[FFmpeg-devel] evaluating the experimental status of ffv1 version 3

[Michael Niedermayer]

On Tue, Oct 02, 2012 at 03:38:28PM +0200, Peter B. wrote:
> Quoting "Peter B." <pb at das-werkstatt.com>:
> >For YUV422 10bit, I'll use the a converted set of the SVT
> >collection from the "Video Quality Experts Group" (VQEG) FTP
> >server [4].
> >== References:
> >[4] ftp://vqeg.its.bldrdoc.gov/HDTV/SVT_exports/SVT_YUV10_Exports_
> 
> In the README of the SVT set at VQEG, it says that the YUV422 10bit
> are 0-padded to 4bits, and ffmpeg says it's "v210x, yuv422p16le".
> 
> What would be the correct way to generate framemd5 checksums for the
> .yuv10 image sequence source, in order to verify against
> yuv422p10le?
> 
> I've created 2 framemd5 files for the source:
> 
> // --------------
> $ ffmpeg -f image2 -s 1920x1080 -r 25 -i
> 'NewMobCal_YUV10_1080i25_/NM_%05d.yuv10' -pix_fmt yuv422p16le -an -f
> framemd5 NewMobCal_YUV10_1080i25-p16le.framemd5
> // --------------
> $ ffmpeg -f image2 -s 1920x1080 -r 25 -i
> 'NewMobCal_YUV10_1080i25_/NM_%05d.yuv10' -pix_fmt yuv422p10le -an -f
> framemd5 NewMobCal_YUV10_1080i25-p10le.framemd5
> // --------------
> 
> Then I create 2 FFv1.3 output files. One with "yuv422p10le" and one
> with "yuv422p16le":
> // --------------
> ffmpeg -f image2 -s 1920x1080 -r 25 -i 'client/VQEG/HDTV/SVT_exports/SVT_YUV10_Exports_/NewMobCal_YUV10_1080i25_/NM_%05d.yuv10'
> -an -vcodec ffv1 -strict experimental -slices 24 -slicecrc 0 -g 1
> -pix_fmt yuv422p16le client/test-p16le.avi
> // --------------
> ffmpeg -f image2 -s 1920x1080 -r 25 -i 'client/VQEG/HDTV/SVT_exports/SVT_YUV10_Exports_/NewMobCal_YUV10_1080i25_/NM_%05d.yuv10'
> -an -vcodec ffv1 -strict experimental -slices 24 -slicecrc 0 -g 1
> -pix_fmt yuv422p10le client/test-p10le.avi
> // --------------
> 
> When comparing the generated p10le.framemd5 of the source with the
> framemd5 from a yuv422p10le ffv1.3, they match.
> 
> However, if I create a yuv422p16le ffv1.3, the framemd5 do *NOT*
> match with the p16le.framemd5 from the source.
> 
> Is there an error on my side?

maybe

the problem is that you are not working with 16bit material and the
encoder just stores 10bit because thats what v210x tells it is in
there, somewhere this ends up introducing a difference


if you first convert to 16bit raw and store that seperately then md5s
match

... -an -pix_fmt yuv422p16le -vcodec rawvideo  test16.nut

[...]
-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

The misfortune of the wise is better than the prosperity of the fool.
-- Epicurus
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 198 bytes
Desc: Digital signature
URL: <http://ffmpeg.org/pipermail/ffmpeg-devel/attachments/20121003/7f76f2d9/attachment.asc>