Any chance of a new build? Yesterdays MPV commits solved some problems for me but I’m missing your sweet sweet TrueHD fixes.
Edit: just noticed you have a build running already ![]()
Any chance of a new build? Yesterdays MPV commits solved some problems for me but I’m missing your sweet sweet TrueHD fixes.
Edit: just noticed you have a build running already ![]()
Just updated the repo(s) an hour ago. The new build should be up in a few hours - its currently compiling. ![]()
The new build is up! ![]()
Thanks, it is working well here.
FYI for anyone using this new build on a TV (or any other display targeting a 2.2 gamma) you should add “treat-srgb-as-power22=both” to your mpv config, as it seems to be targeting the srgb gamma curve by default now and you’ll have crushed blacks without this setting.
Sorry if I’m somewhat bit off-topic. I’m using Mitzsch’s libmpv-2.dll because the version that comes with PLEX HTPC seems to disable hardware decoding for 8K videos, making them unwatchable.
I have observed that the latest build 2025-10-30 19:12 produces much more frame drops than the earlier build 2025-09-21 12:23. I have two logs accordingly, and I attach them along with my mpv.conf.
As a workaround, I have reverted to the old version. Any help would be appreciated.
Plex HTPC libmpv-2 20251030.log (651.5 KB)
Plex HTPC libmpv-2 20250921.log (661.9 KB)
mpv.conf.txt (4.3 KB)
There have been many commits between these two MPV versions, and your mpv.conf seems complex. Could you please start with versions from in between (Releases · mitzsch/mpv-winbuild · GitHub), and if this does not help, try a minimal mpv.conf and check again?
Thank you very much for your reply. As far as I can tell, build 2025-09-21 12:23 was the last “good” version. I have therefore repeated the test comparing it with next one (build 2025-10-04 11:03), using a minimal mpv.conf. Please let me know if you need any additional information.
logs & conf.zip (203.3 KB)
Thanks for testing. So the “problem” must be in a commit between 09/21 and 10/4. I somehow suspect it’s an ffmpeg-related issue. What I also noticed in your logs are the lines =>
Nov 02, 2025 14:27:47.648 [6484] DEBUG - [MPVEngine/mpv] ffmpeg: stream level seek from 214214092 to 352413202
Nov 02, 2025 14:27:47.673 [6484] DEBUG - [MPVEngine/mpv] ffmpeg: stream level seek from 352426342 to 214206618
Nov 02, 2025 14:27:47.675 [6484] DEBUG - [MPVEngine/mpv] ffmpeg: stream level seek from 214219758 to 352426289
Nov 02, 2025 14:27:47.678 [6484] DEBUG - [MPVEngine/mpv] ffmpeg: stream level seek from 352742133 to 214568182
Nov 02, 2025 14:27:47.680 [6484] DEBUG - [MPVEngine/mpv] ffmpeg: stream level seek from 214576039 to 352733719
Nov 02, 2025 14:27:47.682 [6484] DEBUG - [MPVEngine/mpv] ffmpeg: stream level seek from 352741576 to 214568545
Nov 02, 2025 14:27:47.685 [6484] DEBUG - [MPVEngine/mpv] ffmpeg: stream level seek from 214576402 to 352734449
Nov 02, 2025 14:27:47.687 [6484] DEBUG - [MPVEngine/mpv] ffmpeg: stream level seek from 352776789 to 214568926
Nov 02, 2025 14:27:47.690 [6484] DEBUG - [MPVEngine/mpv] ffmpeg: stream level seek from 214576783 to 352749721
Nov 02, 2025 14:27:47.692 [6484] DEBUG - [MPVEngine/mpv] ffmpeg: stream level seek from 352850461 to 214601807
Nov 02, 2025 14:27:47.694 [6484] DEBUG - [MPVEngine/mpv] ffmpeg: stream level seek from 214609664 to 352837945
Nov 02, 2025 14:27:47.696 [6484] DEBUG - [MPVEngine/mpv] ffmpeg: stream level seek from 352845802 to 214602171
Nov 02, 2025 14:27:47.698 [6484] DEBUG - [MPVEngine/mpv] ffmpeg: stream level seek from 214610028 to 352838688
Nov 02, 2025 14:27:47.701 [6484] DEBUG - [MPVEngine/mpv] ffmpeg: stream level seek from 352846545 to 214602546
Nov 02, 2025 14:27:47.703 [6484] DEBUG - [MPVEngine/mpv] ffmpeg: stream level seek from 214610403 to 352846604
Nov 02, 2025 14:27:47.708 [6484] DEBUG - [MPVEngine/mpv] ffmpeg: stream level seek from 354618880 to 214730327
Nov 02, 2025 14:27:47.709 [6484] DEBUG - [MPVEngine/mpv] ffmpeg: stream level seek from 214738184 to 354591448
Nov 02, 2025 14:27:47.712 [6484] DEBUG - [MPVEngine/mpv] ffmpeg: stream level seek from 354633788 to 214730693
Nov 02, 2025 14:27:47.714 [6484] DEBUG - [MPVEngine/mpv] ffmpeg: stream level seek from 214743833 to 354610599
Nov 02, 2025 14:27:47.726 [6484] DEBUG - [MPVEngine/mpv] ffmpeg: stream level seek from 354623739 to 214731063
Nov 02, 2025 14:27:47.738 [6484] DEBUG - [MPVEngine/mpv] ffmpeg: stream level seek from 214738920 to 354610967
Nov 02, 2025 14:27:47.751 [6484] DEBUG - [MPVEngine/mpv] ffmpeg: stream level seek from 354777243 to 214761156
Something is “wrong” with your file. Could you please also try to remux that file and test again? (with mkvtoolnix?)
Thanks for looking into this. I have remuxed the 8K video to .mkv using MKVToolNix (GUI), and when I played it, the ffmpeg: stream level seek entries were gone. But it seems something else is wrong with this video, because now MPV doesn’t fill the buffer correctly, and the video pauses and resumes repeatedly.
So I did the test with another 8K file which doesn’t produce the ffmpeg: stream level seek entries, but has still a significant difference in frame drops. The mpv.conf hasn’t changed. I hope this is useful. Thank you for your support!
logs & conf (2).zip (51.5 KB)
Hm, okay, something is odd with 8K video on your end. I also can not reproduce it due to lack of hardware. Nvidia GPUs, even though supporting 8K decoding, are not very good at it with mpv… (YouTube is not a problem…, a different known issue)
I just initiated a new build that should be up in about 3h… Could you please test that build? Although I don’t think there is anything in there that addresses it, you never know. (There have been over 100 commits to ffmpeg, and some to mpv in that time)
Could you also please use your mpv.conf with standalone mpv? Is behavior the same? If so, you may be better off creating an issue over at the mpv GitHub repo.
Thank you for your comments @Mitzsch. I have tested your new build (2025-11-08 14:36), and it produces about the same amount of frame drops as build 2025-10-04 11:03. FYI, I’m using an AMD Radeon RX 7600, and I found that the issue only occurs with 8K videos with 59.4 or 60 fps. Lower frame rates seem to be OK.
I will look into using standalone MPV at a later time, and will report the issue at the MPV GitHub repo if I can reproduce it. To be clear, I didn’t think that the issue had anything to do with your code changes, I was just hoping that maybe you had some ideas on how to fix it. Thanks for your support.
8K60 is very taxing… At least lower frame rates are good! ![]()
No, I would also have been surprised if my small code change had anything to do with it! (Also, in the time behavior changed, I have not made any changes…)
On the Nvidia side of things, I probably said please use high-performance GPU power mode, but as far as I know, there is no such setting on AMD’s driver control panel…
Please keep us updated! ![]()
So I tested several mpv builds, and found that my issue is related to this specific commit. For some reason, they changed the default for swapchain-depth to 2. I added the line swapchain-depth=3 (the previous default) in my mpv.conf, and the problem disappeared!
New build coming soon!
Sorry for the silence - life is taxing from time to time! ![]()