build(ffmpeg): update to FFmpeg 7.1

[cgutman]

Description

This PR rebases from FFmpeg 6.1 to FFmpeg 7.1.

Notable changes:

• A replacement for 01-amf-colorspace-v3.patch was upstreamed in FFmpeg 7.1, but it requires a fixup to pass-through the proper colorspace into the bitstream for HEVC and AV1. I replaced it with the patch I sent to ffmpeg-devel, but upstream will probably take a better fix that handles both internal RGB->YUV conversion and pre-converted YUV input
• A replacement for 02-idr-on-amf.patch was upstreamed but missed the 7.1 cut, so this patch is replaced with my upstreamed version. This requires changes on the Sunshine side to set the forced_idr option for AMF like we do for QSV and NVENC.
• A replacement for 05-amfenc-new-av1-usages.patch was upstreamed for FFmpeg 7.1, so that patch is dropped completely
• Some minor changes were required to the CBS stuff to adapt to the new VP8 CBS support in FFmpeg 7.1 and the relocation of HEVC-related headers
• 05-vaapi-customized-surface-alignment.patch needed to be reworked due to upstream refactoring of common VAAPI/D3D12VA/Vulkan encoding logic into a single base class

Testing checklist:

• Software
• QSV
• NVENC (Linux)
• AMF
• VideoToolbox
• VAAPI (Intel)
• VAAPI (AMD + alignment fix)

Screenshot

Issues Fixed or Closed

• Closes fix: patch Vulkan synchronization #357

Type of Change

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• Dependency update (updates to dependencies)
• Documentation update (changes to documentation)
• Repository update (changes to repository files, e.g. .github/...)

Checklist

• My code follows the style guidelines of this project
• I have performed a self-review of my own code
• I have commented my code, particularly in hard-to-understand areas
• I have added or updated the in code docstring/documentation-blocks for new or existing methods/components

[cgutman]

There seems to be an extra frame of latency introduced in NVENC that causes a "decoder did not produce IDR frame" warning when the client requests an IDR frame. I don't see anything obvious in upstream changes that would explain it, so I'll probably have to bisect.

[cgutman]

OK, I'm not sure why the behavior changed in FFmpeg, but it looks like setting surfaces=1 for NVENC solves it.

This is ready for merge. I'll open up the Sunshine PR after merge to update the submodule and add the new options for NVENC and AMF.

[DistantThunder]

I don't know if it's of any incidence but FFMPEG release/7.1 branch isn't at the same commit as actual tag release

• https://github.com/FFmpeg/FFmpeg/releases/tag/n7.1

[ReenigneArcher]

@DistantThunder we are following the release/7.1 branch: https://github.com/FFmpeg/FFmpeg/commits/release/7.1/

[moi952]

Hi,
What does FFmpeg 7.1 bring to the use of sunshine?
I saw that vulkan had been implemented but not for encoding is that right?
Thanks and sorry for the inconvenience

[ReenigneArcher]

The notes in this PR should help answer that: #359