Mirroring UVs vs stacking them

dzibarik

What are pros and cons of those two approaches?

I personally use mirroring because (if I understand right) stacking breaks AO baking but may be there is some disadvantage in mirroring I'm not aware of?

Damian Nachman

Stacking is usually better than mirroring.
That being said, you should do it properly, by making sure that the UV's are stacked on parts of the mesh that are not visibile.
For example, if your UV'ing a car then you'll probaly go for these places:

-On the underside of a car, given that the player can't flip the car.
-On certain interior parts of the car, given that the player can't get a good look inside.
And so forth....
Be mindfull of what asset you're working on, and how it will be integrated into the game.

In a nutshell, stacking is better than mirroning, as it leads to less uniformity - but it's always more time consuming to set up.
And in most case, if your stressing enough for UV space that you need to stack UV's, then you'll most likely be adding some mirroning as well.

Deadly Nightshade

Depends on the kind of graphics you are creating.
For mobile games running with simple shaders (maybe not even using normal maps), cutting down on texture size is one of the top prioritites so stacking shells becomes very important. Stacking, straightening, stretching and big varieties in texel density - you can get away with a lot more because things such as texture stretching/pinching might not become as obvious on the final asset.

But for say, current gen, high-detail assets such as FPS-weapons or the like, stacking shells can actually be to your disadvantage when doing the texturing later as you will have a harder battle vs texture seams. Stacking also tends to give you troublems during baking of AO maps on non-exploded meshes. So think long and hard about what shells can be stacked or not.