@yukinoh1989 You can think of baking sort of like data compression. You take a whole bunch of information from your scene(which requires a lot of computational power), and cache only the bits that are most important for you. If you have ever used a site like textures.com, you can think of baking as a way of generating custom maps, often textures, though you can bake all kinds of things: lighting, simulations, animations, and surface information. Of course once the info is compressed, it isn’t as robust or flexible as before, so you have to determine where it fits best in your workflow. That will vary from project to project, though there are some trends for different fields. Games and real time simulations will bake almost all texture information, usually at least some lighting information, and particle effects. For still renders and film, I don’t know that anyone bakes their lights, but textures frequently get baked to cut geometric computation or to preserve details on animation friendly models. For characters specifically, baking maps like normals, ao, and curvature can go a long way for workflow efficiency even when computational power may not be much of a factor when all is said and done. Ultimately, it’s about selectively storing information to improve efficiency, and managing the trade-off between acceptable input and acceptable output.
