TSC Meeting Notes 2023-09-07

Attendance:

  • Cary Phillips
  • Christina Tempelaar-Lietz
  • John Mertic
  • Joseph Goldstone
  • Kimball Thurston
  • Larry Gritz
  • Nick Porcino
  • Peter Hillman
  • Rod Bogart

Discussion:

  • Chromaticity / color space tags - need for a cross project discussion
    • Larry: regarding meeting with Doug from OpenColorIO, color space canonical naming came up again, time to have a OpenColorIO, OpenImageIO, OpenEXR summit to determine a strategy for chromaticity.
    • Kimball: talked to Foundry at Siggraph regarding updating Nuke OCIO plugins. Weta has a known metadata tag for color space internally. 
    • Larry: problem on OCIO side, names are specific to profile. Recently from 2.2 forward they have internal configs. Then can use interchange spaces to translate to built-in configs. Wants OCIO to submit canonical names for color spaces everyone uses. 
    • Kimball: discuss whether to discard primaries. (added note: primaries = chromaticity coordinates.) 
    • Larry: still some complication regarding what heuristics to follow for how to handle multiple color spaces. Also what does the metadata mean - is it archival for source camera or are they expected to be modified.
    • Kimball: should include OpenTimelineIO. Could involve the entire TAC.
    • Cary: OpenEXR is oddly positioned to be the driver of this initiative since this issue is a bit higher level.
  • C2PA
    • Larry: need to pick a name for the metadata. Stage 1 apps can embed it. Needs to cryptographically hash a chunk in the file for a defined boundary of the file.
    • Kimball: there is a hook point for that since library already does some work in a second pass when writing a file.
    • Larry: people are asking for OpenEXR support. They want to be able to include in their spec details for how it should be done.
    • Kimball: would like to not add cryptographic requirements on the library. However having an example code for enabling it, similar to how we allow replacing the malloc library, could work. 
    • Cary: any benefit to doing the signing on the channel data rather than the file? Hashing on the compressed or uncompressed values.
    • Kimball: compressed so you can validate that it is unchanged. Also a thumbnail. Might be a valid use of blockchain - e.g. image could encode Arri took the picture, FX company added effects to it, etc...
    • Kimball: core library has the encode and decode function so you can register your own metadata types but doesn't help because there no hook to save it at the end.
    • Larry: needs a couple pieces.
  • Cary: How to link OpenEXR again existing external libDeflate? 
    • Kimball: have a script at home to run through the permutations. There are 2 config options.
    • Cary: need to specify a particular location. 
  • Dev days issues
    • Cary: only non-C++ ideas augmenting Imath tests, need more ideas