I have a Canon Log 3, S-Log3 or V-Log clip and want to know which constants in an open transform I can check against the manufacturer's document, and which I must take on trust?

Curve constants can be checked against vendor documents for Sony, Panasonic and Canon (CA-B219) as well as ARRI and RED. Primaries cannot always: the Sony summary and Canon's 2014 paper give no numeric primaries in their text (CA-B219, CA-B566), while OCIO 2.6 and ACES carry S-Gamut3, S-Gamut3.Cine and Cinema Gamut values without citing a source; V-Gamut primaries in OCIO equal the Panasonic manual. Curator extension: even with matching primaries, the vendor AP0 matrix depends on the adaptation method. Panasonic's published V-Gamut->AP0 matrix is closest to Bradford (about 2e-4) but not reproduced exactly; OCIO and ACES both use Bradford for V-Gamut. Curator qualifications: Documents and code read 2026-10-22; matrices recomputed by the curator in Python from published primaries (arithmetic only); no transform executed, no footage decoded. Sony raster primaries table not read (existing hold, CA-B566). Panasonic matrix residual unexplained. Curator amendment CA-AM-0065 (2026-10-22): Registry and pin clarified (curator verified on the cycle-2 producer copy C2-B05-B/work/aces_transforms.json, sha256 4c6b0f208fd9d42f equal to the pin). In the v2.0.0+2025.04.04 block, all 48 aces-input-and-colorspaces transformUrl values point at commit cc60becf, which explains the producer citation; the umbrella submodule gitlink 2383d375 is a different commit, so cite both. Entries [18] ARRI (ID LogC4, file CSC.Arri.LogCv4_to_ACES.ctl), [19] BMD (userName 'to ACES2605-1', the same typo as the CTL), [31] Panasonic, [33] RED. Panasonic V-Gamut->AP0: curator recomputation agrees with discovery that no named CAT reproduces the published matrix (Bradford nearest, 2.34e-4); its row sums are 0.999628, 1.000173, 0.999996, so it is not exactly white-preserving; a Bradford adaptation to a destination white (0.32160, 0.33771) reproduces it within 3.37e-6. That white is an inference only; no Panasonic text states it. Limitations: Documents and code read 2026-10-22; matrices recomputed by the curator in Python from published primaries (arithmetic only); no transform executed, no footage decoded. Evidence status: statements located in the cited sources by a correlated desk review (Claude); not executed; not independently reviewed.

{
  "citations": [
    {
      "url": "https://github.com/AcademySoftwareFoundation/OpenColorIO/tree/10a43ec1df03d072ca9a9e2fb3f26e454b0900d8/src/OpenColorIO/transforms/builtins",
      "note": "OCIO builtins SonyCameras.cpp @10a43ec1 lines 22-37, 73-108; OCIO builtins CanonCameras.cpp @10a43ec1 lines 28-33; OCIO builtins PanasonicCameras.cpp @10a43ec1 lines 23-36, 67; ACES CSC.Panasonic.VLog_VGamut_to_ACES.ctl @29b722bc lines 31-32; Panasonic V-Log/V-Gamut Reference Manual Rev.1.0, pdftotext -layout lines 160-172; curator arithmetic (not source code execution)"
    },
    {
      "url": "https://github.com/aces-aswf/aces-input-and-colorspaces/tree/29b722bccd529460696a8382394504cae2e88419",
      "note": "OCIO builtins SonyCameras.cpp @10a43ec1 lines 22-37, 73-108; OCIO builtins CanonCameras.cpp @10a43ec1 lines 28-33; OCIO builtins PanasonicCameras.cpp @10a43ec1 lines 23-36, 67; ACES CSC.Panasonic.VLog_VGamut_to_ACES.ctl @29b722bc lines 31-32; Panasonic V-Log/V-Gamut Reference Manual Rev.1.0, pdftotext -layout lines 160-172; curator arithmetic (not source code execution)"
    }
  ]
}