avconv_qsv: use the actual pixel format provided by lavc

Do not hardcode NV12. This allows 10bit decoding with -hwaccel qsv.
This commit is contained in:
Anton Khirnov 2016-07-13 09:16:35 +02:00
parent 6f40181cad
commit 8b7a9729aa

View File

@ -79,7 +79,7 @@ int qsv_init(AVCodecContext *s)
frames_ctx->width = FFALIGN(s->coded_width, 32);
frames_ctx->height = FFALIGN(s->coded_height, 32);
frames_ctx->format = AV_PIX_FMT_QSV;
frames_ctx->sw_format = AV_PIX_FMT_NV12;
frames_ctx->sw_format = s->sw_pix_fmt;
frames_ctx->initial_pool_size = 32;
frames_hwctx->frame_type = MFX_MEMTYPE_VIDEO_MEMORY_DECODER_TARGET;