aboutsummaryrefslogtreecommitdiffstats
path: root/libavutil/hwcontext_cuda.c
diff options
context:
space:
mode:
authorYogender Gupta <ygupta@nvidia.com>2017-07-18 16:35:53 +0530
committerPhilip Langdale <philipl@overt.org>2017-08-03 15:25:35 -0700
commit3407d8118c7236c94dc0eddabc82041e5c130201 (patch)
tree5d963be1d107db15596135ea41de8a5d2ac9dd01 /libavutil/hwcontext_cuda.c
parent2e8679373ab628a19885645ed5e1271be7797600 (diff)
downloadffmpeg-3407d8118c7236c94dc0eddabc82041e5c130201.tar.gz
hwcontext_cuda : Support YUV444P16 format
Signed-off-by: Philip Langdale <philipl@overt.org>
Diffstat (limited to 'libavutil/hwcontext_cuda.c')
-rw-r--r--libavutil/hwcontext_cuda.c8
1 files changed, 7 insertions, 1 deletions
diff --git a/libavutil/hwcontext_cuda.c b/libavutil/hwcontext_cuda.c
index ed595c3e0f..dfb67bc941 100644
--- a/libavutil/hwcontext_cuda.c
+++ b/libavutil/hwcontext_cuda.c
@@ -37,6 +37,7 @@ static const enum AVPixelFormat supported_formats[] = {
AV_PIX_FMT_YUV444P,
AV_PIX_FMT_P010,
AV_PIX_FMT_P016,
+ AV_PIX_FMT_YUV444P16,
};
static int cuda_frames_get_constraints(AVHWDeviceContext *ctx,
@@ -142,6 +143,9 @@ static int cuda_frames_init(AVHWFramesContext *ctx)
case AV_PIX_FMT_P016:
size = aligned_width * ctx->height * 3;
break;
+ case AV_PIX_FMT_YUV444P16:
+ size = aligned_width * ctx->height * 6;
+ break;
default:
av_log(ctx, AV_LOG_ERROR, "BUG: Pixel format missing from size calculation.");
return AVERROR_BUG;
@@ -161,7 +165,8 @@ static int cuda_get_buffer(AVHWFramesContext *ctx, AVFrame *frame)
int width_in_bytes = ctx->width;
if (ctx->sw_format == AV_PIX_FMT_P010 ||
- ctx->sw_format == AV_PIX_FMT_P016) {
+ ctx->sw_format == AV_PIX_FMT_P016 ||
+ ctx->sw_format == AV_PIX_FMT_YUV444P16) {
width_in_bytes *= 2;
}
aligned_width = FFALIGN(width_in_bytes, CUDA_FRAME_ALIGNMENT);
@@ -188,6 +193,7 @@ static int cuda_get_buffer(AVHWFramesContext *ctx, AVFrame *frame)
frame->linesize[2] = aligned_width / 2;
break;
case AV_PIX_FMT_YUV444P:
+ case AV_PIX_FMT_YUV444P16:
frame->data[0] = frame->buf[0]->data;
frame->data[1] = frame->data[0] + aligned_width * ctx->height;
frame->data[2] = frame->data[1] + aligned_width * ctx->height;