Troy James Sobotka

avcodec_decode_video2 and got_picture_ptr

5

Regarding frameFinished in https://github.com/mpenkov/ffmpeg-tutorial/blob/master/tutorial01.c#L150 It is worth noting here that frameFinished might result in missed frames. Consider the following frame format: ``` F1 F2 F3 F4 II BB BB PP...

input_scale scales alpha currently. This can lead to odd behaviour in many applications. Should there be an option to scale RGB and not alpha if desired?

CCT Conversions

2

If you are keen to flesh this out further, looking into more correct and flexible transforms from CCT to various RGB encodings. See https://www.colour-science.org/api/0.3.1/html/colour.temperature.cct.html https://github.com/jlampel/bl_extra_lights/blob/0e3375e73e5431d09820e4dff0b94f34d80a97a5/conversions.py#L10

Getting to CIE based Colorimetry is Overly Complicated

2

Can we please, please, please provide utilities to go directly to CIE based colourimetry? None of the nonsense present at the SIGGRAPH 2022 discussion is valid to support routing through...

INTERPOLATION: "Best" defaults to linear?

2

https://github.com/imageworks/OpenColorIO/blob/67a4f21db960bc1ce2ee040ff3264d3a9262af53/src/core/Lut3DOp.cpp#L674-L676 https://github.com/imageworks/OpenColorIO/blob/67a4f21db960bc1ce2ee040ff3264d3a9262af53/src/core/Lut1DOp.cpp#L555-L557 This appears by design, however should it not default to tetrahedral for the 3D LUT sampling?

Currently, correcting for colour critical work via a 3D LUT is possible with OCIO, but messy organizationally. If wrapped up into a View transform, the parcel makes it even more...

ociolutimage --colorconvert doesn't work with --extract

2

Summary says it all. ociolutimage doesn't appear to transform the input RGB image when using colorconvert with extract.

[OCIO] Example of breakage

4

https://github.com/cgvirus/Olive-Editor-Community-Effects/blob/94b4b1a97a8d0c5f528764c1bce3060f2ab5c264/bloom/bloom.frag#L42

Parsing error on whitespace

3

`direction:inverse` fails while `direction: inverse` succeeds silently. @Shootfast reports that ociocheck also fails to note the problem.

Given that V4 makes it impossible to deduce the implicit white point of a characterization, how is OCIO V2 dealing with this rather impossibly large obstacle? As per the more...