Re: [FFmpeg-devel] [PATCH] libavcodec/qsvenc: fix mpeg2 encoding
> > > > > > It is closer to be merged, but need to confirm which exact line of code > > > make difference. > > > (Your verification on Windows was appreciated.) > > > > As I pointed, the QSV_RUNTIME_VERSION_ATLEAST(q->ver, 1, 11) check fails > > in Windows. > > This version runs without problems and resolves the bug (in Windows). > > Well, let me to explain with more detail: > > V1: > if (avctx->codec_id != AV_CODEC_ID_MPEG2VIDEO && > > > QSV_RUNTIME_VERSION_ATLEAST(q->ver, 1, 11)) > > For mpeg2 case, it is always equal to "if (0)" , right? > > V2: > if (avctx->codec_id != AV_CODEC_ID_MPEG2VIDEO) > > For mpeg2 case, it is equal to "if (0)" too, right? > > Thus why I asked what the difference between V1 and V2 was. I'm sorry! Now I understand what you're saying. It's my fault. Well then... I feel like you're right. However, I can only comment that the first time I compiled your patch it doesn't work. And for this reason I created this one. And now I can confirm that it resolves the problem. Please feel free to approve whatever you think is best. But something has to be done, because the MPEG-2 QSV encoder in Windows is broken. Regards. A.H. --- ___ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-devel To unsubscribe, visit link above, or email ffmpeg-devel-requ...@ffmpeg.org with subject "unsubscribe".
Re: [FFmpeg-devel] [PATCH] libavcodec/qsvenc: fix mpeg2 encoding
> -Original Message- > From: ffmpeg-devel [mailto:ffmpeg-devel-boun...@ffmpeg.org] On Behalf > Of Andreas Håkon > Sent: Wednesday, May 15, 2019 8:55 PM > To: FFmpeg development discussions and patches > > Subject: Re: [FFmpeg-devel] [PATCH] libavcodec/qsvenc: fix mpeg2 encoding > > Hi, > > > > > The difference? > > > In addition to a few extra little checks, It seems that the > > > QSV_RUNTIME_VERSION_ATLEAST(q->ver, 1, 11) check fails every time in > > > Windows (almost in my environment). > > > > If QSV_RUNTIME_VERSION_ATLEAST(q->ver, 1, 11) always equal to zero, > > then co3 won't be set, If codec is MPEG2, co3 won't be set too. > > I can't see any difference for MPEG2 case. > > Your assumption about QSV_RUNTIME_VERSION_ATLEAST(q->ver, 1, 11) is > not real. > This depends on the run-time driver used. > > > > > Futhermore, it has more sense to apply the check inside the > > > conditional preprocessing of #if QSV_HAVE_CO3 and not outside. > > > > Make sense, but no difference for run-time result? > > I prefer to make the code as more legible as possible. > That's the reason for the other additions of #if QSV_HAVE_CO3. > > > > > As a summary: I confirm that my patch works. And it's based on your > > > proposal. So please comment positively to merge it. > > > > It is closer to be merged, but need to confirm which exact line of code > make difference. > > (Your verification on Windows was appreciated.) > > As I pointed, the QSV_RUNTIME_VERSION_ATLEAST(q->ver, 1, 11) check fails > in Windows. > This version runs without problems and resolves the bug (in Windows). Well, let me to explain with more detail: V1: if (avctx->codec_id != AV_CODEC_ID_MPEG2VIDEO && > QSV_RUNTIME_VERSION_ATLEAST(q->ver, 1, 11)) For mpeg2 case, it is always equal to "if (0)" , right? V2: if (avctx->codec_id != AV_CODEC_ID_MPEG2VIDEO) For mpeg2 case, it is equal to "if (0)" too, right? Thus why I asked what the difference between V1 and V2 was. ___ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-devel To unsubscribe, visit link above, or email ffmpeg-devel-requ...@ffmpeg.org with subject "unsubscribe".
Re: [FFmpeg-devel] [PATCH] libavcodec/qsvenc: fix mpeg2 encoding
Hi, > > The difference? > > In addition to a few extra little checks, It seems that the > > QSV_RUNTIME_VERSION_ATLEAST(q->ver, 1, 11) check fails every time in > > Windows (almost in my environment). > > If QSV_RUNTIME_VERSION_ATLEAST(q->ver, 1, 11) always equal to zero, then co3 > won't be set, > If codec is MPEG2, co3 won't be set too. > I can't see any difference for MPEG2 case. Your assumption about QSV_RUNTIME_VERSION_ATLEAST(q->ver, 1, 11) is not real. This depends on the run-time driver used. > > Futhermore, it has more sense to apply the check inside the conditional > > preprocessing of #if QSV_HAVE_CO3 and not outside. > > Make sense, but no difference for run-time result? I prefer to make the code as more legible as possible. That's the reason for the other additions of #if QSV_HAVE_CO3. > > As a summary: I confirm that my patch works. And it's based on your > > proposal. So please comment positively to merge it. > > It is closer to be merged, but need to confirm which exact line of code make > difference. > (Your verification on Windows was appreciated.) As I pointed, the QSV_RUNTIME_VERSION_ATLEAST(q->ver, 1, 11) check fails in Windows. This version runs without problems and resolves the bug (in Windows). So, I suggest to accept this patch. Regards. A.H. --- ___ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-devel To unsubscribe, visit link above, or email ffmpeg-devel-requ...@ffmpeg.org with subject "unsubscribe".
Re: [FFmpeg-devel] [PATCH] libavcodec/qsvenc: fix mpeg2 encoding
> > > Subject: [FFmpeg-devel] [PATCH] libavcodec/qsvenc: fix mpeg2 > > > encoding Fixes bug #7839 > > > https://trac.ffmpeg.org/ticket/7839 > > > Supersedes: > > > #12935 - https://patchwork.ffmpeg.org/patch/12935/ > > > #12872 - https://patchwork.ffmpeg.org/patch/12872/ > > > Regards. > > > A.H. > > > > It was saidhttps://patchwork.ffmpeg.org/patch/12935/ could not fix > #7839, how this one can work? > > > I compiled your patch and doesn't work. > My path works. Glad to know it. > The difference? > In addition to a few extra little checks, It seems that the > QSV_RUNTIME_VERSION_ATLEAST(q->ver, 1, 11) check fails every time in > Windows (almost in my environment). If QSV_RUNTIME_VERSION_ATLEAST(q->ver, 1, 11) always equal to zero, then co3 won't be set, If codec is MPEG2, co3 won't be set too. I can't see any difference for MPEG2 case. > Futhermore, it has more sense to apply the check inside the conditional > preprocessing of #if QSV_HAVE_CO3 and not outside. Make sense, but no difference for run-time result? > As a summary: I confirm that my patch works. And it's based on your > proposal. So please comment positively to merge it. It is closer to be merged, but need to confirm which exact line of code make difference. (Your verification on Windows was appreciated.) > Thank you! > A.H. ___ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-devel To unsubscribe, visit link above, or email ffmpeg-devel-requ...@ffmpeg.org with subject "unsubscribe".
Re: [FFmpeg-devel] [PATCH] libavcodec/qsvenc: fix mpeg2 encoding
Hi, > > Subject: [FFmpeg-devel] [PATCH] libavcodec/qsvenc: fix mpeg2 encoding > > Fixes bug #7839 > > https://trac.ffmpeg.org/ticket/7839 > > Supersedes: > > #12935 - https://patchwork.ffmpeg.org/patch/12935/ > > #12872 - https://patchwork.ffmpeg.org/patch/12872/ > > Regards. > > A.H. > > It was saidhttps://patchwork.ffmpeg.org/patch/12935/ could not fix #7839, how > this one can work? I compiled your patch and doesn't work. My path works. The difference? In addition to a few extra little checks, It seems that the QSV_RUNTIME_VERSION_ATLEAST(q->ver, 1, 11) check fails every time in Windows (almost in my environment). Futhermore, it has more sense to apply the check inside the conditional preprocessing of #if QSV_HAVE_CO3 and not outside. As a summary: I confirm that my patch works. And it's based on your proposal. So please comment positively to merge it. Thank you! A.H. ___ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-devel To unsubscribe, visit link above, or email ffmpeg-devel-requ...@ffmpeg.org with subject "unsubscribe".
Re: [FFmpeg-devel] [PATCH] libavcodec/qsvenc: fix mpeg2 encoding
> From: ffmpeg-devel [mailto:ffmpeg-devel-boun...@ffmpeg.org] On Behalf > Of Andreas Håkon > Sent: Wednesday, May 15, 2019 12:00 AM > To: FFmpeg development discussions and patches > > Subject: [FFmpeg-devel] [PATCH] libavcodec/qsvenc: fix mpeg2 encoding > > Fixes bug #7839 > https://trac.ffmpeg.org/ticket/7839 > > Supersedes: > #12935 - https://patchwork.ffmpeg.org/patch/12935/ > #12872 - https://patchwork.ffmpeg.org/patch/12872/ > > Regards. > A.H. It was said https://patchwork.ffmpeg.org/patch/12935/ could not fix #7839, how this one can work? ___ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org https://ffmpeg.org/mailman/listinfo/ffmpeg-devel To unsubscribe, visit link above, or email ffmpeg-devel-requ...@ffmpeg.org with subject "unsubscribe".