Hi folks, I’m trying to unwrap the mesh of a stone pillar model I made for use in buildings, but it’s just refusing to unwrap without overlapping itself. The broblem seems to be in the ends of the pillar, so I made a copy of the file and stripped out all but one of the problem sections of the mesh, so that I can upload it here weird-unwrap.blend (453 KB), and there’s a screenshot of it here.
As you can see, It’s basically a shallow cuboid with a circle cut out of the top surface. I marked the seams with the intention that it’d unfold into a wide strip with one small side of the cirle being at one end, and the other part (being shaped a bit like a pitchfork with the middle point missing) at the other end of the unwraped shape. No matter what I’ve tried, I just can’t get the two faces to simply flatten out at either end of the UV map without deforming and overlapping other faces. After the first failed unwrap, I put a tick in the “Correct Aspect” box in the 3D viewport and re-unwrapped it, but it didn’t make any difference; it kept on stretching and overlapping the UVs. I even tried making the UV image map wider and then unwrapped it again, but it just stretched out the UVs in that direction (as shown in the screenshot).
I know the UVs are supposed to scale to use the maximum amount of pixels available whilst still satisying their constraints, but I assume I’m missing some other, fairly fundamental, piece of knowledge, because I’m fairly sure unfolding such a simple shape so that it has a 1:1 ratio of surface area to pixels should be a trivial task.
Daedalus, yeah. The most immediate reason is that the circular hole represents geometry that would be there, but I’ve just removed it for the sake of simplifying the example problem; the more general reason, however, is that I’d like to learn about concepts, rather than performing specific tasks. Richard’s comment about ngons is a good example; I don’t yet understand exactly why they would be more problematic than quad/tri-anglular faces, but it gives me a bit of information to help me work through this problem, gives a bit of an explanation as to why some other attempts at unwrapping went so badly and gives me a name to look out for while I’m learning about Blender or maths in general. Like if I have a list of tutorials in front of me and I see ngons in the title, I could home in on that one in the hopes of finding some more, generally useful, information.
I think I’ve realized why changing the image dimensions on one axis causes stretching/squashing of faces along that axis. My assumption is that it’s because the UV image area is an abstract 1 x 1 space in which X and Y pixel exes are merely the pixel density for a given axis, rather than its actual width and height. The UVs are then unwrapped with respect to the 1 x 1 area which may happen to contain a greater number of pixels on one axis than the other. Is that right? It seems to fit what I’ve been seeing. I’ve been slowly eding forward with getting to grips with the basics (didn’t help that I kept forgetting to select things before trying to change them: like not selecting all before using Ctrl + A and other dumb stuff.
One thing that does baffle me, though, is that when I use Ctrl + P to make the UV island fit the bounds of the image, it just turns into the jumbled mess like in my original post, rather than scaling it down slightly and placing it on the middle of the image.