/3/ - 3DCG

>>480652
Atlasing involves combining the UV’s of multiple objects into a single texture, primarily to reduce the number of textures you have to load and/or the number of draw calls needed to draw a group of objects. Think of it this way-

You have a bedroom. In the room, you have a bed, dresser, nightstand, blinds, and a lamp. If you create individual textures for each object (and each texture has it’s own material in your game engine/modeling app), each object will require 1 draw call to paint on the screen, for a total of 5 draw calls.
Now, take the UV’s of all those objects, and lay them out into a single texture map. You now have 5 objects that share the same texture and material. Your renderer can now batch the calls to draw the objects together, reducing the number of draw calls from 5 down to 1. Because draw calls are one of the more overhead heavy things you can perform on a GPU, ideally you want as few draw calls as possible to display your scene. That is the primary advantage of altlasing, reducing texture memory (for the most part, see caveats below) and reducing draw call overhead on the GPU.

Drawbacks to atlasing primarily revolve around texel resolution. If you take a group of objects and put all their UV islands together on say a 2048x2048 texture map, you will sacrifice texel resolution compared to giving each object say a 512x512 texture map. (however, assuming your UV’s are scaled proportionately based on the actual object size in the world, you easily get consistent texel size).

It’s all a matter of choice based on your hardware restrictions, as well as just how much detail you want and wether you are willing to sacrifice some fidelity for performance.

TL;DR version- atlasing generally reduces RAM use and overhead for resolution, but has it’s drawbacks, and choosing wether to use it or not is a consideration that requires examining multiple factors in your project.