New nvidia-true-hdr command causes frame drops with SVP. Old command does not

[netExtra]

mpv Information

mpv v0.41.0-233-gf28cea85c Copyright © 2000-2026 mpv/MPlayer/mplayer2 projects
built on Feb 28 2026 00:21:47
libplacebo version: v7.360.0 (v7.360.0-4-gf7c2db9-dirty)
FFmpeg version: N-123068-gaa483bc42
FFmpeg library versions:
libavcodec 62.23.103
libavdevice 62.2.100
libavfilter 11.12.100
libavformat 62.10.101
libavutil 60.25.100
libswresample 6.2.100
libswscale 9.4.100

Other Information

- Windows version: Windows 11 25H2
- GPU model, driver and version: Nvida RTX-5090 
- Source of mpv: Shinchiro mpv-x86_64-v3-20260228-git-f28cea8
- Latest known working version:
- Issue started after the following happened: after new nvidia-true-hdr command added.

Reproduction Steps

Base config:
input-ipc-server=mpvpipe
hwdec=d3d11va-copy
d3d11va-zero-copy

log-file="C:\Users\Admin\AppData\Roaming\mpv\mpv-nvidia-true-hdr.log"

pause=yes
geometry=60%
ontop-level=5

vo=gpu-next
profile=high-quality
target-colorspace-hint=auto # auto is for HDR passthrough

Expected Behavior

With this command: vf-add="d3d11vpp=nvidia-true-hdr,format=max-luma=130" I wasn't expecting frame drops with SVP.

Actual Behavior

With the new nvidia-true-hdr command: vf-add="d3d11vpp=nvidia-true-hdr,format=max-luma=130" Frames start to drops after about 20s with SVP.

With the old nvidia-true-hdr command: vf-add="d3d11vpp=format=x2bgr10:nvidia-true-hdr,format=x2bgr10:transfer=pq:primaries=bt.2020" there are no frame drops with SVP.

I made sure to use a recent Schinchiro version: "Shinchiro mpv-x86_64-v3-20260228-git-f28cea8" and enabled the RTX Video indicator to make sure that RTX HDR video was working in both cases. Both log files are attached.

Log File

mpv-new-nvidia-true-hdr command.log

mpv-old nvidia-true-hdr command.log

Sample Files

No response

I carefully read all instruction and confirm that I did the following:

• I tested and confirmed that the issue exists with the latest release version or newer.
• I provided all required information including system and mpv version.
• I produced the log file with the exact same set of files, parameters, and conditions used in "Reproduction Steps", with the addition of --log-file=output.txt.
• I produced the log file while the behaviors described in "Actual Behavior" were actively observed.
• I attached the full, untruncated log file.
• I attached the backtrace in the case of a crash.