More sophisticated parameterization?

Become a member of the CGSociety

Connect, Share, and Learn with our Large Growing CG Art Community. It's Free!

Thread Tools Search this Thread Display Modes
  07 July 2013
Also, why are the relaxation and unfolding in maya so slow?
But hey, you got new icons in maya 2014, didn't you?
  07 July 2013
RoadKill uses the same paradigm that I mentioned. Select a seam on the mesh, and then it generates the patches using the seam as a boundary. While its unfolding is better than maya (not too hard), still it can't compete with ARAP.
  07 July 2013
Well you've got your answer then, use ARAP.

And when you actually work at a studio on a job and you only have access to maya's unfold tools you can tell your boss they don't work and sit there twiddling your thumbs.

I know maya's UV tools aren't the best, latest or most intuitive and they took time to figure out but good UV layout can be done with them. It might take a little longer, and if I have access to roadkill or uvlayout I will use them as they're quicker, but often those tools aren't available so having a good base understanding of how mayas work is useful.

Last edited by Hezza : 07 July 2013 at 10:10 AM.
  07 July 2013
We're all patiently waiting for you to "fix" this, Zoharl. I'm still not sure what's broken, as the rest of us are using UV maps every day and get by just fine. In fact, without standard UV mapping techniques, my architectural texture work would be an utter nightmare.

Go forth and script your solution, my friend!
Commodore 64 @ 1MHz
1541 Floppy Drive

"Like stone we battle the wind... Beat down and strangle the rains..."
  07 July 2013
I usually do the "switch" to LSCM in RoadKill. That is the algorithm (open source blender add for it if I'm not mistaken), that makes it work. Not all work perfectly. Sometimes I have to go back and forth for half a geo head, then mirror, and sew geo and uv's back together.
.. and what is arap?
  07 July 2013
@refract, I gave a link to the ARAP paper in the previous page, where you can look at the figures that compare between the methods. Also I gave a link to their page, where they implemented a small utility that demonstrates the concept.

@infernal, a recap of this thread: roadkill (that was written by research people) provides a more sensible way (IMO) to define a seam. Most people make do with the unfolding algo, or alternatively the LSCM from 2002. Since I don't texture regularly, it was enough for me to export a mesh and texture it with the ARAP demo. Actually even that, as you could glimpse from the previous page, isn't good enough. So I sent it to a friend of mine, who has an implementation of something that called global parameterization. Instead of defining seams, it defines a few singular points (I think they called cone points) over the surface that defines a new subspace. It also uses a sophisticated quadrangulation. The bottom line is that he gave me a perfect param. for my L shape. So sorry, at the moment I don't have the time/motivation to write a decent tool (probably with CGAL) and release it to the public. Just wanted to point out that you are working with decadent tools, and maybe you should require new ones from autodesk (at least once in ten years?); but you probably know that already.
  07 July 2013
Originally Posted by zoharl: So I sent it to a friend of mine, who has an implementation of something that called global parameterization. Instead of defining seams, it defines a few singular points (I think they called cone points) over the surface that defines a new subspace. It also uses a sophisticated quadrangulation. The bottom line is that he gave me a perfect param. for my L shape.

Could you post a screen grab of the UV's that you ended up with? Am quite curious what an automated solution to the L-shape would be.

  07 July 2013
This is what I accomplished with roadkill. I put the seams where I felt was intuitive, as I would have done a cube. There texture isn't continuous across faces, and after a lot of manual tweaking, I still couldn't get it to fit perfect in a side (should have calculated it explicitly)

Here is my friend results. From some reason it consists of 3 patches, and the texture isn't continuous across faces. But it fits perfectly in each side. I need to read the paper to know the exact algorithm, and to discuss with my friend how he used it with Maya's limitations for planar texture.

  07 July 2013
Both are interesting. I think the desire to have the uv's as one complete shell while admirable, is slightly miss guided. Thats from a production perspective, although I think you're in academia/research?

Out the two I'd still pick the roadkill one, but chop it up and stack it in 0,1 so there's not so much wasted uv space. When texturing I usually jump between photoshop then zbrush/mari, so having UV's that are easy to work on/understandable in 2D is important, and the 3d paint apps aren't bothered by seams, so I'd rather have a few shells that make sense, than one unfolded model thats awkward to work on.

Last edited by Hezza : 07 July 2013 at 07:32 PM.
  07 July 2013
+1 for Ptex, but until it's ready for real-time, I'll proceed.

apologies if this has already been suggested, I haven't read all the post, for me I really get inspired by 3d coats UV mapping workflow/results. It's easy to define the cuts and the unfold and layout is recalculated real-time-ish after the cut. Does a fantastic job.

I don't know for sure what you had in mind for an ideal solution zoharl, but my 2 cents is that the best possible single cut parameterization function is still pretty useless for all but a sphere shape.

If you're talking about a practical, not theoretical solution then there are other factors that must be considered, factors that conversely make the solution easier to solve.
You way out minimal distortion against minimal wasted space in a rectangle and for most scenarios human readability. The result is you actually want to divide the shell into pieces that unfold in a logical way and can be arranged individually.
The holy grail for me is a function that makes logical cuts for me. The rest is pretty well solved.
  07 July 2013
Yeah, I'm a student, and my objectives might be different than an artist.

The editing in 3D-coat looks amazing:

(How can you go back to maya uv editor after that?)

Nevertheless, I tested 3D-coat v4 auto-mapping on my L shape and it produced something similar to the 6-planes auto-mapping in maya. Strangely, v3.7 generated something different that seems better. But as we said, we don't expect much in any case from auto tools...
  07 July 2013
Thread automatically closed

This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.
CGTalk Policy/Legalities
Note that as CGTalk Members, you agree to the terms and conditions of using this website.
Thread Closed share thread

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

vB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Society of Digital Artists

Powered by vBulletin
Copyright 2000 - 2006,
Jelsoft Enterprises Ltd.
Minimize Ads
Forum Jump

All times are GMT. The time now is 01:23 PM.

Powered by vBulletin
Copyright ©2000 - 2018, Jelsoft Enterprises Ltd.