On 6/23/15, Nicolas George <geo...@nsup.org> wrote: > Le quintidi 5 messidor, an CCXXIII, Paul B Mahol a ecrit : >> Looks like there are no new comments so I can safely assume current >> state as is is good enough. > > If you want to be really safe, you can observe that there is an unanswered > question and assume that <20 hours is not enough to be sure there will be > no > answer. So here it is right now, since you use ultimatums: > >>> Just to put it into same file? >>> Moving it into same filter would be also possible but how would then >>> new filter be named then? > > That depends on the code, you know it better than me. > > If it often makes sense to compute both the PSNR and the SSIM, then a > single > filter would be preferred, especially if some of the computations can be > used for both.
Well tiny_ssim computes PSNR from SSIM but numbers do not match other implementations. I'm sure that our MSE(and PSNR) calculations in vf_psnr.c matches numbers from some page linked from wikipedia. Computations of PSNR and SSIM have not much in common. So the only code that is actually shared is set_meta() which could be moved to some other place in lavfi. > If this is done in a single filter, various names come to mind: -vf quality > is the most obvious. I just do not like big filters, I made mistake putting multiple different filters in single histogram filter. _______________________________________________ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
