This has two somewhat related facets:
- When passing the output of the renderer to the output system, sometimes it's desirable to pass extra non-color data. The most common thing is an alpha channel, but there can be more, e.g. distance information to the first object intersection or even more detailed data that can be useful for doing e.g. noise-reduction.
- When reading image files, similarly we sometimes may want to read extra data (e.g. an alpha channel), or maybe change the way the basic color data is handled entirely (e.g. for reading normal data).
Currently we have a "Tint" type that adds alpha data to color, which works ok for that specific cases, but can be awkward and limited. Maybe some sort of general channel abstraction would be good...
This has two somewhat related facets:
Currently we have a "Tint" type that adds alpha data to color, which works ok for that specific cases, but can be awkward and limited. Maybe some sort of general channel abstraction would be good...