LSCM mapping gives asymmetrical results

[Noob]

I usually use LSCM mapping a lot for character faces, but I'm having a bit of trouble with AC3D's implementation. I can't seem to get it to give a symmetrical result even on relatively simple shapes. To illustrate this, here is what the latest version of Blender does, and here is what AC3D does.

Would it be possible for AC3D to at least give symmetrical results?

[Andy]

Most LSCM unwrappings will need adjustment but what you see here might be something to do with the vertices that are chosen as anchor points at the start of the LSCM algorithm. AC3D picks a couple at random, perhaps Blender chooses them differently?

[Mossie]

Could it be vertix order tht drives the selection? <thinking out loud>

[Andy]

Not sure if the order of selection would change things. The algorithm starts by 'pinning' down two vertices and unwrapping between these. If these two vertices are different to the vertices that Blender picks, then it will look different.

[Noob]

Originally Posted by Andy

Not sure if the order of selection would change things. The algorithm starts by 'pinning' down two vertices and unwrapping between these. If these two vertices are different to the vertices that Blender picks, then it will look different.

Ahh, so that's how it works. Thanks for the explanation! Would it be possible for us to someday choose the vertices it uses?

[jentron]

Originally Posted by Andy

Not sure if the order of selection would change things. The algorithm starts by 'pinning' down two vertices and unwrapping between these. If these two vertices are different to the vertices that Blender picks, then it will look different.

Would it make sense to choose the second vertex as the mirror of the first? So vertex[1]={x,y,z} and vertex[2]={-x,y,z}? Or as near to {-x,y,z} as possible? I've been playing with the evaluation version a little, and I keep coming out with skewed results that need manual intervention to derotate and unskew even though the shapes are extrusions along an axis.