Questions about uvmaps, texturing, compression and optimization

Hello, can someone explain how engine works in situations which I will try describe below.
When I’m making uv maps and texturing I always wondered how much memory (ofcourse video memory) takes in situations when my meshes have different uv maps?

I’m interested in such variants:
1. I have cube with uv map where every face placed in different shells
2. I have 3 object which shares the same texture but placed in a level with different counts (for example 5 cubes 2 toruses and 1 pyramid)

Is it better to place faces above each other or I can freely place faces into a different shells? How to make cheaper for computer

About compression, in my pc i have normal map with the size 2112 kb, but when I place it to the project it has size ~5000 kb. is there a way to avoid such compression or it is normal situation?

And my last question is about resolution of textures. Can i use non-standart resolutions like 768x768 or 237x237?

  1. Just 13mb
  2. Unwrap just stores additional coordinates for your mesh, not the texture. Memory increase is negligible and is applied to mesh, not texture.
  3. Typically, less UV seams-better
  4. You might be importing a texture from a lossy format. Try importing as TGA or something like that.
  5. You can but should not for a number of reasons.

File size of an image is more related to there resolution of the image rather than what size the original file is, the engine compression settings determine how much it’s compressed.

UV mapping itself is not going to affect memory consumption much, if you add more UV channels then that will increase the amount of memory used since it has to store more data for the model—which means, don’t use additional UV channels if you don’t need them.