Libdovi error with Vapoursynth

[Amaraldo]

The dovi_tool is a great tool. I am trying to create lossless FEL reconstructed Dolby Vision screenshots using dovi_tool and Vapoursynth but I keep getting this error "No entry point found in /opt/homebrew/lib/vapoursynth/libdovi.3.1.2.dylib" when running Vapoursynth scripts.

Is this related to the fact I'm using an m1 mac? I'd appreciate any help or workaround.

[quietvoid]

Why do you have libdovi installed in the VS plugins?
It shouldn't be there.

[Amaraldo]

Ah. My mistake. Where should it be?

Also, is it possible to reconstruct the FEL with lossless re-encoding for short spliced Dolby content? I am trying to create a database comparing the quality of standard HDR10 screenshots with their Dolby Vision counterparts. Lossy re-encoding defeats the purpose as any gains afforded by the FEL are lost in the re-encode. Thanks again!

[quietvoid]

You can simply use awsmfunc.MapDolbyVision in VS.
https://github.com/OpusGang/awsmfunc/blob/master/src/awsmfunc/base.py#L1905

[Amaraldo]

Sorry to keep bugging you but I cannot figure out what I'm doing wrong.
I keep getting this error:
"Error: Failed to retrieve frame 0 with error: placebo.Tonemap: Clip is missing DolbyVisionRPU prop for Dolby Vision mapping!"

The base_layer is a short 3 second crop of a UHD (.mkv but I tried .hevc too). Mediainfo shows profile 7 with BL+EL+RPU. I used dovi_tool to extract the EL and set that to the enhancement_layer.

[quietvoid]

Sounds like your source fillter doesn't support the DolbyVisionRPU frame prop.
You need to build the git version of ffms2 or l-smash.

Both clips need to have the RPU as well.

[Amaraldo]

Does the awsmfunc.MapDolbyVision function work differently to this script?

import vapoursynth as vs
from vapoursynth import core
from pathlib import Path

bl_path = Path("")
bl = core.ffms2.Source(bl_path)
bl = bl.resize.Spline36(format=vs.YUV420P16)

bl_w = bl.width
bl_h = bl.height

el_path = Path("")
el = core.ffms2.Source(el_path)
el = core.resize.Point(el, width=bl_w, height=bl_h, format=vs.YUV420P10)

hdr = core.vsnlq.MapNLQ(bl, el, "RPU.bin")

hdr.set_output()

The awsmfunc.MapDolbyVision returns shader errors to do with _15_tmp, _16_tmp issues but the script above works fine. If they both produce the same output, I can just use the script above.

[quietvoid]

Your script is wrong as it's missing the placebo.Tonemap filter.
It won't produce the correct output.

If you have errors in vs-placebo, open an issue there with more details.
It should work fine so maybe it has something to do with your setup/drivers.