Failure to display a HDR PNG with gpu-next or gpu

[linkmauve]

mpv Information

mpv v0.41.0-459-gda4789c2d Copyright © 2000-2026 mpv/MPlayer/mplayer2 projects
 built on Apr 13 2026 00:40:45
libplacebo version: v7.360.1
FFmpeg version: n8.1
FFmpeg library versions:
   libavcodec      62.28.100
   libavdevice     62.3.100
   libavfilter     11.14.100
   libavformat     62.12.100
   libavutil       60.26.100
   libswresample   6.3.100
   libswscale      9.5.100

Other Information

- Linux version: Arch Linux
- Kernel Version: Linux x280 6.19.11-arch1-1 #1 SMP PREEMPT_DYNAMIC Thu, 02 Apr 2026 23:33:01 +0000 x86_64 GNU/Linux
- GPU Model: 00:02.0 VGA compatible controller [0300]: Intel Corporation Kaby Lake-R GT2 [UHD Graphics 620] [8086:5917] (rev 07)
- Mesa/GPU Driver Version: Mesa 26.1.0-devel (git-276aedc859)
- Window Manager and Version: Niri 25.11.r44 (git-b5640d5)
- Source of mpv: self-built from latest master
- Latest known working version: N/A
- Issue started after the following happened: I downloaded that PNG
![mpv-png.png](https://linkmauve.fr/files/mpv-png.png)

Reproduction Steps

1. Download the sample file.
2. Run mpv --vo=gpu-next HDRFrame_2026-04-12_22-23-38-209.png
3. --vo=gpu also reproduces, but with a different fallback colour.

Expected Behavior

The image would display, just like in Firefox or in --vo=dmabuf-wayland.

Actual Behavior

Instead mpv stays fully magenta (resp. blue with --vo=gpu), and only reacts to keyboard input.

Log File

With --vo=gpu-next: mpv-sucks.txt
With --vo=gpu: mpv-sucks.txt

Sample Files

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.

[N-R-K]

Works fine for me. Try with vulkan disabled (--gpu-api=opengl).

[llyyr]

This only reproduces on anv for me.

[vo/gpu-next/libplacebo] Rejecting texture format 'rgb16' due to misalignment: Row stride 12224 is not a clean multiple of texel size 6! This is likely an API usage bug.

For radv and llvmpipe, rgb16 has texel size 2...

[kasper93]

@linkmauve:

Does this fix the issue for you?

diff --git a/video/mp_image.c b/video/mp_image.c
index 9ba69e6305..dc73b75e5c 100644
--- a/video/mp_image.c
+++ b/video/mp_image.c
@@ -68,7 +68,8 @@ static int mp_image_layout(int imgfmt, int w, int h, int stride_align,
         int alloc_w = mp_chroma_div_up(w, desc.xs[n]);
         int alloc_h = MP_ALIGN_UP(h, 32) >> desc.ys[n];
         int line_bytes = (alloc_w * desc.bpp[n] + 7) / 8;
-        out_stride[n] = MP_ALIGN_NPOT(line_bytes, stride_align);
+        int align = mp_lcm(stride_align, (desc.bpp[n] && desc.bpp[n] % 8 == 0) ? desc.bpp[n] / 8 : 1);
+        out_stride[n] = MP_ALIGN_NPOT(line_bytes, align);
         out_plane_size[n] = out_stride[n] * alloc_h;
     }
     if (desc.flags & MP_IMGFLAG_PAL)

[llyyr]

That resolves the issue for me

[linkmauve]

This fixes it for me too, thanks! And now I see that --vo=dmabuf-wayland has completely wrong colours on Niri, same but different as Firefox.

[llyyr]

Could you open a PR with that diff? so it isn't forgotten