Okay, this is going to be weird, hear me out.

Purpose:

Reduce texture load times, vram use, and therefore lag.

How:

For every texture make there exist and/or be created on first demand a uuid-xxxx alternate version of the texture where the texture is scaled down and saved to the asset servers. The value is determined by the width of the image.

So I might have a uuid, uuid-512, uuid-256.

Execution:

In the build interface, users would have a list of textures on a face and a drop-down menu letting them select the size of each texture on that face.

Why:

Not every texture in the materials on a face needs to be maximum size.

If you have a mesh item with both Blinn-Phong and PBR, a user might have to download up to 6 (maybe 7 if they're using different normals) textures per face for up to 8 faces per object component.

That's 56 possible textures, and you could potentially take 40 or more of them (depending on if the face is small and you can get away with a lower resolution texture) and reduce them in size to take 1/4 as much resources.

So you still have up to 56 textures on an 8 face object but at the memory cost of 26. Currently this type of optimization can only happen if you are the creator, you bought full perm texture assets and export scale down import, or you do crime.

How would creators access these auto scaled textures? Tack the size number at the end, or open the full perm texture and select, "Get scaled down copies."

Anyhow, thanks for reading.