Related to this issue of pixel density, I've often wondered why the CG industry has not architected a new kind of image file, one in which variable pixel density is automatically taken care of.

Imagine an image file format where it was possible to encode multiple regions, each with a different pixel density. Then it would be a simple matter to have one image for color across an entire figure, but allow parts of that image to be higher resolution than other parts.

For example, in your Brad UV layout, I could assign the image as a whole to be 1K pixels, but the region where the head is, I could assign an effective density that is quadruple that, the equivalent of a 4K by 4K image, but ONLY for the area that needed it.

This would so simplify the use of textures. A texture author would always be free to pick and choose where to apply higher resolution, without incurring the storage cost of that increased resolution over the whole image. Users would always just have one file to deal with.

Everything could be in that one image, not just high res heads, but high res eyes, eyelashes, etc. One color map to deal with, one bump map to deal with.

Seems obvious to me.

---

Renderosity forum reply notifications are wonky. If I read a follow-up in a thread, but I don't myself reply, then notifications no longer happen AT ALL on that thread. So if I seem to be ignoring a question, that's why. (Updated September 23, 2019)