XPSNR fails on very similar or identical pictures (score: inf)

[gitoss]

... continuing from #244 :-)

AB-AV1 0.9 fails with an ugly ffmpeg output on very similar or identical pictures because the xpsnr score isn't numeric, but 'inf' (try to run ab-av1.exe xpsnr --reference same.png --distorted same.png

[Parsed_xpsnr_0 @ 0000000000f36540] XPSNR  r: inf
XPSNR average, 1 frames  r: inf  g: inf  b: inf  (minimum: inf)
[out#0/null @ 0000000000f004c0] video:0KiB audio:0KiB subtitle:0KiB other streams:0KiB global headers:0KiB muxing overhead: unknown
frame=    1 fps=0.0 q=-0.0 Lsize=N/A time=00:00:00.04 bitrate=N/A speed=0.434x

A possible hotfix could be to simple limit the xpsnr score to 99 (and map inf to 99, too) - on the logarithmic psnr scale 50-ish is lossless, so 99 should be a safe 'inf' dummy value.

Otherwise thanks for implementing, xpsnr seems to be a bit faster than vmaf which is nice for bulk crf-search image conversion.

[alexheretic]

Thanks, should be straightforward to fix. I'll look when i get some time.

…

On Tue, 24 Dec 2024, 23:37 gitoss, ***@***.***> wrote: ... continuing from #244 <#244> :-) AB-AV1 0.9 fails with an ugly ffmpeg output on very similar or identical pictures because the xpsnr score isn't numeric, but 'inf' (try to run ab-av1.exe xpsnr --reference same.png --distorted same.png [Parsed_xpsnr_0 @ 0000000000f36540] XPSNR r: inf XPSNR average, 1 frames r: inf g: inf b: inf (minimum: inf) [out#0/null @ 0000000000f004c0] video:0KiB audio:0KiB subtitle:0KiB other streams:0KiB global headers:0KiB muxing overhead: unknown frame= 1 fps=0.0 q=-0.0 Lsize=N/A time=00:00:00.04 bitrate=N/A speed=0.434x A possible hotfix could be to simple limit the xpsnr score to 99 (and map inf to 99, too) - on the logarithmic psnr scale 50-ish is lossless, so 99 should be a safe 'inf' dummy value. Otherwise thanks for implementing, xpsnr seems to be a bit faster than vmaf which is nice for bulk crf-search image conversion. — Reply to this email directly, view it on GitHub <#256>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AARZHV45PRNWK62TD6HQRTL2HHV25AVCNFSM6AAAAABUFJCOBGVHI2DSMVQWIX3LMV43ASLTON2WKOZSG42TQMZVGQZDEMQ> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>