I see, the way it was written in the cube was that it was defined as a 1D but the table still held 3x each value. Is that neccesary then for even a 1D lut inside a .cube?
Those files can be used in other software that may only rely on loading the LUT files like compositing software. At least until OCIO/ACES config will have Apple Log included as a built-in transform.
It’s what is sometimes called a 3x1D LUT. The three columns are R, G and B, and could contain three different curves. In this case they all contain the same values on every row. But the format requires three columns, even if they are identical.
For example yes. Or in a custom OCIO configuration. The original .cube that was in their package would also work. And you could also use a matrix node for the primaries conversion rather than a file. But maybe some will find a set of spi1d+spimtx nice to work with which is why I put 'em up. Works nicely in After Effects from what I tested.
The 1D LUT converts the log encoding to linear. Depending on your workflow that may be the only thing you need. But bringing mixed material into a single working color space, meaning the primaries too, will keep the colors consistent.
The spimtx is a conversion matrix from gamut Rec.2020 to AP0 which with linear gamma results in the color space ACES 2065-1. For ACES OCIO configs this is what it needs to convert to and from. If you aren’t implementing it into an OCIO config it is probably better to directly convert to the working space ACEScg to save an extra step, which would mean AP1 primaries instead. You could make an extra file for that or put in values directly into Nuke in a color matrix node.
This is a handy tool to get such conversions. Setting the formatter to Nuke gives you the script that you can copy pase directly into Nuke’s node setup
It’s important to first convert to linear and then apply the matrix. If you need to deliver as roundtrip back in AppleLog you’d do the opposite. Matrix first, then the transfer function.