Which is the best way to unwrap UV's for texture creation?
(I'm currently getting started with modeling & texture creation, and I don't understand the difference between LSCM and ABF)
Which is the best way to unwrap UV's for texture creation?
(I'm currently getting started with modeling & texture creation, and I don't understand the difference between LSCM and ABF)
They're both just algorithms to take a 2d* mesh that is folded in 3d and "spread it out" — i.e. map it — onto a 2d surface with different constraints, but both with the same objective — to create the least distortion of the mapped texture (e.g. imagine if you are using a fabric texture — you don't want the patterns to stretch and distort any more than necessary).
From looking at the abstract of the ABF paper (and contrary to what the Blender docs suggest), ABF's chief virtues is not quality, but better performance with large meshes without loss of quality. Which one is "better" will vary from model to model since both methods are using quite sensible methods to reduce distortion.
*2d as in 2d _manifold_, not 2d as in planar. E.g. a piece of paper approximates a 2d manifold, whether it's flat on a table, or scrunched into a ball.
ABF is an improved version of LSCM. Before the LSCM flattening is performed a preprocess is run which minimises the angular distortion. So ABF is in general the better algorithm. But due to the preprocess it takes approx. twice as long as the plain LSCM.
There is actually a whole family of ABF algorithms these days. Here you can find the paper of the LinABF algorithm which Cheetah3D uses.
https://hal.inria.fr/inria-00186848/PDF/linabf.pdf
I have a question. Sorry I know you are busy.
Why does the shape change in the UV window on some objects when unwrapped between the two algorithms not all objects but some.
Because these are different algorithms. One just flattens the mesh to 2D. The other also tries to keep the angular distortion as small as possible.
Would be sort of useless to have two algorithms if they would generate the same output.:wink:
ABF is an improved version of LSCM. Before the LSCM flattening is performed a preprocess is run which minimises the angular distortion. So ABF is in general the better algorithm. But due to the preprocess it takes approx. twice as long as the plain LSCM.
I'm embarrassed to admit that paper is a bit above my mathematical understanding. So I was looking for a simple answer like: ABF is the one to use to keep angular distortion as small as possible?It's also explained in the paper I've attached in my former answer.