And with some distortion you simply have to live (for that abf is usually the better solution).Īnd while I wrote, Zoohead did the same and was a bit faster. You place them at the hair line, under the jaw, behind the ears. If you create a human head, you usually don't want any seams in the face. How that is done is explained in some forum posts). Most of the time it's necessary (and with the pins in Cheetah it's kind of hard. That's why you usually have additional tools in each program to further work on your uv map.
And then you have to straighten out some parts, because whatever you do, most of the time some parts you really can't use distortion will be distorted. The difficulty is to unwrap the thing in your mind and then place the seams accordingly, already knowing what to do beforehand. Usually it's the best strategy to place your seems somewhere they can't be seen and to think in advance about where that exactly is. With that I could create scratches over all 12 edges in photoshop with two images.Īnd it shows why it's near impossible to create a good tutorial for placing the seams: It depends on the mesh and the situation, i. It also shows why I cheat the hell out of it sometimes and simply use 2 uv maps. You paint, and on the flat pic you get at the end to load into your 3d app everything is distorted the right way if necessary. And distortion isn't a problem, because the painter solves that for you. No matter where your seams are, no matter how many uv islands you have, you just paint over it (which you could do in Cheetah although on a very rudimentary level). This already shows why something like bodypaint in C4d (I think it was available as a separate app) was a big hit 15 years ago and substance painter and other apps who allow you to paint on the model are today. Is the thing animated or not? Can all edges be seen or just some? And then you have to place your edges accordingly. Except you want scratches on the edges, where you have to ask yourself how the die is seen in your scene. You could even have several uv islands (up to 6), and it wouldn't matter. įor a simple die it usually doesn't matter. Please note that in any of the three examples any rectangle could be the top of your box and alone with this we have here a few dozens possibilities to which side of your box what is mapped.Īnd now please think about it, which is the best solution, especially as there are several solutions I didn't show. Now I made 3 uv textures with seams obiously along the edges. Or you could just create 6 rectangles somehow distributed over the uv space. The given standard in Cheetah is all six sides on top of each other (meaning you get the same pic on each side). Take a cube for example, six sides you have to uv map. Where to put the seams? And why there isn't a good tutorial for it. In the end you have to test both algorithms and look what gives you the least distortion. So take what you read in some tutorials with a grain of salt. With hard edge models I sometimes don't get a straight result neither with the one nor the other (for that I have to use another tool atop in such cases).
Because in my experience "conformal" clearly gives the better results than angle based for anorganic geo and at least very often also the better results for organic meshes. To make things more complicated: In Modo conformal is ABF++ better for organic shapes angle based LSCM, better for non-organic shapes, at least according to the documentation, which, I'm not entirely sure, may be wrong in the documentation. While LSCM is "conformal" which would be better suited for non-organic models, the more geometric forms often called "hard edge". Usually (for example in the world of Blender) it's said that "angle based" is better suited for organic models (for example a human being or a pig), actually a better form of LSCM that takes a bit longer because it first does a preprocess step. (In Modo ABF++ stands for "Asynchronous Forward Bounding", which I also had to look up). ABF means angle based flattening, LSCM "Least Squares Conforming Map" (which I had to look up.
0 kommentar(er)
