I am not a designer but I often read that it is really important to have a very nice uv layout and I am not sure why.
I know it is important if you want to use textures which are tiling because otherwise you can run some issues with seems but do we even use tiled textures for our current gen games? (I know that there are some cases where you want tiling textures , for example landscapes)
It seems that many artist just go into Zbrush, generate a uniform UV-layout and then paint onto the model, they may even project textures onto the model and it shouldn’t matter how your UV’s are layed out and where you have placed your seems.
What other reason are there that justify a nice manually unwrapped model?
For lightmaps it matters a lot, because you want to reduce the number of seams and optimize how much of the space that the UV’s cover.
In other cases, there’s similar things to look out for–optimize how much of the space you are using and reduce the amount of seams. In other cases you may want to avoid stretching so that pixels will be the same size across the mesh.
In other cases you want to avoid bleeding over to your UV’s. Games use LOD for textures so it will go to a lower resolution when the object is further away, that can mean that black background pixels can become bigger and bleed onto the UV’s and then the mesh looks bad.
Well in my opinion we are a long ways off of what I would consider a next gen game if the ideals is based on technological improvements. Most of the stuff you see now and think that’s new has been around for a while or even years and more or less someone getting around to taking advantage of an old feature never used.
How important is Uvmapping?
All things games development wise is based on context. With out context than all things are equally important with out bias but if an object “needs” the support of texture then it becomes very important if the object in question is a center piece of your design.
Even then.
As physics based and procedural materials become more popular and common the there will be no need for uvmapping at all, except for lightmapping and I even see that going the way of old ideals, and texture work will someday depend on how many crayons you have in the box.
For now though how something needs to be mapped should be left up to the texture artist and responsible for setting up the mapping the way they want and should be a skill they know how to do on their own.
Huh.
You should learn a little bit more about computer graphics, it’s not only a “design” thing.
Vertices at UV shell borders are doubled if they are exists in multiple UV shells. Like each UV shell got own copy of vertex. If you have too many splits - it will end up in more power hungry object.
Here is the picture:
Same mesh, different uv = different final vertices count.
Furthermore, when you’re baking a normal map, you’re using vertices bitangent and tangent, which are calculated from UV coordinates and if you have bad UV layout - you will have bad normal map.
Thanks for this picture. This is how I tried it in Blender. I assume the one on the left is the way to go. So when I do it like this on a simple house it “peels of” in 2 nice islands. However in UE4 I will have black(ish) edges. When I make all the faces as loose islands it is less.
You can see it in this image; the house on the right hase several islands the left one has 2 nice islands (1 for the roof and 1 for the base)
The left top vertex and the right top vertex are actually the same vertex right? But they have different positions on the image texture so they could have two different color information.
How do graphic engines usually handle this? I mean if you are in the shader and ask for the UV coordiante, it would be crazy if you get back two UV coordinates instead of one. Is one position just overwritten?
This would also mean that every seem will cost you precious texture space right?(At least if they are not stacked on top of each other)
