diff options
author | Michael Niedermayer <michaelni@gmx.at> | 2011-06-30 04:20:53 +0200 |
---|---|---|
committer | Michael Niedermayer <michaelni@gmx.at> | 2011-06-30 04:32:24 +0200 |
commit | 8c0cbb0848430e4de7182dc2fa2f70262d8c77c7 (patch) | |
tree | d08758ea987f08f1d5a8caf44f082adc5586e815 /libavcodec | |
parent | 623bf96678d2009d34e88ed7c2b73c00930cebae (diff) | |
parent | 812f2376eecc19c40a95ade1f7a43b762106e0f7 (diff) | |
download | ffmpeg-8c0cbb0848430e4de7182dc2fa2f70262d8c77c7.tar.gz |
Merge remote-tracking branch 'qatar/master'
* qatar/master:
rational-test: Add proper main() declaration to fix gcc warnings.
configure: Add vdpau and dxva2 to configure results output.
Remove unused, never built libavutil/pca.[ch]
matroskadec: forward parsing errors to caller.
av_find_stream_info: simplify EAGAIN handling.
aacenc: Fix determination of Mid/Side Mode.
psymodel: Remove the single channel analysis function
aacenc: Implement dummy channel group analysis that just calls the single channel analysis for each channel.
psymodel: Add channels and channel groups to the psymodel.
ARM: remove check for PLD instruction
fate: move amr[nw]b test rules into separate files
ogg: fix double free when finding length of small chained oggs.
swscale: implement >8bit scaling support.
build: fix creation of tools dir with make 3.81
build: Mark all-yes Makefile target as phony.
pixfmt: fix YUV422/444 wrong endian comment
build: create output directories as needed
Add new yuv444 pixfmts to avcodec_align_dimensions2
Conflicts:
Makefile
configure
libavutil/pca.c
libavutil/pca.h
libavutil/pixfmt.h
libswscale/swscale.c
libswscale/utils.c
libswscale/x86/swscale_template.c
tests/ref/lavfi/pixdesc
tests/ref/lavfi/pixfmts_copy
tests/ref/lavfi/pixfmts_null
tests/ref/lavfi/pixfmts_scale
tests/ref/lavfi/pixfmts_vflip
Merged-by: Michael Niedermayer <michaelni@gmx.at>
Diffstat (limited to 'libavcodec')
-rw-r--r-- | libavcodec/aaccoder.c | 14 | ||||
-rw-r--r-- | libavcodec/aacenc.c | 13 | ||||
-rw-r--r-- | libavcodec/aacpsy.c | 18 | ||||
-rw-r--r-- | libavcodec/arm/dsputil_arm.S | 7 | ||||
-rw-r--r-- | libavcodec/psymodel.c | 36 | ||||
-rw-r--r-- | libavcodec/psymodel.h | 53 | ||||
-rw-r--r-- | libavcodec/utils.c | 4 |
7 files changed, 107 insertions, 38 deletions
diff --git a/libavcodec/aaccoder.c b/libavcodec/aaccoder.c index 4eba5dd83c..b61af18056 100644 --- a/libavcodec/aaccoder.c +++ b/libavcodec/aaccoder.c @@ -346,7 +346,7 @@ static void encode_window_bands_info(AACEncContext *s, SingleChannelElement *sce float cost_stay_here, cost_get_here; float rd = 0.0f; for (w = 0; w < group_len; w++) { - FFPsyBand *band = &s->psy.psy_bands[s->cur_channel*PSY_MAX_BANDS+(win+w)*16+swb]; + FFPsyBand *band = &s->psy.ch[s->cur_channel].psy_bands[(win+w)*16+swb]; rd += quantize_band_cost(s, sce->coeffs + start + w*128, s->scoefs + start + w*128, size, sce->sf_idx[(win+w)*16+swb], cb, @@ -626,7 +626,7 @@ static void search_for_quantizers_anmr(AVCodecContext *avctx, AACEncContext *s, qmin = INT_MAX; qmax = 0.0f; for (w2 = 0; w2 < sce->ics.group_len[w]; w2++) { - FFPsyBand *band = &s->psy.psy_bands[s->cur_channel*PSY_MAX_BANDS+(w+w2)*16+g]; + FFPsyBand *band = &s->psy.ch[s->cur_channel].psy_bands[(w+w2)*16+g]; if (band->energy <= band->threshold || band->threshold == 0.0f) { sce->zeroes[(w+w2)*16+g] = 1; continue; @@ -655,7 +655,7 @@ static void search_for_quantizers_anmr(AVCodecContext *avctx, AACEncContext *s, float dist = 0; int cb = find_min_book(maxval, sce->sf_idx[w*16+g]); for (w2 = 0; w2 < sce->ics.group_len[w]; w2++) { - FFPsyBand *band = &s->psy.psy_bands[s->cur_channel*PSY_MAX_BANDS+(w+w2)*16+g]; + FFPsyBand *band = &s->psy.ch[s->cur_channel].psy_bands[(w+w2)*16+g]; dist += quantize_band_cost(s, coefs + w2*128, s->scoefs + start + w2*128, sce->ics.swb_sizes[g], q + q0, cb, lambda / band->threshold, INFINITY, NULL); } @@ -728,7 +728,7 @@ static void search_for_quantizers_twoloop(AVCodecContext *avctx, int nz = 0; float uplim = 0.0f; for (w2 = 0; w2 < sce->ics.group_len[w]; w2++) { - FFPsyBand *band = &s->psy.psy_bands[s->cur_channel*PSY_MAX_BANDS+(w+w2)*16+g]; + FFPsyBand *band = &s->psy.ch[s->cur_channel].psy_bands[(w+w2)*16+g]; uplim += band->threshold; if (band->energy <= band->threshold || band->threshold == 0.0f) { sce->zeroes[(w+w2)*16+g] = 1; @@ -1028,7 +1028,7 @@ static void search_for_quantizers_fast(AVCodecContext *avctx, AACEncContext *s, for (w = 0; w < sce->ics.num_windows; w += sce->ics.group_len[w]) { for (g = 0; g < sce->ics.num_swb; g++) { for (w2 = 0; w2 < sce->ics.group_len[w]; w2++) { - FFPsyBand *band = &s->psy.psy_bands[s->cur_channel*PSY_MAX_BANDS+(w+w2)*16+g]; + FFPsyBand *band = &s->psy.ch[s->cur_channel].psy_bands[(w+w2)*16+g]; if (band->energy <= band->threshold) { sce->sf_idx[(w+w2)*16+g] = 218; sce->zeroes[(w+w2)*16+g] = 1; @@ -1066,8 +1066,8 @@ static void search_for_ms(AACEncContext *s, ChannelElement *cpe, if (!cpe->ch[0].zeroes[w*16+g] && !cpe->ch[1].zeroes[w*16+g]) { float dist1 = 0.0f, dist2 = 0.0f; for (w2 = 0; w2 < sce0->ics.group_len[w]; w2++) { - FFPsyBand *band0 = &s->psy.psy_bands[(s->cur_channel+0)*PSY_MAX_BANDS+(w+w2)*16+g]; - FFPsyBand *band1 = &s->psy.psy_bands[(s->cur_channel+1)*PSY_MAX_BANDS+(w+w2)*16+g]; + FFPsyBand *band0 = &s->psy.ch[s->cur_channel+0].psy_bands[(w+w2)*16+g]; + FFPsyBand *band1 = &s->psy.ch[s->cur_channel+1].psy_bands[(w+w2)*16+g]; float minthr = FFMIN(band0->threshold, band1->threshold); float maxthr = FFMAX(band0->threshold, band1->threshold); for (i = 0; i < sce0->ics.swb_sizes[g]; i++) { diff --git a/libavcodec/aacenc.c b/libavcodec/aacenc.c index 2a6dbc8ebd..b877e4eaef 100644 --- a/libavcodec/aacenc.c +++ b/libavcodec/aacenc.c @@ -219,7 +219,7 @@ static av_cold int aac_encode_init(AVCodecContext *avctx) sizes[1] = swb_size_128[i]; lengths[0] = ff_aac_num_swb_1024[i]; lengths[1] = ff_aac_num_swb_128[i]; - ff_psy_init(&s->psy, avctx, 2, sizes, lengths); + ff_psy_init(&s->psy, avctx, 2, sizes, lengths, s->chan_map[0], &s->chan_map[1]); s->psypp = ff_psy_preprocess_init(avctx); s->coder = &ff_aac_coders[2]; @@ -373,7 +373,7 @@ static void adjust_frame_information(AACEncContext *apc, ChannelElement *cpe, in if (msc == 0 || ics0->max_sfb == 0) cpe->ms_mode = 0; else - cpe->ms_mode = msc < ics0->max_sfb ? 1 : 2; + cpe->ms_mode = msc < ics0->max_sfb * ics0->num_windows ? 1 : 2; } } @@ -582,14 +582,17 @@ static int aac_encode_frame(AVCodecContext *avctx, memset(chan_el_counter, 0, sizeof(chan_el_counter)); for (i = 0; i < s->chan_map[0]; i++) { FFPsyWindowInfo* wi = windows + start_ch; + const float *coeffs[2]; tag = s->chan_map[i+1]; chans = tag == TYPE_CPE ? 2 : 1; cpe = &s->cpe[i]; put_bits(&s->pb, 3, tag); put_bits(&s->pb, 4, chan_el_counter[tag]++); + for (ch = 0; ch < chans; ch++) + coeffs[ch] = cpe->ch[ch].coeffs; + s->psy.model->analyze(&s->psy, start_ch, coeffs, wi); for (ch = 0; ch < chans; ch++) { - s->cur_channel = start_ch + ch; - s->psy.model->analyze(&s->psy, s->cur_channel, cpe->ch[ch].coeffs, &wi[ch]); + s->cur_channel = start_ch * 2 + ch; s->coder->search_for_quantizers(avctx, s, &cpe->ch[ch], s->lambda); } cpe->common_window = 0; @@ -605,7 +608,7 @@ static int aac_encode_frame(AVCodecContext *avctx, } } } - s->cur_channel = start_ch; + s->cur_channel = start_ch * 2; if (s->options.stereo_mode && cpe->common_window) { if (s->options.stereo_mode > 0) { IndividualChannelStream *ics = &cpe->ch[0].ics; diff --git a/libavcodec/aacpsy.c b/libavcodec/aacpsy.c index f2b525694b..4152b70bbf 100644 --- a/libavcodec/aacpsy.c +++ b/libavcodec/aacpsy.c @@ -557,8 +557,8 @@ static float calc_reduced_thr_3gpp(AacPsyBand *band, float min_snr, /** * Calculate band thresholds as suggested in 3GPP TS26.403 */ -static void psy_3gpp_analyze(FFPsyContext *ctx, int channel, - const float *coefs, const FFPsyWindowInfo *wi) +static void psy_3gpp_analyze_channel(FFPsyContext *ctx, int channel, + const float *coefs, const FFPsyWindowInfo *wi) { AacPsyContext *pctx = (AacPsyContext*) ctx->model_priv_data; AacPsyChannel *pch = &pctx->ch[channel]; @@ -627,7 +627,7 @@ static void psy_3gpp_analyze(FFPsyContext *ctx, int channel, } /* 5.6.1.3.2 "Calculation of the desired perceptual entropy" */ - ctx->pe[channel] = pe; + ctx->ch[channel].entropy = pe; desired_bits = calc_bit_demand(pctx, pe, ctx->bitres.bits, ctx->bitres.size, wi->num_windows == 8); desired_pe = PSY_3GPP_BITS_TO_PE(desired_bits); /* NOTE: PE correction is kept simple. During initial testing it had very @@ -731,7 +731,7 @@ static void psy_3gpp_analyze(FFPsyContext *ctx, int channel, for (w = 0; w < wi->num_windows*16; w += 16) { for (g = 0; g < num_bands; g++) { AacPsyBand *band = &pch->band[w+g]; - FFPsyBand *psy_band = &ctx->psy_bands[channel*PSY_MAX_BANDS+w+g]; + FFPsyBand *psy_band = &ctx->ch[channel].psy_bands[w+g]; psy_band->threshold = band->thr; psy_band->energy = band->energy; @@ -741,6 +741,16 @@ static void psy_3gpp_analyze(FFPsyContext *ctx, int channel, memcpy(pch->prev_band, pch->band, sizeof(pch->band)); } +static void psy_3gpp_analyze(FFPsyContext *ctx, int channel, + const float **coeffs, const FFPsyWindowInfo *wi) +{ + int ch; + FFPsyChannelGroup *group = ff_psy_find_group(ctx, channel); + + for (ch = 0; ch < group->num_ch; ch++) + psy_3gpp_analyze_channel(ctx, channel + ch, coeffs[ch], &wi[ch]); +} + static av_cold void psy_3gpp_end(FFPsyContext *apc) { AacPsyContext *pctx = (AacPsyContext*) apc->model_priv_data; diff --git a/libavcodec/arm/dsputil_arm.S b/libavcodec/arm/dsputil_arm.S index c614206bac..1247b0fa84 100644 --- a/libavcodec/arm/dsputil_arm.S +++ b/libavcodec/arm/dsputil_arm.S @@ -24,11 +24,6 @@ preserve8 -#if !HAVE_PLD -.macro pld reg -.endm -#endif - #if HAVE_ARMV5TE function ff_prefetch_arm, export=1 subs r2, r2, #1 @@ -37,6 +32,8 @@ function ff_prefetch_arm, export=1 bne ff_prefetch_arm bx lr endfunc +#else +#define pld @ #endif .macro ALIGN_QWORD_D shift, Rd0, Rd1, Rd2, Rd3, Rn0, Rn1, Rn2, Rn3, Rn4 diff --git a/libavcodec/psymodel.c b/libavcodec/psymodel.c index 133a85f5c1..faadb1b870 100644 --- a/libavcodec/psymodel.c +++ b/libavcodec/psymodel.c @@ -25,16 +25,31 @@ extern const FFPsyModel ff_aac_psy_model; -av_cold int ff_psy_init(FFPsyContext *ctx, AVCodecContext *avctx, - int num_lens, - const uint8_t **bands, const int* num_bands) +av_cold int ff_psy_init(FFPsyContext *ctx, AVCodecContext *avctx, int num_lens, + const uint8_t **bands, const int* num_bands, + int num_groups, const uint8_t *group_map) { + int i, j, k = 0; + ctx->avctx = avctx; - ctx->psy_bands = av_mallocz(sizeof(FFPsyBand) * PSY_MAX_BANDS * avctx->channels); + ctx->ch = av_mallocz(sizeof(ctx->ch[0]) * avctx->channels * 2); + ctx->group = av_mallocz(sizeof(ctx->group[0]) * num_groups); ctx->bands = av_malloc (sizeof(ctx->bands[0]) * num_lens); ctx->num_bands = av_malloc (sizeof(ctx->num_bands[0]) * num_lens); memcpy(ctx->bands, bands, sizeof(ctx->bands[0]) * num_lens); memcpy(ctx->num_bands, num_bands, sizeof(ctx->num_bands[0]) * num_lens); + + /* assign channels to groups (with virtual channels for coupling) */ + for (i = 0; i < num_groups; i++) { + /* NOTE: Add 1 to handle the AAC chan_config without modification. + * This has the side effect of allowing an array of 0s to map + * to one channel per group. + */ + ctx->group[i].num_ch = group_map[i] + 1; + for (j = 0; j < ctx->group[i].num_ch * 2; j++) + ctx->group[i].ch[j] = &ctx->ch[k++]; + } + switch (ctx->avctx->codec_id) { case CODEC_ID_AAC: ctx->model = &ff_aac_psy_model; @@ -45,13 +60,24 @@ av_cold int ff_psy_init(FFPsyContext *ctx, AVCodecContext *avctx, return 0; } +FFPsyChannelGroup *ff_psy_find_group(FFPsyContext *ctx, int channel) +{ + int i = 0, ch = 0; + + while (ch <= channel) + ch += ctx->group[i++].num_ch; + + return &ctx->group[i-1]; +} + av_cold void ff_psy_end(FFPsyContext *ctx) { if (ctx->model->end) ctx->model->end(ctx); av_freep(&ctx->bands); av_freep(&ctx->num_bands); - av_freep(&ctx->psy_bands); + av_freep(&ctx->group); + av_freep(&ctx->ch); } typedef struct FFPsyPreprocessContext{ diff --git a/libavcodec/psymodel.h b/libavcodec/psymodel.h index c65614a151..a7b7948cd2 100644 --- a/libavcodec/psymodel.h +++ b/libavcodec/psymodel.h @@ -41,6 +41,23 @@ typedef struct FFPsyBand { } FFPsyBand; /** + * single channel psychoacoustic information + */ +typedef struct FFPsyChannel { + FFPsyBand psy_bands[PSY_MAX_BANDS]; ///< channel bands information + float entropy; ///< total PE for this channel +} FFPsyChannel; + +/** + * psychoacoustic information for an arbitrary group of channels + */ +typedef struct FFPsyChannelGroup { + FFPsyChannel *ch[PSY_MAX_CHANS]; ///< pointers to the individual channels in the group + uint8_t num_ch; ///< number of channels in this group + uint8_t coupling[PSY_MAX_BANDS]; ///< allow coupling for this band in the group +} FFPsyChannelGroup; + +/** * windowing related information */ typedef struct FFPsyWindowInfo { @@ -58,14 +75,14 @@ typedef struct FFPsyContext { AVCodecContext *avctx; ///< encoder context const struct FFPsyModel *model; ///< encoder-specific model functions - FFPsyBand *psy_bands; ///< frame bands information + FFPsyChannel *ch; ///< single channel information + FFPsyChannelGroup *group; ///< channel group information + int num_groups; ///< number of channel groups uint8_t **bands; ///< scalefactor band sizes for possible frame sizes int *num_bands; ///< number of scalefactor bands for possible frame sizes int num_lens; ///< number of scalefactor band sets - float pe[PSY_MAX_CHANS]; ///< total PE for each channel in the frame - struct { int size; ///< size of the bitresevoir in bits int bits; ///< number of bits used in the bitresevoir @@ -95,14 +112,14 @@ typedef struct FFPsyModel { FFPsyWindowInfo (*window)(FFPsyContext *ctx, const int16_t *audio, const int16_t *la, int channel, int prev_type); /** - * Perform psychoacoustic analysis and set band info (threshold, energy). + * Perform psychoacoustic analysis and set band info (threshold, energy) for a group of channels. * - * @param ctx model context - * @param channel audio channel number - * @param coeffs pointer to the transformed coefficients - * @param wi window information + * @param ctx model context + * @param channel channel number of the first channel in the group to perform analysis on + * @param coeffs array of pointers to the transformed coefficients + * @param wi window information for the channels in the group */ - void (*analyze)(FFPsyContext *ctx, int channel, const float *coeffs, const FFPsyWindowInfo *wi); + void (*analyze)(FFPsyContext *ctx, int channel, const float **coeffs, const FFPsyWindowInfo *wi); void (*end) (FFPsyContext *apc); } FFPsyModel; @@ -115,12 +132,24 @@ typedef struct FFPsyModel { * @param num_lens number of possible frame lengths * @param bands scalefactor band lengths for all frame lengths * @param num_bands number of scalefactor bands for all frame lengths + * @param num_groups number of channel groups + * @param group_map array with # of channels in group - 1, for each group * * @return zero if successful, a negative value if not */ -av_cold int ff_psy_init(FFPsyContext *ctx, AVCodecContext *avctx, - int num_lens, - const uint8_t **bands, const int* num_bands); +av_cold int ff_psy_init(FFPsyContext *ctx, AVCodecContext *avctx, int num_lens, + const uint8_t **bands, const int* num_bands, + int num_groups, const uint8_t *group_map); + +/** + * Determine what group a channel belongs to. + * + * @param ctx psymodel context + * @param channel channel to locate the group for + * + * @return pointer to the FFPsyChannelGroup this channel belongs to + */ +FFPsyChannelGroup *ff_psy_find_group(FFPsyContext *ctx, int channel); /** * Cleanup model context at the end. diff --git a/libavcodec/utils.c b/libavcodec/utils.c index fd011cfa19..bc08e350cf 100644 --- a/libavcodec/utils.c +++ b/libavcodec/utils.c @@ -141,6 +141,10 @@ void avcodec_align_dimensions2(AVCodecContext *s, int *width, int *height, int l case PIX_FMT_YUV420P10BE: case PIX_FMT_YUV422P10LE: case PIX_FMT_YUV422P10BE: + case PIX_FMT_YUV444P9LE: + case PIX_FMT_YUV444P9BE: + case PIX_FMT_YUV444P10LE: + case PIX_FMT_YUV444P10BE: w_align= 16; //FIXME check for non mpeg style codecs and use less alignment h_align= 16; if(s->codec_id == CODEC_ID_MPEG2VIDEO || s->codec_id == CODEC_ID_MJPEG || s->codec_id == CODEC_ID_AMV || s->codec_id == CODEC_ID_THP || s->codec_id == CODEC_ID_H264) |