For efficiency purposes I will have many images embedded into a single texture. I was going to make a reusable material that will allow for pulling a section from the texture with start/length parameters. I understand UE supports Paper2D with atlas groups but it seems as though it is for animations. Also, the material I need could be used for UI via/UMG or 3D gameplay. Is there an existing node in the material editor that allows one to pull segments out of a single texture or do I have to create it?
If you’re planning to use it for UMG, I would advise against making a material. You’re better off just referencing unique utexture assets for each UI state for each widget.
So you recommend that every image that shows up in UMG is a separate texture file that can be directly referenced rather than a constant material. Why is that the recommended approach? Also, does UMG support box9 textures?
Because placing them into one material and trying to setup the uvs for the atlas is going be a nightmare, and because I don’t believe the Slate Batcher will recognize them as batchable, so you’ll get no performance win (though I could be wrong on that one) Yes, UMG supports box9 Textures. We just call it a Box Brush, it’s a setting on the brush type when you’re settiing up the brushes on the widgets.
OK. Thank you for the quick response. One more question. Since the recommended approach is to forgo materials what is the most common way to have transparent parts? Simple alpha channel and from png’s, etc?
Yeah just have transparency in the texture and you’re set. Materials are still recommended if you need to pull of craziness you just can’t do with a static texture.
One more thing I need to know related to all of this. Our game will be running on iOS/Android. You suggest we separate out all of our UI elements into separate textures. I thought that textures must be in (power of 2 sizes) such as 128x128, 256x256, 512x512, etc. Most of our UI images will be very distinct sizes. I feel as though we are wasting texture space all over the place if we have to put each in its own size with emptiness surrounding most of it. That is why I was wondering if it would be more efficient to place multiple UI images within a single larger (power of 2) texture. Or am I wrong in that we can have irregular sized texture images such as 128x16, etc?? What do the developers at Epic recommend? This seems to be a very common scenario and yet I’ve not found any documentation on the matter.
For example, let’s say I want to deploy for Android using ETC1.
If I just use standalone textures with alpha, will the cook process compress them?
And, in the case of PVRTC, will the cook compress non-squared textures? if so, it will create a texture atlas in cooking stage right? I’m confused here.