Implemented a workaround for a potential crash when scaling frames in VideoFFmpegComponent.

This commit is contained in:
Leon Styhre 2021-05-15 20:16:48 +02:00
parent c4e542a7a4
commit 282d2fa445

View file

@ -221,7 +221,7 @@ void VideoFFmpegComponent::readFrames()
// We have a video frame that needs conversion to RGBA format.
uint8_t* frameRGBA[4];
int lineSizes[4];
int lineSize[4];
int allocatedSize = 0;
// The pts value is the presentation time, i.e. the time stamp when
@ -231,10 +231,20 @@ void VideoFFmpegComponent::readFrames()
double pts = static_cast<double>(mPacket->dts) *
av_q2d(mVideoStream->time_base);
// Due to some unknown reason, attempting to scale frames where
// coded_width is larger than width leads to graphics corruption or
// crashes. The only workaround I've been able to find is to decrease the
// source width by one pixel. Unfortunately this leads to a noticeably
// softer picture, but as few videos have this issue it's an acceptable
// workaround for now. Possibly this problem is caused by an FFmpeg bug.
int sourceWidth = mVideoCodecContext->width;
if (mVideoCodecContext->coded_width > mVideoCodecContext->width)
sourceWidth--;
// Conversion using libswscale.
struct SwsContext* conversionContext = sws_getContext(
mVideoCodecContext->coded_width,
mVideoCodecContext->coded_height,
sourceWidth,
mVideoCodecContext->height,
mVideoCodecContext->pix_fmt,
mVideoFrame->width,
mVideoFrame->height,
@ -246,7 +256,7 @@ void VideoFFmpegComponent::readFrames()
allocatedSize = av_image_alloc(
frameRGBA,
lineSizes,
lineSize,
mVideoFrame->width,
mVideoFrame->height,
AV_PIX_FMT_RGB32,
@ -257,9 +267,9 @@ void VideoFFmpegComponent::readFrames()
const_cast<uint8_t const* const*>(mVideoFrame->data),
mVideoFrame->linesize,
0,
mVideoCodecContext->coded_height,
mVideoCodecContext->height,
frameRGBA,
lineSizes);
lineSize);
VideoFrame currFrame;