I'm having problems with Apply UV Curves command. The dimensions of the wrapped geometry change. Attached is the simple project I've used to find out the problem. 

Views: 1056

Attachments:

Replies to This Discussion

Yes, this definitely looks buggy to me.  It does the same in V4, I'm surprised it wasn't noticed earlier.  The problem comes from the non-rational aspect of a cylinder surface, the parametrisation seems to be messing up the mapping of the curves to the surface.  If you Rebuild the cylinder with, say, 24 points and degree 3 - which will make it a non-rational surface with uniform parametrization - and then apply the square, it will measure a perfect 100 units curve length on the 4 sides.

----H

Attachments:

Hi Bragagnolo-  I guess the command is doing what it is designed to do, but the parameterization on a cylinder will accelerate and decelerate- you can see this if you apply a repeating texture, for example.  Making an evenly parameterized temporary target surface, e.g.rebuilt to a high point count as Helvetosaur suggests, is a good way out. You'll see similar behavior and find a similar solution with FlowAlongSrf.

-Pascal

pascal@mcneel.com

Yes, the command is doing *literally* what it is designed to do, but probably not what 99% of the users would want or *expect* it to do.  Most people will think of a cylinder as a rolled up sheet of paper (or whatever) and want to apply flat planar curves to it as if they had rerolled the flat sheet into a cylinder.  Paper doesn't care about "parameterization"...  So methinks there should be some expert option to follow the non-rational "special" cylindirical parameterization for those who might want that, but for the average user, treat the cylinder surface as if it was uniformly parameterized and not expect them to know this rather annoying and sophisticated stuff...  My 2¢ anyway...

----H

Hi Helvetosaur...  The thing is, just making a cylinder or cone 'work' with some advanced setting is not going to make the questions go away- there are any number of cases where the uv will not be evenly distributed and ApplyCrv will return 'distorted' results. I'd  say if anything a new tool is needed to help apply 2d curves accurately to developable surfaces, but leave ApplyCrv alone, maybe rename it ApplyCrvUV or something to give a clue as to what it is doing. Deux centimes.

-Pascal

pascal@mcneel.com

I'd  say if anything a new tool is needed to help apply 2d curves accurately to developable surfaces, but leave ApplyCrv alone, maybe rename it ApplyCrvUV or something to give a clue as to what it is doing. Deux centimes.

Je suis d'accord...  ça fait quatre centimes, donc...  :)

I think a new tool will be a good way out, leaving the actual ApplyCrv alone.

 

OK, this actually is a bug Pascal - it has to do with how extrusion objects are treated (the surface in his example is an extrusion).  In the attached, I have mapped a series of 1mm curves to a cylinder NURBS surface, a cylinder extrusion object and a NURBS sphere.  The two NURBS objects are perfect, but the extrusion is off in the same way that Braganolo's objects are...

----H

Attachments:

OK, thanks, I'll poke at this some more.

-Pascal

pascal@mcneel.com

RSS

Translate

© 2014   Created by McNeel Admin.

Badges  |  Report an Issue  |  Terms of Service