVFilterBufferRef *dup = avfilter_ref_buffer(buf_out, AV_PERM_READ);
+            AVFrame *dup = av_frame_clone(buf_out);
 
             av_log(ctx, AV_LOG_DEBUG, "Duplicating frame.\n");
             if (dup)
@@ -237,8 +237,8 @@ static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *buf)
                 ret = AVERROR(ENOMEM);
 
             if (ret < 0) {
-                avfilter_unref_bufferp(&buf_out);
-                avfilter_unref_bufferp(&buf);
+                av_frame_free(&buf_out);
+                av_frame_free(&buf);
                 return ret;
             }
 
@@ -249,7 +249,7 @@ static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *buf)
                                     outlink->time_base) + s->frames_out;
 
         if ((ret = ff_filter_frame(outlink, buf_out)) < 0) {
-            avfilter_unref_bufferp(&buf);
+            av_frame_free(&buf);
             return ret;
         }
 
diff --git a/libavfilter/vf_frei0r.c b/libavfilter/vf_frei0r.c
index 955d0b9f83..fab0d157ec 100644
--- a/libavfilter/vf_frei0r.c
+++ b/libavfilter/vf_frei0r.c
@@ -346,24 +346,24 @@ static int query_formats(AVFilterContext *ctx)
     return 0;
 }
 
-static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *in)
+static int filter_frame(AVFilterLink *inlink, AVFrame *in)
 {
     Frei0rContext *frei0r = inlink->dst->priv;
     AVFilterLink *outlink = inlink->dst->outputs[0];
-    AVFilterBufferRef *out;
+    AVFrame *out;
 
-    out = ff_get_video_buffer(outlink, AV_PERM_WRITE, outlink->w, outlink->h);
+    out = ff_get_video_buffer(outlink, outlink->w, outlink->h);
     if (!out) {
-        avfilter_unref_bufferp(&in);
+        av_frame_free(&in);
         return AVERROR(ENOMEM);
     }
-    avfilter_copy_buffer_ref_props(out, in);
+    av_frame_copy_props(out, in);
 
     frei0r->update(frei0r->instance, in->pts * av_q2d(inlink->time_base) * 1000,
                    (const uint32_t *)in->data[0],
                    (uint32_t *)out->data[0]);
 
-    avfilter_unref_bufferp(&in);
+    av_frame_free(&in);
 
     return ff_filter_frame(outlink, out);
 }
@@ -374,7 +374,6 @@ static const AVFilterPad avfilter_vf_frei0r_inputs[] = {
         .type         = AVMEDIA_TYPE_VIDEO,
         .config_props = config_input_props,
         .filter_frame = filter_frame,
-        .min_perms    = AV_PERM_READ
     },
     { NULL }
 };
@@ -454,19 +453,18 @@ static int source_config_props(AVFilterLink *outlink)
 static int source_request_frame(AVFilterLink *outlink)
 {
     Frei0rContext *frei0r = outlink->src->priv;
-    AVFilterBufferRef *picref = ff_get_video_buffer(outlink, AV_PERM_WRITE, outlink->w, outlink->h);
+    AVFrame *frame = ff_get_video_buffer(outlink, outlink->w, outlink->h);
 
-    if (!picref)
+    if (!frame)
         return AVERROR(ENOMEM);
 
-    picref->video->pixel_aspect = (AVRational) {1, 1};
-    picref->pts = frei0r->pts++;
-    picref->pos = -1;
+    frame->sample_aspect_ratio = (AVRational) {1, 1};
+    frame->pts = frei0r->pts++;
 
-    frei0r->update(frei0r->instance, av_rescale_q(picref->pts, frei0r->time_base, (AVRational){1,1000}),
-                   NULL, (uint32_t *)picref->data[0]);
+    frei0r->update(frei0r->instance, av_rescale_q(frame->pts, frei0r->time_base, (AVRational){1,1000}),
+                   NULL, (uint32_t *)frame->data[0]);
 
-    return ff_filter_frame(outlink, picref);
+    return ff_filter_frame(outlink, frame);
 }
 
 static const AVFilterPad avfilter_vsrc_frei0r_src_outputs[] = {
diff --git a/libavfilter/vf_gradfun.c b/libavfilter/vf_gradfun.c
index 2c9a9763c1..05412cffcc 100644
--- a/libavfilter/vf_gradfun.c
+++ b/libavfilter/vf_gradfun.c
@@ -182,26 +182,26 @@ static int config_input(AVFilterLink *inlink)
     return 0;
 }
 
-static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *in)
+static int filter_frame(AVFilterLink *inlink, AVFrame *in)
 {
     GradFunContext *gf = inlink->dst->priv;
     AVFilterLink *outlink = inlink->dst->outputs[0];
-    AVFilterBufferRef *out;
+    AVFrame *out;
     int p, direct;
 
-    if ((in->perms & AV_PERM_WRITE) && !(in->perms & AV_PERM_PRESERVE)) {
+    if (av_frame_is_writable(in)) {
         direct = 1;
         out = in;
     } else {
-        out = ff_get_video_buffer(outlink, AV_PERM_WRITE, outlink->w, outlink->h);
+        out = ff_get_video_buffer(outlink, outlink->w, outlink->h);
         if (!out) {
-            avfilter_unref_bufferp(&in);
+            av_frame_free(&in);
             return AVERROR(ENOMEM);
         }
 
-        avfilter_copy_buffer_ref_props(out, in);
-        out->video->w = outlink->w;
-        out->video->h = outlink->h;
+        av_frame_copy_props(out, in);
+        out->width  = outlink->w;
+        out->height = outlink->h;
     }
 
     for (p = 0; p < 4 && in->data[p]; p++) {
@@ -221,7 +221,7 @@ static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *in)
     }
 
     if (!direct)
-        avfilter_unref_bufferp(&in);
+        av_frame_free(&in);
 
     return ff_filter_frame(outlink, out);
 }
@@ -232,7 +232,6 @@ static const AVFilterPad avfilter_vf_gradfun_inputs[] = {
         .type         = AVMEDIA_TYPE_VIDEO,
         .config_props = config_input,
         .filter_frame = filter_frame,
-        .min_perms    = AV_PERM_READ,
     },
     { NULL }
 };
diff --git a/libavfilter/vf_hflip.c b/libavfilter/vf_hflip.c
index 85a1d92720..c8a1ca252e 100644
--- a/libavfilter/vf_hflip.c
+++ b/libavfilter/vf_hflip.c
@@ -84,21 +84,21 @@ static int config_props(AVFilterLink *inlink)
     return 0;
 }
 
-static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *in)
+static int filter_frame(AVFilterLink *inlink, AVFrame *in)
 {
     AVFilterContext *ctx  = inlink->dst;
     FlipContext *flip     = ctx->priv;
     AVFilterLink *outlink = ctx->outputs[0];
-    AVFilterBufferRef *out;
+    AVFrame *out;
     uint8_t *inrow, *outrow;
     int i, j, plane, step, hsub, vsub;
 
-    out = ff_get_video_buffer(outlink, AV_PERM_WRITE, outlink->w, outlink->h);
+    out = ff_get_video_buffer(outlink, outlink->w, outlink->h);
     if (!out) {
-        avfilter_unref_bufferp(&in);
+        av_frame_free(&in);
         return AVERROR(ENOMEM);
     }
-    avfilter_copy_buffer_ref_props(out, in);
+    av_frame_copy_props(out, in);
 
     for (plane = 0; plane < 4 && in->data[plane]; plane++) {
         step = flip->max_step[plane];
@@ -107,7 +107,7 @@ static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *in)
 
         outrow = out->data[plane];
         inrow  = in ->data[plane] + ((inlink->w >> hsub) - 1) * step;
-        for (i = 0; i < in->video->h >> vsub; i++) {
+        for (i = 0; i < in->height >> vsub; i++) {
             switch (step) {
             case 1:
                 for (j = 0; j < (inlink->w >> hsub); j++)
@@ -153,7 +153,7 @@ static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *in)
         }
     }
 
-    avfilter_unref_bufferp(&in);
+    av_frame_free(&in);
     return ff_filter_frame(outlink, out);
 }
 
@@ -163,7 +163,6 @@ static const AVFilterPad avfilter_vf_hflip_inputs[] = {
         .type         = AVMEDIA_TYPE_VIDEO,
         .filter_frame = filter_frame,
         .config_props = config_props,
-        .min_perms    = AV_PERM_READ,
     },
     { NULL }
 };
diff --git a/libavfilter/vf_hqdn3d.c b/libavfilter/vf_hqdn3d.c
index 6161b5e6f4..3251b6ffc1 100644
--- a/libavfilter/vf_hqdn3d.c
+++ b/libavfilter/vf_hqdn3d.c
@@ -305,39 +305,39 @@ static int config_input(AVFilterLink *inlink)
     return 0;
 }
 
-static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *in)
+static int filter_frame(AVFilterLink *inlink, AVFrame *in)
 {
     HQDN3DContext *hqdn3d = inlink->dst->priv;
     AVFilterLink *outlink = inlink->dst->outputs[0];
-    AVFilterBufferRef *out;
+    AVFrame *out;
     int direct, c;
 
-    if ((in->perms & AV_PERM_WRITE) && !(in->perms & AV_PERM_PRESERVE)) {
+    if (av_frame_is_writable(in)) {
         direct = 1;
         out = in;
     } else {
-        out = ff_get_video_buffer(outlink, AV_PERM_WRITE, outlink->w, outlink->h);
+        out = ff_get_video_buffer(outlink, outlink->w, outlink->h);
         if (!out) {
-            avfilter_unref_bufferp(&in);
+            av_frame_free(&in);
             return AVERROR(ENOMEM);
         }
 
-        avfilter_copy_buffer_ref_props(out, in);
-        out->video->w = outlink->w;
-        out->video->h = outlink->h;
+        av_frame_copy_props(out, in);
+        out->width  = outlink->w;
+        out->height = outlink->h;
     }
 
     for (c = 0; c < 3; c++) {
         denoise(hqdn3d, in->data[c], out->data[c],
                 hqdn3d->line, &hqdn3d->frame_prev[c],
-                in->video->w >> (!!c * hqdn3d->hsub),
-                in->video->h >> (!!c * hqdn3d->vsub),
+                in->width  >> (!!c * hqdn3d->hsub),
+                in->height >> (!!c * hqdn3d->vsub),
                 in->linesize[c], out->linesize[c],
                 hqdn3d->coefs[c?2:0], hqdn3d->coefs[c?3:1]);
     }
 
     if (!direct)
-        avfilter_unref_bufferp(&in);
+        av_frame_free(&in);
 
     return ff_filter_frame(outlink, out);
 }
diff --git a/libavfilter/vf_libopencv.c b/libavfilter/vf_libopencv.c
index 25b1c5030b..56bc6daea1 100644
--- a/libavfilter/vf_libopencv.c
+++ b/libavfilter/vf_libopencv.c
@@ -35,7 +35,7 @@
 #include "internal.h"
 #include "video.h"
 
-static void fill_iplimage_from_picref(IplImage *img, const AVFilterBufferRef *picref, enum AVPixelFormat pixfmt)
+static void fill_iplimage_from_frame(IplImage *img, const AVFrame *frame, enum AVPixelFormat pixfmt)
 {
     IplImage *tmpimg;
     int depth, channels_nb;
@@ -45,18 +45,18 @@ static void fill_iplimage_from_picref(IplImage *img, const AVFilterBufferRef *pi
     else if (pixfmt == AV_PIX_FMT_BGR24) { depth = IPL_DEPTH_8U;  channels_nb = 3; }
     else return;
 
-    tmpimg = cvCreateImageHeader((CvSize){picref->video->w, picref->video->h}, depth, channels_nb);
+    tmpimg = cvCreateImageHeader((CvSize){frame->width, frame->height}, depth, channels_nb);
     *img = *tmpimg;
-    img->imageData = img->imageDataOrigin = picref->data[0];
+    img->imageData = img->imageDataOrigin = frame->data[0];
     img->dataOrder = IPL_DATA_ORDER_PIXEL;
     img->origin    = IPL_ORIGIN_TL;
-    img->widthStep = picref->linesize[0];
+    img->widthStep = frame->linesize[0];
 }
 
-static void fill_picref_from_iplimage(AVFilterBufferRef *picref, const IplImage *img, enum AVPixelFormat pixfmt)
+static void fill_frame_from_iplimage(AVFrame *frame, const IplImage *img, enum AVPixelFormat pixfmt)
 {
-    picref->linesize[0] = img->widthStep;
-    picref->data[0]     = img->imageData;
+    frame->linesize[0] = img->widthStep;
+    frame->data[0]     = img->imageData;
 }
 
 static int query_formats(AVFilterContext *ctx)
@@ -351,27 +351,27 @@ static av_cold void uninit(AVFilterContext *ctx)
     memset(ocv, 0, sizeof(*ocv));
 }
 
-static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *in)
+static int filter_frame(AVFilterLink *inlink, AVFrame *in)
 {
     AVFilterContext *ctx = inlink->dst;
     OCVContext *ocv = ctx->priv;
     AVFilterLink *outlink= inlink->dst->outputs[0];
-    AVFilterBufferRef *out;
+    AVFrame *out;
     IplImage inimg, outimg;
 
-    out = ff_get_video_buffer(outlink, AV_PERM_WRITE, outlink->w, outlink->h);
+    out = ff_get_video_buffer(outlink, outlink->w, outlink->h);
     if (!out) {
-        avfilter_unref_bufferp(&in);
+        av_frame_free(&in);
         return AVERROR(ENOMEM);
     }
-    avfilter_copy_buffer_ref_props(out, in);
+    av_frame_copy_props(out, in);
 
-    fill_iplimage_from_picref(&inimg , in , inlink->format);
-    fill_iplimage_from_picref(&outimg, out, inlink->format);
+    fill_iplimage_from_frame(&inimg , in , inlink->format);
+    fill_iplimage_from_frame(&outimg, out, inlink->format);
     ocv->end_frame_filter(ctx, &inimg, &outimg);
-    fill_picref_from_iplimage(out, &outimg, inlink->format);
+    fill_frame_from_iplimage(out, &outimg, inlink->format);
 
-    avfilter_unref_bufferp(&in);
+    av_frame_free(&in);
 
     return ff_filter_frame(outlink, out);
 }
@@ -381,7 +381,6 @@ static const AVFilterPad avfilter_vf_ocv_inputs[] = {
         .name       = "default",
         .type       = AVMEDIA_TYPE_VIDEO,
         .filter_frame = filter_frame,
-        .min_perms  = AV_PERM_READ
     },
     { NULL }
 };
diff --git a/libavfilter/vf_lut.c b/libavfilter/vf_lut.c
index f265795fa8..f0cb48bd4a 100644
--- a/libavfilter/vf_lut.c
+++ b/libavfilter/vf_lut.c
@@ -295,28 +295,28 @@ static int config_props(AVFilterLink *inlink)
     return 0;
 }
 
-static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *in)
+static int filter_frame(AVFilterLink *inlink, AVFrame *in)
 {
     AVFilterContext *ctx = inlink->dst;
     LutContext *lut = ctx->priv;
     AVFilterLink *outlink = ctx->outputs[0];
-    AVFilterBufferRef *out;
+    AVFrame *out;
     uint8_t *inrow, *outrow, *inrow0, *outrow0;
     int i, j, k, plane;
 
-    out = ff_get_video_buffer(outlink, AV_PERM_WRITE, outlink->w, outlink->h);
+    out = ff_get_video_buffer(outlink, outlink->w, outlink->h);
     if (!out) {
-        avfilter_unref_bufferp(&in);
+        av_frame_free(&in);
         return AVERROR(ENOMEM);
     }
-    avfilter_copy_buffer_ref_props(out, in);
+    av_frame_copy_props(out, in);
 
     if (lut->is_rgb) {
         /* packed */
         inrow0  = in ->data[0];
         outrow0 = out->data[0];
 
-        for (i = 0; i < in->video->h; i ++) {
+        for (i = 0; i < in->height; i ++) {
             inrow  = inrow0;
             outrow = outrow0;
             for (j = 0; j < inlink->w; j++) {
@@ -337,7 +337,7 @@ static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *in)
             inrow  = in ->data[plane];
             outrow = out->data[plane];
 
-            for (i = 0; i < in->video->h >> vsub; i ++) {
+            for (i = 0; i < in->height >> vsub; i ++) {
                 for (j = 0; j < inlink->w>>hsub; j++)
                     outrow[j] = lut->lut[plane][inrow[j]];
                 inrow  += in ->linesize[plane];
@@ -346,7 +346,7 @@ static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *in)
         }
     }
 
-    avfilter_unref_bufferp(&in);
+    av_frame_free(&in);
     return ff_filter_frame(outlink, out);
 }
 
@@ -355,7 +355,7 @@ static const AVFilterPad inputs[] = {
       .type            = AVMEDIA_TYPE_VIDEO,
       .filter_frame    = filter_frame,
       .config_props    = config_props,
-      .min_perms       = AV_PERM_READ, },
+    },
     { .name = NULL}
 };
 static const AVFilterPad outputs[] = {
diff --git a/libavfilter/vf_overlay.c b/libavfilter/vf_overlay.c
index 8741d4805d..426be0ce3f 100644
--- a/libavfilter/vf_overlay.c
+++ b/libavfilter/vf_overlay.c
@@ -70,8 +70,8 @@ typedef struct {
 
     char x_expr[256], y_expr[256];
 
-    AVFilterBufferRef *main;
-    AVFilterBufferRef *over_prev, *over_next;
+    AVFrame *main;
+    AVFrame *over_prev, *over_next;
 } OverlayContext;
 
 static av_cold int init(AVFilterContext *ctx, const char *args)
@@ -91,9 +91,9 @@ static av_cold void uninit(AVFilterContext *ctx)
 {
     OverlayContext *s = ctx->priv;
 
-    avfilter_unref_bufferp(&s->main);
-    avfilter_unref_bufferp(&s->over_prev);
-    avfilter_unref_bufferp(&s->over_next);
+    av_frame_free(&s->main);
+    av_frame_free(&s->over_prev);
+    av_frame_free(&s->over_next);
 }
 
 static int query_formats(AVFilterContext *ctx)
@@ -194,17 +194,17 @@ static int config_output(AVFilterLink *outlink)
 }
 
 static void blend_frame(AVFilterContext *ctx,
-                        AVFilterBufferRef *dst, AVFilterBufferRef *src,
+                        AVFrame *dst, AVFrame *src,
                         int x, int y)
 {
     OverlayContext *over = ctx->priv;
     int i, j, k;
     int width, height;
-    int overlay_end_y = y + src->video->h;
+    int overlay_end_y = y + src->height;
     int end_y, start_y;
 
-    width = FFMIN(dst->video->w - x, src->video->w);
-    end_y = FFMIN(dst->video->h, overlay_end_y);
+    width = FFMIN(dst->width - x, src->width);
+    end_y = FFMIN(dst->height, overlay_end_y);
     start_y = FFMAX(y, 0);
     height = end_y - start_y;
 
@@ -269,7 +269,7 @@ static void blend_frame(AVFilterContext *ctx,
     }
 }
 
-static int filter_frame_main(AVFilterLink *inlink, AVFilterBufferRef *frame)
+static int filter_frame_main(AVFilterLink *inlink, AVFrame *frame)
 {
     OverlayContext *s = inlink->dst->priv;
 
@@ -279,7 +279,7 @@ static int filter_frame_main(AVFilterLink *inlink, AVFilterBufferRef *frame)
     return 0;
 }
 
-static int filter_frame_overlay(AVFilterLink *inlink, AVFilterBufferRef *frame)
+static int filter_frame_overlay(AVFilterLink *inlink, AVFrame *frame)
 {
     OverlayContext *s = inlink->dst->priv;
 
@@ -335,8 +335,8 @@ static int request_frame(AVFilterLink *outlink)
     while (s->main->pts != AV_NOPTS_VALUE &&
            s->over_next->pts != AV_NOPTS_VALUE &&
            av_compare_ts(s->over_next->pts, tb_over, s->main->pts, tb_main) < 0) {
-        avfilter_unref_bufferp(&s->over_prev);
-        FFSWAP(AVFilterBufferRef*, s->over_prev, s->over_next);
+        av_frame_free(&s->over_prev);
+        FFSWAP(AVFrame*, s->over_prev, s->over_next);
 
         ret = ff_request_frame(ctx->inputs[OVERLAY]);
         if (ret == AVERROR_EOF)
@@ -349,8 +349,8 @@ static int request_frame(AVFilterLink *outlink)
         s->over_next->pts == AV_NOPTS_VALUE ||
         !av_compare_ts(s->over_next->pts, tb_over, s->main->pts, tb_main)) {
         blend_frame(ctx, s->main, s->over_next, s->x, s->y);
-        avfilter_unref_bufferp(&s->over_prev);
-        FFSWAP(AVFilterBufferRef*, s->over_prev, s->over_next);
+        av_frame_free(&s->over_prev);
+        FFSWAP(AVFrame*, s->over_prev, s->over_next);
     } else if (s->over_prev) {
         blend_frame(ctx, s->main, s->over_prev, s->x, s->y);
     }
@@ -364,8 +364,7 @@ static const AVFilterPad avfilter_vf_overlay_inputs[] = {
         .type         = AVMEDIA_TYPE_VIDEO,
         .config_props = config_input_main,
         .filter_frame = filter_frame_main,
-        .min_perms    = AV_PERM_READ,
-        .rej_perms    = AV_PERM_REUSE2 | AV_PERM_PRESERVE,
+        .needs_writable = 1,
         .needs_fifo   = 1,
     },
     {
@@ -373,8 +372,6 @@ static const AVFilterPad avfilter_vf_overlay_inputs[] = {
         .type         = AVMEDIA_TYPE_VIDEO,
         .config_props = config_input_overlay,
         .filter_frame = filter_frame_overlay,
-        .min_perms    = AV_PERM_READ,
-        .rej_perms    = AV_PERM_REUSE2,
         .needs_fifo   = 1,
     },
     { NULL }
diff --git a/libavfilter/vf_pad.c b/libavfilter/vf_pad.c
index f1a890eb1d..1c969e59a9 100644
--- a/libavfilter/vf_pad.c
+++ b/libavfilter/vf_pad.c
@@ -253,101 +253,127 @@ static int config_output(AVFilterLink *outlink)
     return 0;
 }
 
-static AVFilterBufferRef *get_video_buffer(AVFilterLink *inlink, int perms, int w, int h)
+static AVFrame *get_video_buffer(AVFilterLink *inlink, int w, int h)
 {
     PadContext *pad = inlink->dst->priv;
 
-    AVFilterBufferRef *picref = ff_get_video_buffer(inlink->dst->outputs[0], perms,
-                                                    w + (pad->w - pad->in_w),
-                                                    h + (pad->h - pad->in_h));
+    AVFrame *frame = ff_get_video_buffer(inlink->dst->outputs[0],
+                                         w + (pad->w - pad->in_w),
+                                         h + (pad->h - pad->in_h));
     int plane;
 
-    if (!picref)
+    if (!frame)
         return NULL;
 
-    picref->video->w = w;
-    picref->video->h = h;
+    frame->width  = w;
+    frame->height = h;
 
-    for (plane = 0; plane < 4 && picref->data[plane]; plane++) {
+    for (plane = 0; plane < 4 && frame->data[plane]; plane++) {
         int hsub = (plane == 1 || plane == 2) ? pad->hsub : 0;
         int vsub = (plane == 1 || plane == 2) ? pad->vsub : 0;
 
-        picref->data[plane] += (pad->x >> hsub) * pad->line_step[plane] +
-            (pad->y >> vsub) * picref->linesize[plane];
+        frame->data[plane] += (pad->x >> hsub) * pad->line_step[plane] +
+            (pad->y >> vsub) * frame->linesize[plane];
     }
 
-    return picref;
+    return frame;
 }
 
-static int does_clip(PadContext *pad, AVFilterBufferRef *outpicref, int plane, int hsub, int vsub, int x, int y)
+/* check whether each plane in this buffer can be padded without copying */
+static int buffer_needs_copy(PadContext *s, AVFrame *frame, AVBufferRef *buf)
 {
-    int64_t x_in_buf, y_in_buf;
+    int planes[4] = { -1, -1, -1, -1}, *p = planes;
+    int i, j;
 
-    x_in_buf =  outpicref->data[plane] - outpicref->buf->data[plane]
-             +  (x >> hsub) * pad      ->line_step[plane]
-             +  (y >> vsub) * outpicref->linesize [plane];
+    /* get all planes in this buffer */
+    for (i = 0; i < FF_ARRAY_ELEMS(planes) && frame->data[i]; i++) {
+        if (av_frame_get_plane_buffer(frame, i) == buf)
+            *p++ = i;
+    }
 
-    if(x_in_buf < 0 || x_in_buf % pad->line_step[plane])
-        return 1;
-    x_in_buf /= pad->line_step[plane];
+    /* for each plane in this buffer, check that it can be padded without
+     * going over buffer bounds or other planes */
+    for (i = 0; i < FF_ARRAY_ELEMS(planes) && planes[i] >= 0; i++) {
+        int hsub = (planes[i] == 1 || planes[i] == 2) ? s->hsub : 0;
+        int vsub = (planes[i] == 1 || planes[i] == 2) ? s->vsub : 0;
+
+        uint8_t *start = frame->data[planes[i]];
+        uint8_t *end   = start + (frame->height >> hsub) *
+                                 frame->linesize[planes[i]];
+
+        /* amount of free space needed before the start and after the end
+         * of the plane */
+        ptrdiff_t req_start = (s->x >> hsub) * s->line_step[planes[i]] +
+                              (s->y >> vsub) * frame->linesize[planes[i]];
+        ptrdiff_t req_end   = ((s->w - s->x - frame->width) >> hsub) *
+                              s->line_step[planes[i]] +
+                              (s->y >> vsub) * frame->linesize[planes[i]];
+
+        if (frame->linesize[planes[i]] < (s->w >> hsub) * s->line_step[planes[i]])
+            return 1;
+        if (start - buf->data < req_start ||
+            (buf->data + buf->size) - end < req_end)
+            return 1;
+
+#define SIGN(x) ((x) > 0 ? 1 : -1)
+        for (j = 0; j < FF_ARRAY_ELEMS(planes) & planes[j] >= 0; j++) {
+            int hsub1 = (planes[j] == 1 || planes[j] == 2) ? s->hsub : 0;
+            uint8_t *start1 = frame->data[planes[j]];
+            uint8_t *end1   = start1 + (frame->height >> hsub1) *
+                                       frame->linesize[planes[j]];
+            if (i == j)
+                continue;
+
+            if (SIGN(start - end1) != SIGN(start - end1 - req_start) ||
+                SIGN(end - start1) != SIGN(end - start1 + req_end))
+                return 1;
+        }
+    }
 
-    av_assert0(outpicref->buf->linesize[plane]>0); //while reference can use negative linesize the main buffer should not
+    return 0;
+}
 
-    y_in_buf = x_in_buf / outpicref->buf->linesize[plane];
-    x_in_buf %= outpicref->buf->linesize[plane];
+static int frame_needs_copy(PadContext *s, AVFrame *frame)
+{
+    int i;
 
-    if(   y_in_buf<<vsub >= outpicref->buf->h
-       || x_in_buf<<hsub >= outpicref->buf->w)
+    if (!av_frame_is_writable(frame))
         return 1;
+
+    for (i = 0; i < FF_ARRAY_ELEMS(frame->buf) && frame->buf[i]; i++)
+        if (buffer_needs_copy(s, frame, frame->buf[i]))
+            return 1;
     return 0;
 }
 
-static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *in)
+static int filter_frame(AVFilterLink *inlink, AVFrame *in)
 {
     PadContext *pad = inlink->dst->priv;
-    AVFilterBufferRef *out = avfilter_ref_buffer(in, ~0);
-    int plane, needs_copy;
-
-    if (!out) {
-        avfilter_unref_bufferp(&in);
-        return AVERROR(ENOMEM);
-    }
+    AVFrame *out;
+    int needs_copy = frame_needs_copy(pad, in);
 
-    for (plane = 0; plane < 4 && out->data[plane]; plane++) {
-        int hsub = (plane == 1 || plane == 2) ? pad->hsub : 0;
-        int vsub = (plane == 1 || plane == 2) ? pad->vsub : 0;
-
-        av_assert0(out->buf->w > 0 && out->buf->h > 0);
-
-        if (out->format != out->buf->format) //unsupported currently
-            break;
-
-        out->data[plane] -= (pad->x  >> hsub) * pad->line_step[plane] +
-                            (pad->y  >> vsub) * out->linesize [plane];
-
-        if (does_clip(pad, out, plane, hsub, vsub, 0,                   0) ||
-            does_clip(pad, out, plane, hsub, vsub, 0,          pad->h - 1) ||
-            does_clip(pad, out, plane, hsub, vsub, pad->w - 1,          0) ||
-            does_clip(pad, out, plane, hsub, vsub, pad->w - 1, pad->h - 1))
-            break;
-    }
-    needs_copy = plane < 4 && out->data[plane];
     if (needs_copy) {
         av_log(inlink->dst, AV_LOG_DEBUG, "Direct padding impossible allocating new frame\n");
-        avfilter_unref_buffer(out);
-        out = ff_get_video_buffer(inlink->dst->outputs[0], AV_PERM_WRITE | AV_PERM_NEG_LINESIZES,
+        out = ff_get_video_buffer(inlink->dst->outputs[0],
                                   FFMAX(inlink->w, pad->w),
                                   FFMAX(inlink->h, pad->h));
         if (!out) {
-            avfilter_unref_bufferp(&in);
+            av_frame_free(&in);
             return AVERROR(ENOMEM);
         }
 
-        avfilter_copy_buffer_ref_props(out, in);
-    }
+        av_frame_copy_props(out, in);
+    } else {
+        int i;
 
-    out->video->w = pad->w;
-    out->video->h = pad->h;
+        out = in;
+        for (i = 0; i < FF_ARRAY_ELEMS(out->data) && out->data[i]; i++) {
+            int hsub = (i == 1 || i == 2) ? pad->hsub : 0;
+            int vsub = (i == 1 || i == 2) ? pad->vsub : 0;
+            out->data[i] -= (pad->x >> hsub) * pad->line_step[i] +
+                            (pad->y >> vsub) * out->linesize[i];
+        }
+    }
 
     /* top bar */
     if (pad->y) {
@@ -365,21 +391,25 @@ static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *in)
 
     /* left border */
     ff_draw_rectangle(out->data, out->linesize, pad->line, pad->line_step,
-                      pad->hsub, pad->vsub, 0, pad->y, pad->x, in->video->h);
+                      pad->hsub, pad->vsub, 0, pad->y, pad->x, in->height);
 
     if (needs_copy) {
         ff_copy_rectangle(out->data, out->linesize, in->data, in->linesize,
                           pad->line_step, pad->hsub, pad->vsub,
-                          pad->x, pad->y, 0, in->video->w, in->video->h);
+                          pad->x, pad->y, 0, in->width, in->height);
     }
 
     /* right border */
     ff_draw_rectangle(out->data, out->linesize,
                       pad->line, pad->line_step, pad->hsub, pad->vsub,
                       pad->x + pad->in_w, pad->y, pad->w - pad->x - pad->in_w,
-                      in->video->h);
+                      in->height);
+
+    out->width  = pad->w;
+    out->height = pad->h;
 
-    avfilter_unref_bufferp(&in);
+    if (in != out)
+        av_frame_free(&in);
     return ff_filter_frame(inlink->dst->outputs[0], out);
 }
 
diff --git a/libavfilter/vf_pixdesctest.c b/libavfilter/vf_pixdesctest.c
index a1e982c9bc..65f7eceb72 100644
--- a/libavfilter/vf_pixdesctest.c
+++ b/libavfilter/vf_pixdesctest.c
@@ -52,21 +52,20 @@ static int config_props(AVFilterLink *inlink)
     return 0;
 }
 
-static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *in)
+static int filter_frame(AVFilterLink *inlink, AVFrame *in)
 {
     PixdescTestContext *priv = inlink->dst->priv;
     AVFilterLink *outlink    = inlink->dst->outputs[0];
-    AVFilterBufferRef *out;
+    AVFrame *out;
     int i, c, w = inlink->w, h = inlink->h;
 
-    out = ff_get_video_buffer(outlink, AV_PERM_WRITE,
-                              outlink->w, outlink->h);
+    out = ff_get_video_buffer(outlink, outlink->w, outlink->h);
     if (!out) {
-        avfilter_unref_bufferp(&in);
+        av_frame_free(&in);
         return AVERROR(ENOMEM);
     }
 
-    avfilter_copy_buffer_ref_props(out, in);
+    av_frame_copy_props(out, in);
 
     for (i = 0; i < 4; i++) {
         int h = outlink->h;
@@ -102,7 +101,7 @@ static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *in)
         }
     }
 
-    avfilter_unref_bufferp(&in);
+    av_frame_free(&in);
     return ff_filter_frame(outlink, out);
 }
 
@@ -112,7 +111,6 @@ static const AVFilterPad avfilter_vf_pixdesctest_inputs[] = {
         .type         = AVMEDIA_TYPE_VIDEO,
         .filter_frame = filter_frame,
         .config_props = config_props,
-        .min_perms    = AV_PERM_READ,
     },
     { NULL }
 };
diff --git a/libavfilter/vf_scale.c b/libavfilter/vf_scale.c
index 7f189a2215..cb2b0a3654 100644
--- a/libavfilter/vf_scale.c
+++ b/libavfilter/vf_scale.c
@@ -257,11 +257,11 @@ fail:
     return ret;
 }
 
-static int filter_frame(AVFilterLink *link, AVFilterBufferRef *in)
+static int filter_frame(AVFilterLink *link, AVFrame *in)
 {
     ScaleContext *scale = link->dst->priv;
     AVFilterLink *outlink = link->dst->outputs[0];
-    AVFilterBufferRef *out;
+    AVFrame *out;
     const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(link->format);
 
     if (!scale->sws)
@@ -270,25 +270,25 @@ static int filter_frame(AVFilterLink *link, AVFilterBufferRef *in)
     scale->hsub = desc->log2_chroma_w;
     scale->vsub = desc->log2_chroma_h;
 
-    out = ff_get_video_buffer(outlink, AV_PERM_WRITE, outlink->w, outlink->h);
+    out = ff_get_video_buffer(outlink, outlink->w, outlink->h);
     if (!out) {
-        avfilter_unref_bufferp(&in);
+        av_frame_free(&in);
         return AVERROR(ENOMEM);
     }
 
-    avfilter_copy_buffer_ref_props(out, in);
-    out->video->w = outlink->w;
-    out->video->h = outlink->h;
+    av_frame_copy_props(out, in);
+    out->width  = outlink->w;
+    out->height = outlink->h;
 
-    av_reduce(&out->video->pixel_aspect.num, &out->video->pixel_aspect.den,
-              (int64_t)in->video->pixel_aspect.num * outlink->h * link->w,
-              (int64_t)in->video->pixel_aspect.den * outlink->w * link->h,
+    av_reduce(&out->sample_aspect_ratio.num, &out->sample_aspect_ratio.den,
+              (int64_t)in->sample_aspect_ratio.num * outlink->h * link->w,
+              (int64_t)in->sample_aspect_ratio.den * outlink->w * link->h,
               INT_MAX);
 
-    sws_scale(scale->sws, in->data, in->linesize, 0, in->video->h,
+    sws_scale(scale->sws, in->data, in->linesize, 0, in->height,
               out->data, out->linesize);
 
-    avfilter_unref_bufferp(&in);
+    av_frame_free(&in);
     return ff_filter_frame(outlink, out);
 }
 
@@ -297,7 +297,6 @@ static const AVFilterPad avfilter_vf_scale_inputs[] = {
         .name        = "default",
         .type        = AVMEDIA_TYPE_VIDEO,
         .filter_frame = filter_frame,
-        .min_perms   = AV_PERM_READ,
     },
     { NULL }
 };
diff --git a/libavfilter/vf_select.c b/libavfilter/vf_select.c
index 674151d4ef..7839e44168 100644
--- a/libavfilter/vf_select.c
+++ b/libavfilter/vf_select.c
@@ -108,7 +108,6 @@ enum var_name {
     VAR_PREV_SELECTED_N,
 
     VAR_KEY,
-    VAR_POS,
 
     VAR_VARS_NB
 };
@@ -134,7 +133,7 @@ static av_cold int init(AVFilterContext *ctx, const char *args)
         return ret;
     }
 
-    select->pending_frames = av_fifo_alloc(FIFO_SIZE*sizeof(AVFilterBufferRef*));
+    select->pending_frames = av_fifo_alloc(FIFO_SIZE*sizeof(AVFrame*));
     if (!select->pending_frames) {
         av_log(ctx, AV_LOG_ERROR, "Failed to allocate pending frames buffer.\n");
         return AVERROR(ENOMEM);
@@ -181,35 +180,33 @@ static int config_input(AVFilterLink *inlink)
 #define D2TS(d)  (isnan(d) ? AV_NOPTS_VALUE : (int64_t)(d))
 #define TS2D(ts) ((ts) == AV_NOPTS_VALUE ? NAN : (double)(ts))
 
-static int select_frame(AVFilterContext *ctx, AVFilterBufferRef *picref)
+static int select_frame(AVFilterContext *ctx, AVFrame *frame)
 {
     SelectContext *select = ctx->priv;
     AVFilterLink *inlink = ctx->inputs[0];
     double res;
 
     if (isnan(select->var_values[VAR_START_PTS]))
-        select->var_values[VAR_START_PTS] = TS2D(picref->pts);
+        select->var_values[VAR_START_PTS] = TS2D(frame->pts);
     if (isnan(select->var_values[VAR_START_T]))
-        select->var_values[VAR_START_T] = TS2D(picref->pts) * av_q2d(inlink->time_base);
+        select->var_values[VAR_START_T] = TS2D(frame->pts) * av_q2d(inlink->time_base);
 
-    select->var_values[VAR_PTS] = TS2D(picref->pts);
-    select->var_values[VAR_T  ] = TS2D(picref->pts) * av_q2d(inlink->time_base);
-    select->var_values[VAR_POS] = picref->pos == -1 ? NAN : picref->pos;
-    select->var_values[VAR_PREV_PTS] = TS2D(picref ->pts);
+    select->var_values[VAR_PTS] = TS2D(frame->pts);
+    select->var_values[VAR_T  ] = TS2D(frame->pts) * av_q2d(inlink->time_base);
+    select->var_values[VAR_PREV_PTS] = TS2D(frame->pts);
 
     select->var_values[VAR_INTERLACE_TYPE] =
-        !picref->video->interlaced     ? INTERLACE_TYPE_P :
-        picref->video->top_field_first ? INTERLACE_TYPE_T : INTERLACE_TYPE_B;
-    select->var_values[VAR_PICT_TYPE] = picref->video->pict_type;
+        !frame->interlaced_frame     ? INTERLACE_TYPE_P :
+        frame->top_field_first ? INTERLACE_TYPE_T : INTERLACE_TYPE_B;
+    select->var_values[VAR_PICT_TYPE] = frame->pict_type;
 
     res = av_expr_eval(select->expr, select->var_values, NULL);
     av_log(inlink->dst, AV_LOG_DEBUG,
-           "n:%d pts:%d t:%f pos:%d interlace_type:%c key:%d pict_type:%c "
+           "n:%d pts:%d t:%f interlace_type:%c key:%d pict_type:%c "
            "-> select:%f\n",
            (int)select->var_values[VAR_N],
            (int)select->var_values[VAR_PTS],
            select->var_values[VAR_T],
-           (int)select->var_values[VAR_POS],
            select->var_values[VAR_INTERLACE_TYPE] == INTERLACE_TYPE_P ? 'P' :
            select->var_values[VAR_INTERLACE_TYPE] == INTERLACE_TYPE_T ? 'T' :
            select->var_values[VAR_INTERLACE_TYPE] == INTERLACE_TYPE_B ? 'B' : '?',
@@ -228,7 +225,7 @@ static int select_frame(AVFilterContext *ctx, AVFilterBufferRef *picref)
     return res;
 }
 
-static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *frame)
+static int filter_frame(AVFilterLink *inlink, AVFrame *frame)
 {
     SelectContext *select = inlink->dst->priv;
 
@@ -239,7 +236,7 @@ static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *frame)
             if (!av_fifo_space(select->pending_frames)) {
                 av_log(inlink->dst, AV_LOG_ERROR,
                        "Buffering limit reached, cannot cache more frames\n");
-                avfilter_unref_bufferp(&frame);
+                av_frame_free(&frame);
             } else
                 av_fifo_generic_write(select->pending_frames, &frame,
                                       sizeof(frame), NULL);
@@ -248,7 +245,7 @@ static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *frame)
         return ff_filter_frame(inlink->dst->outputs[0], frame);
     }
 
-    avfilter_unref_bufferp(&frame);
+    av_frame_free(&frame);
     return 0;
 }
 
@@ -260,10 +257,10 @@ static int request_frame(AVFilterLink *outlink)
     select->select = 0;
 
     if (av_fifo_size(select->pending_frames)) {
-        AVFilterBufferRef *picref;
+        AVFrame *frame;
 
-        av_fifo_generic_read(select->pending_frames, &picref, sizeof(picref), NULL);
-        return ff_filter_frame(outlink, picref);
+        av_fifo_generic_read(select->pending_frames, &frame, sizeof(frame), NULL);
+        return ff_filter_frame(outlink, frame);
     }
 
     while (!select->select) {
@@ -294,20 +291,20 @@ static int poll_frame(AVFilterLink *outlink)
         select->cache_frames = 0;
     }
 
-    return av_fifo_size(select->pending_frames)/sizeof(AVFilterBufferRef *);
+    return av_fifo_size(select->pending_frames)/sizeof(AVFrame*);
 }
 
 static av_cold void uninit(AVFilterContext *ctx)
 {
     SelectContext *select = ctx->priv;
-    AVFilterBufferRef *picref;
+    AVFrame *frame;
 
     av_expr_free(select->expr);
     select->expr = NULL;
 
     while (select->pending_frames &&
-           av_fifo_generic_read(select->pending_frames, &picref, sizeof(picref), NULL) == sizeof(picref))
-        avfilter_unref_buffer(picref);
+           av_fifo_generic_read(select->pending_frames, &frame, sizeof(frame), NULL) == sizeof(frame))
+        av_frame_free(&frame);
     av_fifo_free(select->pending_frames);
     select->pending_frames = NULL;
 }
diff --git a/libavfilter/vf_setpts.c b/libavfilter/vf_setpts.c
index 79cadd41ca..02fb4bd0c5 100644
--- a/libavfilter/vf_setpts.c
+++ b/libavfilter/vf_setpts.c
@@ -40,7 +40,6 @@ static const char *const var_names[] = {
     "N",           ///< frame number (starting at zero)
     "PHI",         ///< golden ratio
     "PI",          ///< greek pi
-    "POS",         ///< original position in the file of the frame
     "PREV_INPTS",  ///< previous  input PTS
     "PREV_OUTPTS", ///< previous output PTS
     "PTS",         ///< original pts in the file of the frame
@@ -57,7 +56,6 @@ enum var_name {
     VAR_N,
     VAR_PHI,
     VAR_PI,
-    VAR_POS,
     VAR_PREV_INPTS,
     VAR_PREV_OUTPTS,
     VAR_PTS,
@@ -108,7 +106,7 @@ static int config_input(AVFilterLink *inlink)
 #define D2TS(d)  (isnan(d) ? AV_NOPTS_VALUE : (int64_t)(d))
 #define TS2D(ts) ((ts) == AV_NOPTS_VALUE ? NAN : (double)(ts))
 
-static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *frame)
+static int filter_frame(AVFilterLink *inlink, AVFrame *frame)
 {
     SetPTSContext *setpts = inlink->dst->priv;
     int64_t in_pts = frame->pts;
@@ -117,9 +115,8 @@ static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *frame)
     if (isnan(setpts->var_values[VAR_STARTPTS]))
         setpts->var_values[VAR_STARTPTS] = TS2D(frame->pts);
 
-    setpts->var_values[VAR_INTERLACED] = frame->video->interlaced;
+    setpts->var_values[VAR_INTERLACED] = frame->interlaced_frame;
     setpts->var_values[VAR_PTS       ] = TS2D(frame->pts);
-    setpts->var_values[VAR_POS       ] = frame->pos == -1 ? NAN : frame->pos;
     setpts->var_values[VAR_RTCTIME   ] = av_gettime();
 
     d = av_expr_eval(setpts->expr, setpts->var_values, NULL);
@@ -127,10 +124,10 @@ static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *frame)
 
 #ifdef DEBUG
     av_log(inlink->dst, AV_LOG_DEBUG,
-           "n:%"PRId64" interlaced:%d pos:%"PRId64" pts:%"PRId64" t:%f -> pts:%"PRId64" t:%f\n",
+           "n:%"PRId64" interlaced:%d pts:%"PRId64" t:%f -> pts:%"PRId64" t:%f\n",
            (int64_t)setpts->var_values[VAR_N],
            (int)setpts->var_values[VAR_INTERLACED],
-           frame->pos, in_pts, in_pts * av_q2d(inlink->time_base),
+           in_pts, in_pts * av_q2d(inlink->time_base),
            frame->pts, frame->pts * av_q2d(inlink->time_base));
 #endif
 
diff --git a/libavfilter/vf_settb.c b/libavfilter/vf_settb.c
index a5720726e4..325dc741ce 100644
--- a/libavfilter/vf_settb.c
+++ b/libavfilter/vf_settb.c
@@ -108,7 +108,7 @@ static int config_output_props(AVFilterLink *outlink)
     return 0;
 }
 
-static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *frame)
+static int filter_frame(AVFilterLink *inlink, AVFrame *frame)
 {
     AVFilterContext *ctx = inlink->dst;
     AVFilterLink *outlink = ctx->outputs[0];
diff --git a/libavfilter/vf_showinfo.c b/libavfilter/vf_showinfo.c
index c89b0287d8..7970810c5d 100644
--- a/libavfilter/vf_showinfo.c
+++ b/libavfilter/vf_showinfo.c
@@ -41,7 +41,7 @@ static av_cold int init(AVFilterContext *ctx, const char *args)
     return 0;
 }
 
-static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *frame)
+static int filter_frame(AVFilterLink *inlink, AVFrame *frame)
 {
     AVFilterContext *ctx = inlink->dst;
     ShowInfoContext *showinfo = ctx->priv;
@@ -50,7 +50,7 @@ static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *frame)
     int i, plane, vsub = desc->log2_chroma_h;
 
     for (plane = 0; frame->data[plane] && plane < 4; plane++) {
-        size_t linesize = av_image_get_linesize(frame->format, frame->video->w, plane);
+        size_t linesize = av_image_get_linesize(frame->format, frame->width, plane);
         uint8_t *data = frame->data[plane];
         int h = plane == 1 || plane == 2 ? inlink->h >> vsub : inlink->h;
 
@@ -62,18 +62,18 @@ static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *frame)
     }
 
     av_log(ctx, AV_LOG_INFO,
-           "n:%d pts:%"PRId64" pts_time:%f pos:%"PRId64" "
+           "n:%d pts:%"PRId64" pts_time:%f "
            "fmt:%s sar:%d/%d s:%dx%d i:%c iskey:%d type:%c "
            "checksum:%u plane_checksum:[%u %u %u %u]\n",
            showinfo->frame,
-           frame->pts, frame->pts * av_q2d(inlink->time_base), frame->pos,
+           frame->pts, frame->pts * av_q2d(inlink->time_base),
            desc->name,
-           frame->video->pixel_aspect.num, frame->video->pixel_aspect.den,
-           frame->video->w, frame->video->h,
-           !frame->video->interlaced     ? 'P' :         /* Progressive  */
-           frame->video->top_field_first ? 'T' : 'B',    /* Top / Bottom */
-           frame->video->key_frame,
-           av_get_picture_type_char(frame->video->pict_type),
+           frame->sample_aspect_ratio.num, frame->sample_aspect_ratio.den,
+           frame->width, frame->height,
+           !frame->interlaced_frame ? 'P' :         /* Progressive  */
+           frame->top_field_first   ? 'T' : 'B',    /* Top / Bottom */
+           frame->key_frame,
+           av_get_picture_type_char(frame->pict_type),
            checksum, plane_checksum[0], plane_checksum[1], plane_checksum[2], plane_checksum[3]);
 
     showinfo->frame++;
@@ -86,7 +86,6 @@ static const AVFilterPad avfilter_vf_showinfo_inputs[] = {
         .type             = AVMEDIA_TYPE_VIDEO,
         .get_video_buffer = ff_null_get_video_buffer,
         .filter_frame     = filter_frame,
-        .min_perms        = AV_PERM_READ,
     },
     { NULL }
 };
diff --git a/libavfilter/vf_transpose.c b/libavfilter/vf_transpose.c
index d7a173989f..d1a77c4bfd 100644
--- a/libavfilter/vf_transpose.c
+++ b/libavfilter/vf_transpose.c
@@ -121,35 +121,35 @@ static int config_props_output(AVFilterLink *outlink)
     return 0;
 }
 
-static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *in)
+static int filter_frame(AVFilterLink *inlink, AVFrame *in)
 {
     AVFilterLink *outlink = inlink->dst->outputs[0];
     TransContext *trans = inlink->dst->priv;
-    AVFilterBufferRef *out;
+    AVFrame *out;
     int plane;
 
-    out = ff_get_video_buffer(outlink, AV_PERM_WRITE, outlink->w, outlink->h);
+    out = ff_get_video_buffer(outlink, outlink->w, outlink->h);
     if (!out) {
-        avfilter_unref_bufferp(&in);
+        av_frame_free(&in);
         return AVERROR(ENOMEM);
     }
 
     out->pts = in->pts;
 
-    if (in->video->pixel_aspect.num == 0) {
-        out->video->pixel_aspect = in->video->pixel_aspect;
+    if (in->sample_aspect_ratio.num == 0) {
+        out->sample_aspect_ratio = in->sample_aspect_ratio;
     } else {
-        out->video->pixel_aspect.num = in->video->pixel_aspect.den;
-        out->video->pixel_aspect.den = in->video->pixel_aspect.num;
+        out->sample_aspect_ratio.num = in->sample_aspect_ratio.den;
+        out->sample_aspect_ratio.den = in->sample_aspect_ratio.num;
     }
 
     for (plane = 0; out->data[plane]; plane++) {
         int hsub = plane == 1 || plane == 2 ? trans->hsub : 0;
         int vsub = plane == 1 || plane == 2 ? trans->vsub : 0;
         int pixstep = trans->pixsteps[plane];
-        int inh  = in->video->h>>vsub;
-        int outw = out->video->w>>hsub;
-        int outh = out->video->h>>vsub;
+        int inh  = in->height  >> vsub;
+        int outw = out->width  >> hsub;
+        int outh = out->height >> vsub;
         uint8_t *dst, *src;
         int dstlinesize, srclinesize;
         int x, y;
@@ -194,7 +194,7 @@ static int filter_frame(AVFilterLink *inlink, AVFilterBufferRef *in)
         }
     }
 
-    avfilter_unref_bufferp(&in);
+    av_frame_free(&in);
     return ff_filter_frame(outlink, out);
 }
 
@@ -203,7 +203,6 @@ static const AVFilterPad avfilter_vf_transpose_inputs[] = {
         .name        = "default",
         .type        = AVMEDIA_TYPE_VIDEO,
         .filter_frame = filter_frame,
-        .min_perms   = AV_PERM_READ,
     },
     { NULL }
 };
diff --git a/libavfilter/vf_unsharp.c b/libavfilter/vf_unsharp.c
index b446937898..fab1a6a815 100644
--- a/libavfilter/vf_unsharp.c
+++ b/libavfilter/vf_unsharp.c
@@ -214,26 +214,26 @@ static av_cold void uninit(AVFilterContext *ctx)
     free_filter_param(&unsharp->chroma);
 }
 
-static int filter_frame(AVFilterLink *link, AVFilterBufferRef *in)
+static int filter_frame(AVFilterLink *link, AVFrame *in)
 {
     UnsharpContext *unsharp = link->dst->priv;
     AVFilterLink *outlink   = link->dst->outputs[0];
-    AVFilterBufferRef *out;
+    AVFrame *out;
     int cw = SHIFTUP(link->w, unsharp->hsub);
     int ch = SHIFTUP(link->h, unsharp->vsub);
 
-    out = ff_get_video_buffer(outlink, AV_PERM_WRITE, outlink->w, outlink->h);
+    out = ff_get_video_buffer(outlink, outlink->w, outlink->h);
     if (!out) {
-        avfilter_unref_bufferp(&in);
+        av_frame_free(&in);
         return AVERROR(ENOMEM);
     }
-    avfilter_copy_buffer_ref_props(out, in);
+    av_frame_copy_props(out, in);
 
     apply_unsharp(out->data[0], out->linesize[0], in->data[0], in->linesize[0], link->w, link->h, &unsharp->luma);
     apply_unsharp(out->data[1], out->linesize[1], in->data[1], in->linesize[1], cw,      ch,      &unsharp->chroma);
     apply_unsharp(out->data[2], out->linesize[2], in->data[2], in->linesize[2], cw,      ch,      &unsharp->chroma);
 
-    avfilter_unref_bufferp(&in);
+    av_frame_free(&in);
     return ff_filter_frame(outlink, out);
 }
 
@@ -243,7 +243,6 @@ static const AVFilterPad avfilter_vf_unsharp_inputs[] = {
         .type         = AVMEDIA_TYPE_VIDEO,
         .filter_frame = filter_frame,
         .config_props = config_props,
-        .min_perms    = AV_PERM_READ,
     },
     { NULL }
 };
diff --git a/libavfilter/vf_vflip.c b/libavfilter/vf_vflip.c
index 5e6e9653bd..589a05916c 100644
--- a/libavfilter/vf_vflip.c
+++ b/libavfilter/vf_vflip.c
@@ -43,33 +43,29 @@ static int config_input(AVFilterLink *link)
     return 0;
 }
 
-static AVFilterBufferRef *get_video_buffer(AVFilterLink *link, int perms,
-                                        int w, int h)
+static AVFrame *get_video_buffer(AVFilterLink *link, int w, int h)
 {
     FlipContext *flip = link->dst->priv;
-    AVFilterBufferRef *picref;
+    AVFrame *frame;
     int i;
 
-    if (!(perms & AV_PERM_NEG_LINESIZES))
-        return ff_default_get_video_buffer(link, perms, w, h);
-
-    picref = ff_get_video_buffer(link->dst->outputs[0], perms, w, h);
-    if (!picref)
+    frame = ff_get_video_buffer(link->dst->outputs[0], w, h);
+    if (!frame)
         return NULL;
 
     for (i = 0; i < 4; i ++) {
         int vsub = i == 1 || i == 2 ? flip->vsub : 0;
 
-        if (picref->data[i]) {
-            picref->data[i] += ((h >> vsub)-1) * picref->linesize[i];
-            picref->linesize[i] = -picref->linesize[i];
+        if (frame->data[i]) {
+            frame->data[i] += ((h >> vsub) - 1) * frame->linesize[i];
+            frame->linesize[i] = -frame->linesize[i];
         }
     }
 
-    return picref;
+    return frame;
 }
 
-static int filter_frame(AVFilterLink *link, AVFilterBufferRef *frame)
+static int filter_frame(AVFilterLink *link, AVFrame *frame)
 {
     FlipContext *flip = link->dst->priv;
     int i;
diff --git a/libavfilter/vf_yadif.c b/libavfilter/vf_yadif.c
index d1670902bb..37e9887421 100644
--- a/libavfilter/vf_yadif.c
+++ b/libavfilter/vf_yadif.c
@@ -31,8 +31,6 @@
 #undef NDEBUG
 #include <assert.h>
 
-#define PERM_RWP AV_PERM_WRITE | AV_PERM_PRESERVE | AV_PERM_REUSE
-
 #define CHECK(j)\
     {   int score = FFABS(cur[mrefs + off_left + (j)] - cur[prefs + off_left - (j)])\
                   + FFABS(cur[mrefs  +(j)] - cur[prefs  -(j)])\
@@ -167,15 +165,15 @@ static void filter_edges_16bit(void *dst1, void *prev1, void *cur1, void *next1,
     FILTER(w - 3, w)
 }
 
-static void filter(AVFilterContext *ctx, AVFilterBufferRef *dstpic,
+static void filter(AVFilterContext *ctx, AVFrame *dstpic,
                    int parity, int tff)
 {
     YADIFContext *yadif = ctx->priv;
     int y, i;
 
     for (i = 0; i < yadif->csp->nb_components; i++) {
-        int w = dstpic->video->w;
-        int h = dstpic->video->h;
+        int w = dstpic->width;
+        int h = dstpic->height;
         int refs = yadif->cur->linesize[i];
         int df = (yadif->csp->comp[i].depth_minus1 + 8) / 8;
         int l_edge, l_edge_pix;
@@ -225,23 +223,22 @@ static void filter(AVFilterContext *ctx, AVFilterBufferRef *dstpic,
     emms_c();
 }
 
-static AVFilterBufferRef *get_video_buffer(AVFilterLink *link, int perms,
-                                           int w, int h)
+static AVFrame *get_video_buffer(AVFilterLink *link, int w, int h)
 {
-    AVFilterBufferRef *picref;
+    AVFrame *frame;
     int width  = FFALIGN(w, 32);
     int height = FFALIGN(h + 2, 32);
     int i;
 
-    picref = ff_default_get_video_buffer(link, perms, width, height);
+    frame = ff_default_get_video_buffer(link, width, height);
 
-    picref->video->w = w;
-    picref->video->h = h;
+    frame->width  = w;
+    frame->height = h;
 
     for (i = 0; i < 3; i++)
-        picref->data[i] += picref->linesize[i];
+        frame->data[i] += frame->linesize[i];
 
-    return picref;
+    return frame;
 }
 
 static int return_frame(AVFilterContext *ctx, int is_second)
@@ -251,19 +248,19 @@ static int return_frame(AVFilterContext *ctx, int is_second)
     int tff, ret;
 
     if (yadif->parity == -1) {
-        tff = yadif->cur->video->interlaced ?
-              yadif->cur->video->top_field_first : 1;
+        tff = yadif->cur->interlaced_frame ?
+              yadif->cur->top_field_first : 1;
     } else {
         tff = yadif->parity ^ 1;
     }
 
     if (is_second) {
-        yadif->out = ff_get_video_buffer(link, PERM_RWP, link->w, link->h);
+        yadif->out = ff_get_video_buffer(link, link->w, link->h);
         if (!yadif->out)
             return AVERROR(ENOMEM);
 
-        avfilter_copy_buffer_ref_props(yadif->out, yadif->cur);
-        yadif->out->video->interlaced = 0;
+        av_frame_copy_props(yadif->out, yadif->cur);
+        yadif->out->interlaced_frame = 0;
     }
 
     filter(ctx, yadif->out, tff ^ !is_second, tff);
@@ -284,7 +281,7 @@ static int return_frame(AVFilterContext *ctx, int is_second)
     return ret;
 }
 
-static int filter_frame(AVFilterLink *link, AVFilterBufferRef *picref)
+static int filter_frame(AVFilterLink *link, AVFrame *frame)
 {
     AVFilterContext *ctx = link->dst;
     YADIFContext *yadif = ctx->priv;
@@ -293,36 +290,35 @@ static int filter_frame(AVFilterLink *link, AVFilterBufferRef *picref)
         return_frame(ctx, 1);
 
     if (yadif->prev)
-        avfilter_unref_buffer(yadif->prev);
+        av_frame_free(&yadif->prev);
     yadif->prev = yadif->cur;
     yadif->cur  = yadif->next;
-    yadif->next = picref;
+    yadif->next = frame;
 
     if (!yadif->cur)
         return 0;
 
-    if (yadif->auto_enable && !yadif->cur->video->interlaced) {
-        yadif->out  = avfilter_ref_buffer(yadif->cur, AV_PERM_READ);
+    if (yadif->auto_enable && !yadif->cur->interlaced_frame) {
+        yadif->out  = av_frame_clone(yadif->cur);
         if (!yadif->out)
             return AVERROR(ENOMEM);
 
-        avfilter_unref_bufferp(&yadif->prev);
+        av_frame_free(&yadif->prev);
         if (yadif->out->pts != AV_NOPTS_VALUE)
             yadif->out->pts *= 2;
         return ff_filter_frame(ctx->outputs[0], yadif->out);
     }
 
     if (!yadif->prev &&
-        !(yadif->prev = avfilter_ref_buffer(yadif->cur, AV_PERM_READ)))
+        !(yadif->prev = av_frame_clone(yadif->cur)))
         return AVERROR(ENOMEM);
 
-    yadif->out = ff_get_video_buffer(ctx->outputs[0], PERM_RWP,
-                                     link->w, link->h);
+    yadif->out = ff_get_video_buffer(ctx->outputs[0], link->w, link->h);
     if (!yadif->out)
         return AVERROR(ENOMEM);
 
-    avfilter_copy_buffer_ref_props(yadif->out, yadif->cur);
-    yadif->out->video->interlaced = 0;
+    av_frame_copy_props(yadif->out, yadif->cur);
+    yadif->out->interlaced_frame = 0;
 
     if (yadif->out->pts != AV_NOPTS_VALUE)
         yadif->out->pts *= 2;
@@ -349,8 +345,7 @@ static int request_frame(AVFilterLink *link)
         ret  = ff_request_frame(link->src->inputs[0]);
 
         if (ret == AVERROR_EOF && yadif->next) {
-            AVFilterBufferRef *next =
-                avfilter_ref_buffer(yadif->next, AV_PERM_READ);
+            AVFrame *next = av_frame_clone(yadif->next);
 
             if (!next)
                 return AVERROR(ENOMEM);
@@ -389,7 +384,7 @@ static int poll_frame(AVFilterLink *link)
     }
     assert(yadif->next || !val);
 
-    if (yadif->auto_enable && yadif->next && !yadif->next->video->interlaced)
+    if (yadif->auto_enable && yadif->next && !yadif->next->interlaced_frame)
         return val;
 
     return val * ((yadif->mode&1)+1);
@@ -399,9 +394,9 @@ static av_cold void uninit(AVFilterContext *ctx)
 {
     YADIFContext *yadif = ctx->priv;
 
-    if (yadif->prev) avfilter_unref_bufferp(&yadif->prev);
-    if (yadif->cur ) avfilter_unref_bufferp(&yadif->cur );
-    if (yadif->next) avfilter_unref_bufferp(&yadif->next);
+    if (yadif->prev) av_frame_free(&yadif->prev);
+    if (yadif->cur ) av_frame_free(&yadif->cur );
+    if (yadif->next) av_frame_free(&yadif->next);
 }
 
 static int query_formats(AVFilterContext *ctx)
diff --git a/libavfilter/video.c b/libavfilter/video.c
index cb68ca4162..23bf867a82 100644
--- a/libavfilter/video.c
+++ b/libavfilter/video.c
@@ -19,6 +19,7 @@
 #include <string.h>
 #include <stdio.h>
 
+#include "libavutil/buffer.h"
 #include "libavutil/imgutils.h"
 #include "libavutil/mem.h"
 
@@ -26,77 +27,34 @@
 #include "internal.h"
 #include "video.h"
 
-#ifdef DEBUG
-static char *ff_get_ref_perms_string(char *buf, size_t buf_size, int perms)
+AVFrame *ff_null_get_video_buffer(AVFilterLink *link, int w, int h)
 {
-    snprintf(buf, buf_size, "%s%s%s%s%s%s",
-             perms & AV_PERM_READ      ? "r" : "",
-             perms & AV_PERM_WRITE     ? "w" : "",
-             perms & AV_PERM_PRESERVE  ? "p" : "",
-             perms & AV_PERM_REUSE     ? "u" : "",
-             perms & AV_PERM_REUSE2    ? "U" : "",
-             perms & AV_PERM_NEG_LINESIZES ? "n" : "");
-    return buf;
-}
-#endif
-
-static void ff_dlog_ref(void *ctx, AVFilterBufferRef *ref, int end)
-{
-    av_unused char buf[16];
-    av_dlog(ctx,
-            "ref[%p buf:%p refcount:%d perms:%s data:%p linesize[%d, %d, %d, %d] pts:%"PRId64" pos:%"PRId64,
-            ref, ref->buf, ref->buf->refcount, ff_get_ref_perms_string(buf, sizeof(buf), ref->perms), ref->data[0],
-            ref->linesize[0], ref->linesize[1], ref->linesize[2], ref->linesize[3],
-            ref->pts, ref->pos);
-
-    if (ref->video) {
-        av_dlog(ctx, " a:%d/%d s:%dx%d i:%c iskey:%d type:%c",
-                ref->video->pixel_aspect.num, ref->video->pixel_aspect.den,
-                ref->video->w, ref->video->h,
-                !ref->video->interlaced     ? 'P' :         /* Progressive  */
-                ref->video->top_field_first ? 'T' : 'B',    /* Top / Bottom */
-                ref->video->key_frame,
-                av_get_picture_type_char(ref->video->pict_type));
-    }
-    if (ref->audio) {
-        av_dlog(ctx, " cl:%"PRId64"d n:%d r:%d p:%d",
-                ref->audio->channel_layout,
-                ref->audio->nb_samples,
-                ref->audio->sample_rate,
-                ref->audio->planar);
-    }
-
-    av_dlog(ctx, "]%s", end ? "\n" : "");
-}
-
-AVFilterBufferRef *ff_null_get_video_buffer(AVFilterLink *link, int perms, int w, int h)
-{
-    return ff_get_video_buffer(link->dst->outputs[0], perms, w, h);
+    return ff_get_video_buffer(link->dst->outputs[0], w, h);
 }
 
 /* TODO: set the buffer's priv member to a context structure for the whole
  * filter chain.  This will allow for a buffer pool instead of the constant
  * alloc & free cycle currently implemented. */
-AVFilterBufferRef *ff_default_get_video_buffer(AVFilterLink *link, int perms, int w, int h)
+AVFrame *ff_default_get_video_buffer(AVFilterLink *link, int w, int h)
 {
-    int linesize[4];
-    uint8_t *data[4];
-    AVFilterBufferRef *picref = NULL;
+    AVFrame *frame = av_frame_alloc();
+    int ret;
 
-    // +2 is needed for swscaler, +16 to be SIMD-friendly
-    if (av_image_alloc(data, linesize, w, h, link->format, 16) < 0)
+    if (!frame)
         return NULL;
 
-    picref = avfilter_get_video_buffer_ref_from_arrays(data, linesize,
-                                                       perms, w, h, link->format);
-    if (!picref) {
-        av_free(data[0]);
-        return NULL;
-    }
+    frame->width  = w;
+    frame->height = h;
+    frame->format = link->format;
 
-    return picref;
+    ret = av_frame_get_buffer(frame, 32);
+    if (ret < 0)
+        av_frame_free(&frame);
+
+    return frame;
 }
 
+#if FF_API_AVFILTERBUFFER
 AVFilterBufferRef *
 avfilter_get_video_buffer_ref_from_arrays(uint8_t *data[4], int linesize[4], int perms,
                                           int w, int h, enum AVPixelFormat format)
@@ -141,25 +99,20 @@ fail:
     av_free(pic);
     return NULL;
 }
+#endif
 
-AVFilterBufferRef *ff_get_video_buffer(AVFilterLink *link, int perms, int w, int h)
+AVFrame *ff_get_video_buffer(AVFilterLink *link, int w, int h)
 {
-    AVFilterBufferRef *ret = NULL;
+    AVFrame *ret = NULL;
 
     av_unused char buf[16];
     FF_DPRINTF_START(NULL, get_video_buffer); ff_dlog_link(NULL, link, 0);
-    av_dlog(NULL, " perms:%s w:%d h:%d\n", ff_get_ref_perms_string(buf, sizeof(buf), perms), w, h);
 
     if (link->dstpad->get_video_buffer)
-        ret = link->dstpad->get_video_buffer(link, perms, w, h);
+        ret = link->dstpad->get_video_buffer(link, w, h);
 
     if (!ret)
-        ret = ff_default_get_video_buffer(link, perms, w, h);
-
-    if (ret)
-        ret->type = AVMEDIA_TYPE_VIDEO;
-
-    FF_DPRINTF_START(NULL, get_video_buffer); ff_dlog_link(NULL, link, 0); av_dlog(NULL, " returning "); ff_dlog_ref(NULL, ret, 1);
+        ret = ff_default_get_video_buffer(link, w, h);
 
     return ret;
 }
diff --git a/libavfilter/video.h b/libavfilter/video.h
index be93810bcd..f7e8e34008 100644
--- a/libavfilter/video.h
+++ b/libavfilter/video.h
@@ -21,22 +21,19 @@
 
 #include "avfilter.h"
 
-AVFilterBufferRef *ff_default_get_video_buffer(AVFilterLink *link,
-                                               int perms, int w, int h);
-AVFilterBufferRef *ff_null_get_video_buffer(AVFilterLink *link, int perms, int w, int h);
+AVFrame *ff_default_get_video_buffer(AVFilterLink *link, int w, int h);
+AVFrame *ff_null_get_video_buffer(AVFilterLink *link, int w, int h);
 
 /**
  * Request a picture buffer with a specific set of permissions.
  *
  * @param link  the output link to the filter from which the buffer will
  *              be requested
- * @param perms the required access permissions
  * @param w     the minimum width of the buffer to allocate
  * @param h     the minimum height of the buffer to allocate
  * @return      A reference to the buffer. This must be unreferenced with
  *              avfilter_unref_buffer when you are finished with it.
  */
-AVFilterBufferRef *ff_get_video_buffer(AVFilterLink *link, int perms,
-                                       int w, int h);
+AVFrame *ff_get_video_buffer(AVFilterLink *link, int w, int h);
 
 #endif /* AVFILTER_VIDEO_H */
diff --git a/libavfilter/vsink_nullsink.c b/libavfilter/vsink_nullsink.c
index 71d2b3ec50..9ab67e12f8 100644
--- a/libavfilter/vsink_nullsink.c
+++ b/libavfilter/vsink_nullsink.c
@@ -20,9 +20,9 @@
 #include "internal.h"
 #include "libavutil/internal.h"
 
-static int filter_frame(AVFilterLink *link, AVFilterBufferRef *frame)
+static int filter_frame(AVFilterLink *link, AVFrame *frame)
 {
-    avfilter_unref_bufferp(&frame);
+    av_frame_free(&frame);
     return 0;
 }
 
diff --git a/libavfilter/vsrc_color.c b/libavfilter/vsrc_color.c
index c0a4e1c3b2..9d9aab3f2a 100644
--- a/libavfilter/vsrc_color.c
+++ b/libavfilter/vsrc_color.c
@@ -146,19 +146,18 @@ static int color_config_props(AVFilterLink *inlink)
 static int color_request_frame(AVFilterLink *link)
 {
     ColorContext *color = link->src->priv;
-    AVFilterBufferRef *picref = ff_get_video_buffer(link, AV_PERM_WRITE, color->w, color->h);
+    AVFrame *frame = ff_get_video_buffer(link, color->w, color->h);
 
-    if (!picref)
+    if (!frame)
         return AVERROR(ENOMEM);
 
-    picref->video->pixel_aspect = (AVRational) {1, 1};
-    picref->pts                 = color->pts++;
-    picref->pos                 = -1;
+    frame->sample_aspect_ratio = (AVRational) {1, 1};
+    frame->pts                 = color->pts++;
 
-    ff_draw_rectangle(picref->data, picref->linesize,
+    ff_draw_rectangle(frame->data, frame->linesize,
                       color->line, color->line_step, color->hsub, color->vsub,
                       0, 0, color->w, color->h);
-    return ff_filter_frame(link, picref);
+    return ff_filter_frame(link, frame);
 }
 
 static const AVFilterPad avfilter_vsrc_color_outputs[] = {
diff --git a/libavfilter/vsrc_movie.c b/libavfilter/vsrc_movie.c
index 0023d55bbc..2a893bd517 100644
--- a/libavfilter/vsrc_movie.c
+++ b/libavfilter/vsrc_movie.c
@@ -54,7 +54,6 @@ typedef struct {
     AVFrame *frame;   ///< video frame to store the decoded images in
 
     int w, h;
-    AVFilterBufferRef *picref;
 } MovieContext;
 
 #define OFFSET(x) offsetof(MovieContext, x)
@@ -142,16 +141,13 @@ static int movie_init(AVFilterContext *ctx)
         return AVERROR(EINVAL);
     }
 
+    movie->codec_ctx->refcounted_frames = 1;
+
     if ((ret = avcodec_open2(movie->codec_ctx, codec, NULL)) < 0) {
         av_log(ctx, AV_LOG_ERROR, "Failed to open codec\n");
         return ret;
     }
 
-    if (!(movie->frame = avcodec_alloc_frame()) ) {
-        av_log(ctx, AV_LOG_ERROR, "Failed to alloc frame\n");
-        return AVERROR(ENOMEM);
-    }
-
     movie->w = movie->codec_ctx->width;
     movie->h = movie->codec_ctx->height;
 
@@ -196,8 +192,7 @@ static av_cold void uninit(AVFilterContext *ctx)
         avcodec_close(movie->codec_ctx);
     if (movie->format_ctx)
         avformat_close_input(&movie->format_ctx);
-    avfilter_unref_buffer(movie->picref);
-    avcodec_free_frame(&movie->frame);
+    av_frame_free(&movie->frame);
 }
 
 static int query_formats(AVFilterContext *ctx)
@@ -225,41 +220,29 @@ static int movie_get_frame(AVFilterLink *outlink)
     MovieContext *movie = outlink->src->priv;
     AVPacket pkt;
     int ret, frame_decoded;
-    AVStream *st = movie->format_ctx->streams[movie->stream_index];
+    AVStream av_unused *st = movie->format_ctx->streams[movie->stream_index];
 
     if (movie->is_done == 1)
         return 0;
 
+    movie->frame = av_frame_alloc();
+    if (!movie->frame)
+        return AVERROR(ENOMEM);
+
     while ((ret = av_read_frame(movie->format_ctx, &pkt)) >= 0) {
         // Is this a packet from the video stream?
         if (pkt.stream_index == movie->stream_index) {
-            movie->codec_ctx->reordered_opaque = pkt.pos;
             avcodec_decode_video2(movie->codec_ctx, movie->frame, &frame_decoded, &pkt);
 
             if (frame_decoded) {
-                /* FIXME: avoid the memcpy */
-                movie->picref = ff_get_video_buffer(outlink, AV_PERM_WRITE | AV_PERM_PRESERVE |
-                                                    AV_PERM_REUSE2, outlink->w, outlink->h);
-                av_image_copy(movie->picref->data, movie->picref->linesize,
-                              movie->frame->data,  movie->frame->linesize,
-                              movie->picref->format, outlink->w, outlink->h);
-                avfilter_copy_frame_props(movie->picref, movie->frame);
-
-                /* FIXME: use a PTS correction mechanism as that in
-                 * ffplay.c when some API will be available for that */
-                /* use pkt_dts if pkt_pts is not available */
-                movie->picref->pts = movie->frame->pkt_pts == AV_NOPTS_VALUE ?
-                    movie->frame->pkt_dts : movie->frame->pkt_pts;
-
-                movie->picref->pos                    = movie->frame->reordered_opaque;
-                if (!movie->frame->sample_aspect_ratio.num)
-                    movie->picref->video->pixel_aspect = st->sample_aspect_ratio;
+                if (movie->frame->pkt_pts != AV_NOPTS_VALUE)
+                    movie->frame->pts = movie->frame->pkt_pts;
                 av_dlog(outlink->src,
-                        "movie_get_frame(): file:'%s' pts:%"PRId64" time:%f pos:%"PRId64" aspect:%d/%d\n",
-                        movie->file_name, movie->picref->pts,
-                        (double)movie->picref->pts * av_q2d(st->time_base),
-                        movie->picref->pos,
-                        movie->picref->video->pixel_aspect.num, movie->picref->video->pixel_aspect.den);
+                        "movie_get_frame(): file:'%s' pts:%"PRId64" time:%f aspect:%d/%d\n",
+                        movie->file_name, movie->frame->pts,
+                        (double)movie->frame->pts * av_q2d(st->time_base),
+                        movie->frame->sample_aspect_ratio.num,
+                        movie->frame->sample_aspect_ratio.den);
                 // We got it. Free the packet since we are returning
                 av_free_packet(&pkt);
 
@@ -287,8 +270,8 @@ static int request_frame(AVFilterLink *outlink)
     if ((ret = movie_get_frame(outlink)) < 0)
         return ret;
 
-    ret = ff_filter_frame(outlink, movie->picref);
-    movie->picref = NULL;
+    ret = ff_filter_frame(outlink, movie->frame);
+    movie->frame = NULL;
 
     return ret;
 }
diff --git a/libavfilter/vsrc_testsrc.c b/libavfilter/vsrc_testsrc.c
index 632bd278ac..674d6d716c 100644
--- a/libavfilter/vsrc_testsrc.c
+++ b/libavfilter/vsrc_testsrc.c
@@ -53,7 +53,7 @@ typedef struct {
     char *duration;             ///< total duration of the generated video
     AVRational sar;             ///< sample aspect ratio
 
-    void (* fill_picture_fn)(AVFilterContext *ctx, AVFilterBufferRef *picref);
+    void (* fill_picture_fn)(AVFilterContext *ctx, AVFrame *frame);
 
     /* only used by rgbtest */
     int rgba_map[4];
@@ -130,24 +130,23 @@ static int config_props(AVFilterLink *outlink)
 static int request_frame(AVFilterLink *outlink)
 {
     TestSourceContext *test = outlink->src->priv;
-    AVFilterBufferRef *picref;
+    AVFrame *frame;
 
     if (test->max_pts >= 0 && test->pts > test->max_pts)
         return AVERROR_EOF;
-    picref = ff_get_video_buffer(outlink, AV_PERM_WRITE, test->w, test->h);
-    if (!picref)
+    frame = ff_get_video_buffer(outlink, test->w, test->h);
+    if (!frame)
         return AVERROR(ENOMEM);
 
-    picref->pts = test->pts++;
-    picref->pos = -1;
-    picref->video->key_frame = 1;
-    picref->video->interlaced = 0;
-    picref->video->pict_type = AV_PICTURE_TYPE_I;
-    picref->video->pixel_aspect = test->sar;
+    frame->pts                 = test->pts++;
+    frame->key_frame           = 1;
+    frame->interlaced_frame    = 0;
+    frame->pict_type           = AV_PICTURE_TYPE_I;
+    frame->sample_aspect_ratio = test->sar;
     test->nb_frame++;
-    test->fill_picture_fn(outlink->src, picref);
+    test->fill_picture_fn(outlink->src, frame);
 
-    return ff_filter_frame(outlink, picref);
+    return ff_filter_frame(outlink, frame);
 }
 
 #if CONFIG_TESTSRC_FILTER
@@ -235,7 +234,7 @@ static void draw_digit(int digit, uint8_t *dst, unsigned dst_linesize,
 
 #define GRADIENT_SIZE (6 * 256)
 
-static void test_fill_picture(AVFilterContext *ctx, AVFilterBufferRef *picref)
+static void test_fill_picture(AVFilterContext *ctx, AVFrame *frame)
 {
     TestSourceContext *test = ctx->priv;
     uint8_t *p, *p0;
@@ -249,9 +248,9 @@ static void test_fill_picture(AVFilterContext *ctx, AVFilterBufferRef *picref)
     int seg_size;
     int second;
     int i;
-    uint8_t *data = picref->data[0];
-    int width  = picref->video->w;
-    int height = picref->video->h;
+    uint8_t *data = frame->data[0];
+    int width  = frame->width;
+    int height = frame->height;
 
     /* draw colored bars and circle */
     radius = (width + height) / 4;
@@ -281,11 +280,11 @@ static void test_fill_picture(AVFilterContext *ctx, AVFilterBufferRef *picref)
         }
         quad0 += dquad_y;
         dquad_y += 2;
-        p0 += picref->linesize[0];
+        p0 += frame->linesize[0];
     }
 
     /* draw sliding color line */
-    p = data + picref->linesize[0] * height * 3/4;
+    p = data + frame->linesize[0] * height * 3/4;
     grad = (256 * test->nb_frame * test->time_base.num / test->time_base.den) %
         GRADIENT_SIZE;
     rgrad = 0;
@@ -314,8 +313,8 @@ static void test_fill_picture(AVFilterContext *ctx, AVFilterBufferRef *picref)
             grad -= GRADIENT_SIZE;
     }
     for (y = height / 8; y > 0; y--) {
-        memcpy(p, p - picref->linesize[0], 3 * width);
-        p += picref->linesize[0];
+        memcpy(p, p - frame->linesize[0], 3 * width);
+        p += frame->linesize[0];
     }
 
     /* draw digits */
@@ -324,10 +323,10 @@ static void test_fill_picture(AVFilterContext *ctx, AVFilterBufferRef *picref)
         second = test->nb_frame * test->time_base.num / test->time_base.den;
         x = width - (width - seg_size * 64) / 2;
         y = (height - seg_size * 13) / 2;
-        p = data + (x*3 + y * picref->linesize[0]);
+        p = data + (x*3 + y * frame->linesize[0]);
         for (i = 0; i < 8; i++) {
             p -= 3 * 8 * seg_size;
-            draw_digit(second % 10, p, picref->linesize[0], seg_size);
+            draw_digit(second % 10, p, frame->linesize[0], seg_size);
             second /= 10;
             if (second == 0)
                 break;
@@ -427,13 +426,13 @@ static void rgbtest_put_pixel(uint8_t *dst, int dst_linesize,
     }
 }
 
-static void rgbtest_fill_picture(AVFilterContext *ctx, AVFilterBufferRef *picref)
+static void rgbtest_fill_picture(AVFilterContext *ctx, AVFrame *frame)
 {
     TestSourceContext *test = ctx->priv;
-    int x, y, w = picref->video->w, h = picref->video->h;
+    int x, y, w = frame->width, h = frame->height;
 
     for (y = 0; y < h; y++) {
-         for (x = 0; x < picref->video->w; x++) {
+         for (x = 0; x < w; x++) {
              int c = 256*x/w;
              int r = 0, g = 0, b = 0;
 
@@ -441,7 +440,7 @@ static void rgbtest_fill_picture(AVFilterContext *ctx, AVFilterBufferRef *picref
              else if (3*y < 2*h) g = c;
              else                b = c;
 
-             rgbtest_put_pixel(picref->data[0], picref->linesize[0], x, y, r, g, b,
+             rgbtest_put_pixel(frame->data[0], frame->linesize[0], x, y, r, g, b,
                                ctx->outputs[0]->format, test->rgba_map);
          }
      }
diff --git a/libavfilter/yadif.h b/libavfilter/yadif.h
index 18c8624f1f..a9cbecf471 100644
--- a/libavfilter/yadif.h
+++ b/libavfilter/yadif.h
@@ -46,10 +46,10 @@ typedef struct YADIFContext {
      */
     int auto_enable;
 
-    AVFilterBufferRef *cur;
-    AVFilterBufferRef *next;
-    AVFilterBufferRef *prev;
-    AVFilterBufferRef *out;
+    AVFrame *cur;
+    AVFrame *next;
+    AVFrame *prev;
+    AVFrame *out;
 
     /**
      * Required alignment for filter_line