I’ve been learning more about ACES and I like the fact that it uses a standard 16-bit half float linear format but it got me wondering, would there be an advantage to be able to specify an offset for 18% grey?
The reason I’m wondering this is because floats naturally form a sort of slog curve that’s centered around zero. Values around zero are able to account for very small changes but as values get further and further from zero they become less precise. Usually we would want slog formats centered around where we want the most precision, so maybe around 18% grey for example, and then very large and very small values can be less precise.
I’m by no means an expert but if you could provide a grey value I’d imagine that it could be used to offset the values before writing them and then offset them back before using them as linear, maybe as optional metadata?