Re: [FFmpeg-devel] avcodec/prores_ks : use official quant matrix

2018-04-22 Thread Martin Vignali
2018-04-14 11:31 GMT+02:00 Martin Vignali : > > > >> > For the proxy chroma version is also the same matrix use in prores_aw >> >> Patch is OK then, assumming FATE still passes. >> >> >> Ok thanks for comments. > It passes fate test for me (but prores_ks have few fate test !) > > I will wait few d

Re: [FFmpeg-devel] avcodec/prores_ks : use official quant matrix

2018-04-14 Thread Martin Vignali
> > > > For the proxy chroma version is also the same matrix use in prores_aw > > Patch is OK then, assumming FATE still passes. > > > Ok thanks for comments. It passes fate test for me (but prores_ks have few fate test !) I will wait few days, if someone else have comments, before apply. Martin

Re: [FFmpeg-devel] avcodec/prores_ks : use official quant matrix

2018-04-13 Thread Paul B Mahol
On 4/13/18, Martin Vignali wrote: >> You sure those are actually used in binary? >> >> > luma, chroma matrix in prores is write in each frame. > So i take these values, from a real sample (prores made with official > encoder (and for xq, you can also, compare with arri sample footage) > > For the

Re: [FFmpeg-devel] avcodec/prores_ks : use official quant matrix

2018-04-13 Thread Martin Vignali
> You sure those are actually used in binary? > > luma, chroma matrix in prores is write in each frame. So i take these values, from a real sample (prores made with official encoder (and for xq, you can also, compare with arri sample footage) For the proxy chroma version is also the same matrix us

Re: [FFmpeg-devel] avcodec/prores_ks : use official quant matrix

2018-04-13 Thread Paul B Mahol
On 4/13/18, Martin Vignali wrote: > Hello, > > Patch in attach, add support to a separate luma and chroma matrix > for prores encoding > > like proxy and xq codec version, doesn't have the same luma and chroma > matrix > > The official xq luma matrix is not use for now (need to fix something > els