ffmpeg-headless: enable additional encoders

[eclairevoyant]

Description of changes

Should help with #271863

This adds some extremely helpful and popular encoders in by default:

• openjpeg
• celt
• libwebp
• libaom

Closure size for ffmpeg-headless went up 18.5 MiB, and even less of a difference for the "small" variant.
(I tested this on the master branch before rebasing because I didn't want to spend all year on a mass rebuild)

$ nix store diff-closures nixpkgs#ffmpeg-headless^bin .#ffmpeg-headless^bin
celt: ∅ → 0.11.3, +168.4 KiB
ffmpeg-headless: +70.0 KiB
giflib: ∅ → 5.2.2, +398.7 KiB
lcms2: ∅ → 2.16, +466.2 KiB
lerc: ∅ → 4.0.0, +840.2 KiB
libaom: ∅ → 3.9.0, +8047.8 KiB
libdeflate: ∅ → 1.20, +427.0 KiB
libtiff: ∅ → 4.6.0, +655.9 KiB
libvmaf: ∅ → 3.0.0, +2665.0 KiB
libwebp: ∅ → 1.4.0, +2559.7 KiB
openjpeg: ∅ → 2.5.2, +1525.1 KiB
zstd: ∅ → 1.5.6, +1158.0 KiB

$ nvd diff $(nix build nixpkgs#ffmpeg-headless^bin --print-out-paths --no-link) $(nix build .#ffmpeg-headless^bin --print-out-paths --no-link)
<<< /nix/store/4n60lnj3zkjpasd4c56bzhpx2m8lc1sx-ffmpeg-headless-6.1.1-bin
>>> /nix/store/884f487w5hac6rs94jq6hq5zqkxdv666-ffmpeg-headless-6.1.1-bin
Added packages:
[A.]  #01  celt        0.11.3
[A.]  #02  giflib      5.2.2
[A.]  #03  lcms2       2.16
[A.]  #04  lerc        4.0.0
[A.]  #05  libaom      3.9.0
[A.]  #06  libdeflate  1.20
[A.]  #07  libtiff     4.6.0
[A.]  #08  libvmaf     3.0.0
[A.]  #09  libwebp     1.4.0 x2
[A.]  #10  openjpeg    2.5.2
[A.]  #11  zstd        1.5.6
Closure size: 66 -> 78 (15 paths added, 3 paths removed, delta +12, disk usage +18.5MiB).

Things done

• Built on platform(s)
  • x86_64-linux
  • aarch64-linux
  • x86_64-darwin
  • aarch64-darwin
• For non-Linux: Is sandboxing enabled in nix.conf? (See Nix manual)
  • sandbox = relaxed
  • sandbox = true
• Tested, as applicable:
  • NixOS test(s) (look inside nixos/tests)
  • and/or package tests
  • or, for functions and "core" functionality, tests in lib/tests or pkgs/test
  • made sure NixOS tests are linked to the relevant packages
• Tested compilation of all packages that depend on this change using nix-shell -p nixpkgs-review --run "nixpkgs-review rev HEAD". Note: all changes have to be committed, also see nixpkgs-review usage
• Tested basic functionality of all binary files (usually in ./result/bin/)
• 24.11 Release Notes (or backporting 23.11 and 24.05 Release notes)
  • (Package updates) Added a release notes entry if the change is major or breaking
  • (Module updates) Added a release notes entry if the change is significant
  • (Module addition) Added a release notes entry if adding a new NixOS module
• Fits CONTRIBUTING.md.

---

Add a 👍 reaction to pull requests you find important.

[jopejoe1]

Maybe even worth including aom and jxl seeing #330125 and looking at the mpv manual they are listed as supported formats for screenshots https://mpv.io/manual/master/#options-screenshot-format

[eclairevoyant]

jxl is too big (100 megs)
aom i thought would be redundant as we already have svt-av1 and rav1e

[jopejoe1]

jxl is too big (100 megs)

ooff

aom i thought would be redundant as we already have svt-av1 and rav1e

mpv defaults to that one https://mpv.io/manual/master/#options-screenshot-avif-encoder

[eclairevoyant]

aom i thought would be redundant as we already have svt-av1 and rav1e

mpv defaults to that one

right but that doesn't preclude using another av1 encoder (aom is slow so the default is probably suboptimal). anyway i'll take a look at closure size later

[Atemu]

aom is slow so the default is probably suboptimal

Keep in mind that the purpose here is singular frames where iframe quality matters most and speed is almost secondary. The performance of either encoder in motion pictures has little relevance here.

[eclairevoyant]

aom adds another 10 megs, up to you if you think it's worth it
I personally am not super convinced by needing that much space for a thumbnailer but 🤷

[Atemu]

That sounds acceptable, especially considering that you likely already have libaom in your final closure via libavif.

Searching around, I gathered from discussions that aom has greater compression efficiency for avif purposes and that SVT-AV1 has odd resolution and memory limitations. I have sadly not found any data on this despite my best efforts.

[eclairevoyant]

added libaom, desc updated.

[Atemu]

Could you add the PR message content to the commit message?