Artifacts on 4k video encoded with ffmpeg placebo preset

[olodar]

Hello, I encoded 4k video with ffmpeg like this:

ffmpeg -i orig.mp4 -preset placebo -c:a copy -c:v libx264 -crf 30 final.mkv

but when I trying play this video I see random artifacts.

[Attachment 73653 - Click to enlarge]
Also I noticed this problem appears only on 4k videos (FullHD is okay) AND if playing with hardware acceleration (when stream decoding by Video Decode gpu block) AND only with placebo preset (veryslow and any another preset doesn't "produce" this artifacts), but if I play with software decoding, everything plays ok without any artifacts.
Thanks!

[jagabo]

Try reducing b-frames to 8.

ffmpeg -i orig.mp4 -preset placebo -c:a copy -c:v libx264 -crf 30 -x264opts bframes=8 final.mkv

[olodar]

Try reducing b-frames to 8.

ffmpeg -i orig.mp4 -preset placebo -c:a copy -c:v libx264 -crf 30 -x264opts bframes=8 final.mkv

Thanks for your answer! Unfortunately it's not help:

[Attachment 73664 - Click to enlarge]

[ProWo]

...but if I play with software decoding, everything plays ok without any artifacts.

That means your problem isn't an encoding problem, but a playing problem.
Forget hardware acceleration for 4k videos on your PC.

[jagabo]

Try reducing b-frames to 8.

ffmpeg -i orig.mp4 -preset placebo -c:a copy -c:v libx264 -crf 30 -x264opts bframes=8 final.mkv

Thanks for your answer! Unfortunately it's not help:

I thought that would be it. There are a few more differences between veryslow and placebo you can try changing:

veryslow:  cabac=1 / ref=16 / deblock=1:0:0 / analyse=0x3:0x133 / me=umh  / subme=10 / psy=1 / psy_rd=1.00:0.00 / mixed_ref=1 / me_range=24 / chroma_me=1 / trellis=2 / 8x8dct=1 / cqm=0 / deadzone=21,11 / fast_pskip=1 / chroma_qp_offset=-2 / threads=6 / sliced_threads=0 / nr=0 / decimate=1 / interlaced=0 / bluray_compat=0 / constrained_intra=0 / bframes=8  / b_pyramid=2 / b_adapt=2 / b_bias=0 / direct=3 / weightb=1 / open_gop=0 / weightp=2 / keyint=250 / keyint_min=23 / scenecut=40 / intra_refresh=0 / rc_lookahead=60 / rc=crf / mbtree=1 / crf=16.0 / qcomp=0.60 / qpmin=0 / qpmax=69 / qpstep=4 / ip_ratio=1.40                 / aq=1:1.00

placebo:   cabac=1 / ref=16 / deblock=1:0:0 / analyse=0x3:0x133 / me=tesa / subme=11 / psy=1 / psy_rd=1.00:0.00 / mixed_ref=1 / me_range=24 / chroma_me=1 / trellis=2 / 8x8dct=1 / cqm=0 / deadzone=21,11 / fast_pskip=0 / chroma_qp_offset=-2 / threads=6 / sliced_threads=0 / nr=0 / decimate=1 / interlaced=0 / bluray_compat=0 / constrained_intra=0 / bframes=16 / b_pyramid=2 / b_adapt=2 / b_bias=0 / direct=3 / weightb=1 / open_gop=0 / weightp=2 / keyint=250 / keyint_min=23 / scenecut=40 / intra_refresh=0 / rc_lookahead=60 / rc=crf / mbtree=1 / crf=16.0 / qcomp=0.60 / qpmin=0 / qpmax=69 / qpstep=4 / ip_ratio=1.40                 / aq=1:1.00

Or just use the veryslow preset. You won't see any difference anyway.

[olodar]

...but if I play with software decoding, everything plays ok without any artifacts.

That means your problem isn't an encoding problem, but a playing problem.
Forget hardware acceleration for 4k videos on your PC.

I tried on different devices - smartphone, laptop, PC - everywhere with hardware decoding artifacts appears. Besides, there's no artifacts with any other preset, this problem only with placebo preset.

[rgr]

Can you attach the problematic fragment?

[olodar]

Can you attach the problematic fragment?

Sure - https://kazakhcloud.gsgt.ru/index.php/s/zs77fEqbqZM8K5T

Also, I was a little wrong, in fact, "placebo" and "veryslow" gives artifacts too. "slower" already encodes good.

[rgr]

Can you attach the problematic fragment?

Sure - https://kazakhcloud.gsgt.ru/index.php/s/zs77fEqbqZM8K5T

Also, I was a little wrong, in fact, "placebo" and "veryslow" gives artifacts too. "slower" already encodes good.

"Internal Server Error

The server encountered an internal error and was unable to complete your request.
Please contact the server administrator if this error reappears multiple times, please include the technical details below in your report.
More details can be found in the server log."

[jagabo]

I was a little wrong, in fact, "placebo" and "veryslow" gives artifacts too. "slower" already encodes good.

In that case try setting bframes=3 and ref=8.

[olodar]

Can you attach the problematic fragment?

Sure - https://kazakhcloud.gsgt.ru/index.php/s/zs77fEqbqZM8K5T

Also, I was a little wrong, in fact, "placebo" and "veryslow" gives artifacts too. "slower" already encodes good.

"Internal Server Error

The server encountered an internal error and was unable to complete your request.
Please contact the server administrator if this error reappears multiple times, please include the technical details below in your report.
More details can be found in the server log."

Excuse me, reup: https://kazakhcloud.gsgt.ru/index.php/s/pFJCGdzgWp4gk98

I was a little wrong, in fact, "placebo" and "veryslow" gives artifacts too. "slower" already encodes good.

In that case try setting bframes=3 and ref=8.

No, some artifacts are still appears. BTW, "x264opts" is deprecated, need use "x264-params" instead - https://trac.ffmpeg.org/wiki/Encode/H.264 "Warning: Do not use the option x264opts, as it will eventually be removed. Use x264-params instead.".

[rgr]

Can you attach the problematic fragment?

Sure - https://kazakhcloud.gsgt.ru/index.php/s/zs77fEqbqZM8K5T

Also, I was a little wrong, in fact, "placebo" and "veryslow" gives artifacts too. "slower" already encodes good.

"Internal Server Error

The server encountered an internal error and was unable to complete your request.
Please contact the server administrator if this error reappears multiple times, please include the technical details below in your report.
More details can be found in the server log."

Excuse me, reup: https://kazakhcloud.gsgt.ru/index.php/s/pFJCGdzgWp4gk98

Download doesn't work -- just nothing happens...

[davexnet]

I was able download it on my phone using Firefox/uBlock origin

[olodar]

Can you attach the problematic fragment?

Sure - https://kazakhcloud.gsgt.ru/index.php/s/zs77fEqbqZM8K5T

Also, I was a little wrong, in fact, "placebo" and "veryslow" gives artifacts too. "slower" already encodes good.

"Internal Server Error

The server encountered an internal error and was unable to complete your request.
Please contact the server administrator if this error reappears multiple times, please include the technical details below in your report.
More details can be found in the server log."

Excuse me, reup: https://kazakhcloud.gsgt.ru/index.php/s/pFJCGdzgWp4gk98

Download doesn't work -- just nothing happens...

Try direct link then - https://kazakhcloud.gsgt.ru/index.php/s/pFJCGdzgWp4gk98/download/SGCAM_20230827_035840892.mp4

[olodar]

I still haven't found solution. Those glitches (artifacts) appears with placebo > veryslow > slower presets. Slow and higher already ok. So, only workaround I have found is use x265 codec for 4k videos. For example, parameters -c:v libx265 -crf 36 -preset placebo offers very good quality (in my opinion). For FullHD videos x264 is still my favorite. However, x265 compression time much higher than x264:

time ffmpeg -i VID_20220129_113412338.mp4 -preset placebo -crf 25 -c:a copy -c:v libx264 x264.mkv

real 36m40.473s
user 424m23.824s
sys 0m15.119s

time ffmpeg -i VID_20220129_113412338.mp4 -preset placebo -crf 30 -c:a copy -c:v libx265 x265.mkv

real 198m33.424s
user 2307m57.016s
sys 0m10.698s

Diff in sizes:
84M x264.mkv
33M x265.mkv

[jagabo]

I was a little wrong, in fact, "placebo" and "veryslow" gives artifacts too. "slower" already encodes good.

Then try reducing b-fframes to 3 and/or ref to 8. Other difference can be seen here:

REM slower     cabac=1 / ref=8  / deblock=1:0:0 / analyse=0x3:0x133 / me=umh  / subme=9  / psy=1 / psy_rd=1.00:0.00 / mixed_ref=1 / me_range=16 / chroma_me=1 / trellis=2 / 8x8dct=1 / cqm=0 / deadzone=21,11 / fast_pskip=1 / chroma_qp_offset=-2 / threads=6 / sliced_threads=0 / nr=0 / decimate=1 / interlaced=0 / bluray_compat=0 / constrained_intra=0 / bframes=3  / b_pyramid=2 / b_adapt=2 / b_bias=0 / direct=3 / weightb=1 / open_gop=0 / weightp=2 / keyint=250 / keyint_min=23 / scenecut=40 / intra_refresh=0 / rc_lookahead=60 / rc=crf / mbtree=1 / crf=18.0 / qcomp=0.60 / qpmin=0 / qpmax=69 / qpstep=4 / ip_ratio=1.40                 / aq=1:1.00
REM veryslow:  cabac=1 / ref=16 / deblock=1:0:0 / analyse=0x3:0x133 / me=umh  / subme=10 / psy=1 / psy_rd=1.00:0.00 / mixed_ref=1 / me_range=24 / chroma_me=1 / trellis=2 / 8x8dct=1 / cqm=0 / deadzone=21,11 / fast_pskip=1 / chroma_qp_offset=-2 / threads=6 / sliced_threads=0 / nr=0 / decimate=1 / interlaced=0 / bluray_compat=0 / constrained_intra=0 / bframes=8  / b_pyramid=2 / b_adapt=2 / b_bias=0 / direct=3 / weightb=1 / open_gop=0 / weightp=2 / keyint=250 / keyint_min=23 / scenecut=40 / intra_refresh=0 / rc_lookahead=60 / rc=crf / mbtree=1 / crf=16.0 / qcomp=0.60 / qpmin=0 / qpmax=69 / qpstep=4 / ip_ratio=1.40                 / aq=1:1.00
REM placebo:   cabac=1 / ref=16 / deblock=1:0:0 / analyse=0x3:0x133 / me=tesa / subme=11 / psy=1 / psy_rd=1.00:0.00 / mixed_ref=1 / me_range=24 / chroma_me=1 / trellis=2 / 8x8dct=1 / cqm=0 / deadzone=21,11 / fast_pskip=0 / chroma_qp_offset=-2 / threads=6 / sliced_threads=0 / nr=0 / decimate=1 / interlaced=0 / bluray_compat=0 / constrained_intra=0 / bframes=16 / b_pyramid=2 / b_adapt=2 / b_bias=0 / direct=3 / weightb=1 / open_gop=0 / weightp=2 / keyint=250 / keyint_min=23 / scenecut=40 / intra_refresh=0 / rc_lookahead=60 / rc=crf / mbtree=1 / crf=16.0 / qcomp=0.60 / qpmin=0 / qpmax=69 / qpstep=4 / ip_ratio=1.40                 / aq=1:1.00

[rgr]

I don't have any artifacts.

ffmpeg -i SGCAM_20230827_035840892.mp4 -preset placebo -crf 25 -c:a copy -c:v libx264 x264.mkv

[Selur]

Decoding the clip from rgr I see artifacts when cuda hardware decoding is used.
https://imgsli.com/MjY4NzE0
(side note: MPC HC switches to software coding for the clip)

[rgr]

Yes indeed. I thought MPC-HC did hardware decoding. Why so? However, it does not load the CPU much, I see that VLC loads the graphics processor more.

[olodar]

I was a little wrong, in fact, "placebo" and "veryslow" gives artifacts too. "slower" already encodes good.

Then try reducing b-fframes to 3 and/or ref to 8. Other difference can be seen here:

REM slower     cabac=1 / ref=8  / deblock=1:0:0 / analyse=0x3:0x133 / me=umh  / subme=9  / psy=1 / psy_rd=1.00:0.00 / mixed_ref=1 / me_range=16 / chroma_me=1 / trellis=2 / 8x8dct=1 / cqm=0 / deadzone=21,11 / fast_pskip=1 / chroma_qp_offset=-2 / threads=6 / sliced_threads=0 / nr=0 / decimate=1 / interlaced=0 / bluray_compat=0 / constrained_intra=0 / bframes=3  / b_pyramid=2 / b_adapt=2 / b_bias=0 / direct=3 / weightb=1 / open_gop=0 / weightp=2 / keyint=250 / keyint_min=23 / scenecut=40 / intra_refresh=0 / rc_lookahead=60 / rc=crf / mbtree=1 / crf=18.0 / qcomp=0.60 / qpmin=0 / qpmax=69 / qpstep=4 / ip_ratio=1.40                 / aq=1:1.00
REM veryslow:  cabac=1 / ref=16 / deblock=1:0:0 / analyse=0x3:0x133 / me=umh  / subme=10 / psy=1 / psy_rd=1.00:0.00 / mixed_ref=1 / me_range=24 / chroma_me=1 / trellis=2 / 8x8dct=1 / cqm=0 / deadzone=21,11 / fast_pskip=1 / chroma_qp_offset=-2 / threads=6 / sliced_threads=0 / nr=0 / decimate=1 / interlaced=0 / bluray_compat=0 / constrained_intra=0 / bframes=8  / b_pyramid=2 / b_adapt=2 / b_bias=0 / direct=3 / weightb=1 / open_gop=0 / weightp=2 / keyint=250 / keyint_min=23 / scenecut=40 / intra_refresh=0 / rc_lookahead=60 / rc=crf / mbtree=1 / crf=16.0 / qcomp=0.60 / qpmin=0 / qpmax=69 / qpstep=4 / ip_ratio=1.40                 / aq=1:1.00
REM placebo:   cabac=1 / ref=16 / deblock=1:0:0 / analyse=0x3:0x133 / me=tesa / subme=11 / psy=1 / psy_rd=1.00:0.00 / mixed_ref=1 / me_range=24 / chroma_me=1 / trellis=2 / 8x8dct=1 / cqm=0 / deadzone=21,11 / fast_pskip=0 / chroma_qp_offset=-2 / threads=6 / sliced_threads=0 / nr=0 / decimate=1 / interlaced=0 / bluray_compat=0 / constrained_intra=0 / bframes=16 / b_pyramid=2 / b_adapt=2 / b_bias=0 / direct=3 / weightb=1 / open_gop=0 / weightp=2 / keyint=250 / keyint_min=23 / scenecut=40 / intra_refresh=0 / rc_lookahead=60 / rc=crf / mbtree=1 / crf=16.0 / qcomp=0.60 / qpmin=0 / qpmax=69 / qpstep=4 / ip_ratio=1.40                 / aq=1:1.00

Yes, I tried to play with those parameters, didn't found what exactly affects.

I don't have any artifacts.

ffmpeg -i SGCAM_20230827_035840892.mp4 -preset placebo -crf 25 -c:a copy -c:v libx264 x264.mkv

[Attachment 79535 - Click to enlarge]

Decoding the clip from rgr I see artifacts when cuda hardware decoding is used.
https://imgsli.com/MjY4NzE0
(side note: MPC HC switches to software coding for the clip)

Yes, for some reason hardware decoders produces artifacts, while software decoders plays well.

[jagabo]

Hardware decoders don't support all possible h.264 parameters.

So far I've found that ffmpeg x264lib with preset=placebo, ref=5, trellis=1, and subme=8 plays properly with the SGCAM source in post #14 on my Nvidia RTX 4060 with VLC and hardware decoding.

ffmpeg -i SGCAM_20230827_035840892.mp4 -c:a copy -c:v libx264 -preset placebo -x264opts ref=5:trellis=1:subme=8 x264.placebo+ref=5+trellis=1+subme=8.mp4

Leaving any one of those modifiers out leads to playback errors, especially trellis. By the way the default values when using preset placebo are ref=16, trellis=2, and subme=11.