avcodec/cuvid: add support for hardware deinterlacing

Currently does not work with the ffmpeg cli tool, due do it using the
old one in one out API.
Anything using the new API, like mpv, can make use of it, provided it is
prepared for a decoder modifying the framerate and outputing multiple
frames per input. FFmpeg itself is not.
This commit is contained in:
Timo Rothenpieler 2016-09-18 02:42:52 +02:00
parent 3b24020b54
commit 0b420886a4
1 changed files with 82 additions and 21 deletions

View File

@ -25,16 +25,19 @@
#include "libavutil/hwcontext_cuda.h"
#include "libavutil/fifo.h"
#include "libavutil/log.h"
#include "libavutil/opt.h"
#include "avcodec.h"
#include "internal.h"
#include <nvcuvid.h>
#define MAX_FRAME_COUNT 20
#define MAX_FRAME_COUNT 25
typedef struct CuvidContext
{
AVClass *avclass;
CUvideodecoder cudecoder;
CUvideoparser cuparser;
@ -45,6 +48,9 @@ typedef struct CuvidContext
AVFifoBuffer *frame_queue;
int deint_mode;
int64_t prev_pts;
int internal_error;
int ever_flushed;
int decoder_flushing;
@ -56,6 +62,13 @@ typedef struct CuvidContext
CUVIDEOFORMATEX cuparse_ext;
} CuvidContext;
typedef struct CuvidParsedFrame
{
CUVIDPARSERDISPINFO dispinfo;
int second_field;
int is_deinterlacing;
} CuvidParsedFrame;
static int check_cu(AVCodecContext *avctx, CUresult err, const char *func)
{
const char *err_name;
@ -86,7 +99,7 @@ static int CUDAAPI cuvid_handle_video_sequence(void *opaque, CUVIDEOFORMAT* form
AVHWFramesContext *hwframe_ctx = (AVHWFramesContext*)ctx->hwframe->data;
CUVIDDECODECREATEINFO cuinfo;
av_log(avctx, AV_LOG_TRACE, "pfnSequenceCallback\n");
av_log(avctx, AV_LOG_TRACE, "pfnSequenceCallback, progressive_sequence=%d\n", format->progressive_sequence);
ctx->internal_error = 0;
@ -97,7 +110,7 @@ static int CUDAAPI cuvid_handle_video_sequence(void *opaque, CUVIDEOFORMAT* form
(AVRational){ format->display_aspect_ratio.x, format->display_aspect_ratio.y },
(AVRational){ avctx->width, avctx->height }));
if (!format->progressive_sequence)
if (!format->progressive_sequence && ctx->deint_mode == cudaVideoDeinterlaceMode_Weave)
avctx->flags |= AV_CODEC_FLAG_INTERLACED_DCT;
else
avctx->flags &= ~AV_CODEC_FLAG_INTERLACED_DCT;
@ -169,7 +182,14 @@ static int CUDAAPI cuvid_handle_video_sequence(void *opaque, CUVIDEOFORMAT* form
cuinfo.ulNumOutputSurfaces = 1;
cuinfo.ulCreationFlags = cudaVideoCreate_PreferCUVID;
cuinfo.DeinterlaceMode = cudaVideoDeinterlaceMode_Weave;
if (format->progressive_sequence) {
ctx->deint_mode = cuinfo.DeinterlaceMode = cudaVideoDeinterlaceMode_Weave;
} else {
cuinfo.DeinterlaceMode = ctx->deint_mode;
}
if (ctx->deint_mode != cudaVideoDeinterlaceMode_Weave)
avctx->framerate = av_mul_q(avctx->framerate, (AVRational){2, 1});
ctx->internal_error = CHECK_CU(cuvidCreateDecoder(&ctx->cudecoder, &cuinfo));
if (ctx->internal_error < 0)
@ -208,12 +228,18 @@ static int CUDAAPI cuvid_handle_picture_display(void *opaque, CUVIDPARSERDISPINF
{
AVCodecContext *avctx = opaque;
CuvidContext *ctx = avctx->priv_data;
av_log(avctx, AV_LOG_TRACE, "pfnDisplayPicture\n");
CuvidParsedFrame parsed_frame = { *dispinfo, 0, 0 };
ctx->internal_error = 0;
av_fifo_generic_write(ctx->frame_queue, dispinfo, sizeof(CUVIDPARSERDISPINFO), NULL);
if (ctx->deint_mode == cudaVideoDeinterlaceMode_Weave) {
av_fifo_generic_write(ctx->frame_queue, &parsed_frame, sizeof(CuvidParsedFrame), NULL);
} else {
parsed_frame.is_deinterlacing = 1;
av_fifo_generic_write(ctx->frame_queue, &parsed_frame, sizeof(CuvidParsedFrame), NULL);
parsed_frame.second_field = 1;
av_fifo_generic_write(ctx->frame_queue, &parsed_frame, sizeof(CuvidParsedFrame), NULL);
}
return 1;
}
@ -234,7 +260,7 @@ static int cuvid_decode_packet(AVCodecContext *avctx, const AVPacket *avpkt)
if (is_flush && avpkt && avpkt->size)
return AVERROR_EOF;
if (av_fifo_size(ctx->frame_queue) / sizeof(CUVIDPARSERDISPINFO) > MAX_FRAME_COUNT - 2 && avpkt && avpkt->size)
if (av_fifo_size(ctx->frame_queue) / sizeof(CuvidParsedFrame) > MAX_FRAME_COUNT - 2 && avpkt && avpkt->size)
return AVERROR(EAGAIN);
if (ctx->bsf && avpkt && avpkt->size) {
@ -330,20 +356,20 @@ static int cuvid_output_frame(AVCodecContext *avctx, AVFrame *frame)
return ret;
if (av_fifo_size(ctx->frame_queue)) {
CUVIDPARSERDISPINFO dispinfo;
CuvidParsedFrame parsed_frame;
CUVIDPROCPARAMS params;
unsigned int pitch = 0;
int offset = 0;
int i;
av_fifo_generic_read(ctx->frame_queue, &dispinfo, sizeof(CUVIDPARSERDISPINFO), NULL);
av_fifo_generic_read(ctx->frame_queue, &parsed_frame, sizeof(CuvidParsedFrame), NULL);
memset(&params, 0, sizeof(params));
params.progressive_frame = dispinfo.progressive_frame;
params.second_field = 0;
params.top_field_first = dispinfo.top_field_first;
params.progressive_frame = parsed_frame.dispinfo.progressive_frame;
params.second_field = parsed_frame.second_field;
params.top_field_first = parsed_frame.dispinfo.top_field_first;
ret = CHECK_CU(cuvidMapVideoFrame(ctx->cudecoder, dispinfo.picture_index, &mapped_frame, &pitch, &params));
ret = CHECK_CU(cuvidMapVideoFrame(ctx->cudecoder, parsed_frame.dispinfo.picture_index, &mapped_frame, &pitch, &params));
if (ret < 0)
goto error;
@ -419,9 +445,20 @@ static int cuvid_output_frame(AVCodecContext *avctx, AVFrame *frame)
frame->width = avctx->width;
frame->height = avctx->height;
if (avctx->pkt_timebase.num && avctx->pkt_timebase.den)
frame->pts = av_rescale_q(dispinfo.timestamp, (AVRational){1, 10000000}, avctx->pkt_timebase);
frame->pts = av_rescale_q(parsed_frame.dispinfo.timestamp, (AVRational){1, 10000000}, avctx->pkt_timebase);
else
frame->pts = dispinfo.timestamp;
frame->pts = parsed_frame.dispinfo.timestamp;
if (parsed_frame.second_field) {
if (ctx->prev_pts == INT64_MIN) {
ctx->prev_pts = frame->pts;
frame->pts += (avctx->pkt_timebase.den * avctx->framerate.den) / (avctx->pkt_timebase.num * avctx->framerate.num);
} else {
int pts_diff = (frame->pts - ctx->prev_pts) / 2;
ctx->prev_pts = frame->pts;
frame->pts += pts_diff;
}
}
/* CUVIDs opaque reordering breaks the internal pkt logic.
* So set pkt_pts and clear all the other pkt_ fields.
@ -431,10 +468,10 @@ static int cuvid_output_frame(AVCodecContext *avctx, AVFrame *frame)
av_frame_set_pkt_duration(frame, 0);
av_frame_set_pkt_size(frame, -1);
frame->interlaced_frame = !dispinfo.progressive_frame;
frame->interlaced_frame = !parsed_frame.is_deinterlacing && !parsed_frame.dispinfo.progressive_frame;
if (!dispinfo.progressive_frame)
frame->top_field_first = dispinfo.top_field_first;
if (frame->interlaced_frame)
frame->top_field_first = parsed_frame.dispinfo.top_field_first;
} else if (ctx->decoder_flushing) {
ret = AVERROR_EOF;
} else {
@ -461,6 +498,11 @@ static int cuvid_decode_frame(AVCodecContext *avctx, void *data, int *got_frame,
av_log(avctx, AV_LOG_TRACE, "cuvid_decode_frame\n");
if (ctx->deint_mode != cudaVideoDeinterlaceMode_Weave) {
av_log(avctx, AV_LOG_ERROR, "Deinterlacing is not supported via the old API\n");
return AVERROR(EINVAL);
}
if (!ctx->decoder_flushing) {
ret = cuvid_decode_packet(avctx, avpkt);
if (ret < 0)
@ -568,7 +610,7 @@ static av_cold int cuvid_decode_init(AVCodecContext *avctx)
return ret;
}
ctx->frame_queue = av_fifo_alloc(MAX_FRAME_COUNT * sizeof(CUVIDPARSERDISPINFO));
ctx->frame_queue = av_fifo_alloc(MAX_FRAME_COUNT * sizeof(CuvidParsedFrame));
if (!ctx->frame_queue) {
ret = AVERROR(ENOMEM);
goto error;
@ -763,6 +805,8 @@ static av_cold int cuvid_decode_init(AVCodecContext *avctx)
ctx->ever_flushed = 0;
ctx->prev_pts = INT64_MIN;
if (!avctx->pkt_timebase.num || !avctx->pkt_timebase.den)
av_log(avctx, AV_LOG_WARNING, "Invalid pkt_timebase, passing timestamps as-is.\n");
@ -790,7 +834,7 @@ static void cuvid_flush(AVCodecContext *avctx)
av_fifo_freep(&ctx->frame_queue);
ctx->frame_queue = av_fifo_alloc(MAX_FRAME_COUNT * sizeof(CUVIDPARSERDISPINFO));
ctx->frame_queue = av_fifo_alloc(MAX_FRAME_COUNT * sizeof(CuvidParsedFrame));
if (!ctx->frame_queue) {
av_log(avctx, AV_LOG_ERROR, "Failed to recreate frame queue on flush\n");
return;
@ -831,7 +875,23 @@ static void cuvid_flush(AVCodecContext *avctx)
av_log(avctx, AV_LOG_ERROR, "CUDA reinit on flush failed\n");
}
#define OFFSET(x) offsetof(CuvidContext, x)
#define VD AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_DECODING_PARAM
static const AVOption options[] = {
{ "deint", "Set deinterlacing mode", OFFSET(deint_mode), AV_OPT_TYPE_INT, { .i64 = cudaVideoDeinterlaceMode_Weave }, cudaVideoDeinterlaceMode_Weave, cudaVideoDeinterlaceMode_Adaptive, VD, "deint" },
{ "weave", "Weave deinterlacing (do nothing)", 0, AV_OPT_TYPE_CONST, { .i64 = cudaVideoDeinterlaceMode_Weave }, 0, 0, VD, "deint" },
{ "bob", "Bob deinterlacing", 0, AV_OPT_TYPE_CONST, { .i64 = cudaVideoDeinterlaceMode_Bob }, 0, 0, VD, "deint" },
{ "adaptive", "Adaptive deinterlacing", 0, AV_OPT_TYPE_CONST, { .i64 = cudaVideoDeinterlaceMode_Adaptive }, 0, 0, VD, "deint" },
{ NULL }
};
#define DEFINE_CUVID_CODEC(x, X) \
static const AVClass x##_cuvid_class = { \
.class_name = #x "_cuvid", \
.item_name = av_default_item_name, \
.option = options, \
.version = LIBAVUTIL_VERSION_INT, \
}; \
AVHWAccel ff_##x##_cuvid_hwaccel = { \
.name = #x "_cuvid", \
.type = AVMEDIA_TYPE_VIDEO, \
@ -844,6 +904,7 @@ static void cuvid_flush(AVCodecContext *avctx)
.type = AVMEDIA_TYPE_VIDEO, \
.id = AV_CODEC_ID_##X, \
.priv_data_size = sizeof(CuvidContext), \
.priv_class = &x##_cuvid_class, \
.init = cuvid_decode_init, \
.close = cuvid_decode_end, \
.decode = cuvid_decode_frame, \