View Full Version : Seams in Nurbs!
11 November 2002, 03:24 AM
Can some Lightwave expert tell me why I am getting seams in the cheek of my chick's face? Thanks!
11 November 2002, 03:33 AM
you might try spinning some quads to get rid of the seams.. I've had this problem before and that takes care of mine.. I'm no expert, but thats what seems to work for me..Nice model by the way
11 November 2002, 03:36 AM
That's a CHICK???
Wow, serious muscle definition there. :eek:
Please post a wireframe. I might not be of much help, but those here that can be will like to see a wireframe to see if you have problems in the mesh. This only really happens to me if I have n-gons (polys with more than 4 sides) or some 2-siders lying around. If you've used Extender, this might be the problem. Also, when more than 4 polys meet at one point, seams will often occur.
Post that wire and someone will have better info, I'm sure...
p.s. Nice model, btw, and I wasn't critiquing...But that's one muscular chick...
11 November 2002, 03:40 AM
You might have points that arent welded there. You might have point polys there. You might have internal polys sharing that edge, it could be some opengl artifact....it could be a lot of things that are impossible to troubleshoot without seeing the actual model. :)
11 November 2002, 03:41 AM
in sub-d mode
press "ctrl+u" (unweld)
press "m" (merge points)
you should be able to see the "trouble polygons" now
delete them and rebuild.
11 November 2002, 04:01 AM
Here's the wireframe!
Thanks for the suggestions, I'll try 'em and let you know
(Ignore the blue arrows.. those are Sasquatch hair stylers)
11 November 2002, 06:38 AM
Hey thanks Yon Dabuda, that did the trick!!
11 November 2002, 09:42 AM
These are my findings about this whole thing...
These seamy nurb polys occur randomly whenever I use the unweld tool. Unweld seems to be a scary customer. From now on, anytime I need to use unweld, I will cut any involved polygons into a new layer, and unweld them where it's in a controlled environment. Do my UV business, re-merge points, repair any polys that got the seam-bug, then copy and paste the modified area back into the main object; re-merge & pray to Bongo, God of polygons that there are no new seam-bugged polys.
In the meantime, it looks like I've got 4 hours of of seam-repair ahead of me. :annoyed:
11 November 2002, 09:54 AM
i dont think its the unweld tool, i never use it, but still get it.
i would guess that it has to do with symmetry, a small glitch with its workings, although i could be way off.
11 November 2002, 04:27 PM
In my experience I've never had problems with the unweld or weld tool, but I have had problems with the merge point tool not merging coincident points. If you get seams in your renders I can tell you it's because you've got some badly constructed geometry in that location. You can use the statistics window, "w", to find single or 2 point polys. In most case though, I've found it is because the merge point tool has not merged some coincident points.
Also the opengl display will show seams that don't appear when when rendered. Here is and image showing that case. The first two are from Modeler the last two are renders from different angles showing that the seam does in fact not show up.
11 November 2002, 04:30 PM
Sorry, image size was to big. Trying again.
11 November 2002, 04:42 PM
The seams youre seeing are opengl artifacts caused by having more then 4 polygons sharing a single vertex in subpatch mode.
Typically they dont show up when frozen or rendered.
As far as unweld goes, Ive never had any problems with my geometry going funky after using it. Usually for me those seams show up if an edge or two has some unshared vertices, 2 point polys, or you have DUPLICATE polys in that area. I would hit the m key to merge any duplicate points and then hit shift i to UNIFY any duplicate geometry.
11 November 2002, 04:52 PM
Cool wire on the nose there.
The reason why you might see seams inside Modeler but none in Layout (or in renders) is because Layout subdivides and triples the mesh instead of keeping it in some special format like Modeler does (when tab'ing/subpatching).
Generally I try to stay away from intersections where one point is connecting 6 or more polys. 5 is mostly ok, but no more.
If you'd like to try and get rid of the seams, the two tris you have on the side of the nose could be merged to create a quad or a so called "diamond patch". For the seam in the center of the nose I'm not sure. You could try cutting in som new patches (using edgetools perhaps), one on top and one at the bottom. That'll do away with the 6 poly intersections I don't like so much, but yield a couple of 5 poly intersections quite close to eachother, which I don't like either. Whatever works best.
6 polygon intersections are avoidable, 5 poly aren't.
Just came up with a feature request. As is, the point statistics only shows points with max ">4" polys attached to them. I'd really like that to be configurable or atleast bumped up to ">5".
11 November 2002, 08:01 PM
I have another theory then....
When I modeled my figure, she was 36' tall. So recently, I scaled her down to 5'8". When I did that, alot of the points probably had lots of floating point digits added to them- making it more likely for numbers to be averaged weirdly and causing points not to re-merge correctly.
Anyway, it's just a theory. Anyone know a way of snapping points to a really fine grid? Like rounding them up to the nearest hundredth or something?
11 November 2002, 10:29 PM
Not an unlikely theory. Use Quantize (under Modify>Deform)! It'll let you define intervals to which all selected points will snap to.
11 November 2002, 03:45 AM
Damnit.. Quantize didn't do the trick either. I did it with a step value of 200 um. The points noticably shuffled, but new "damaged" polys were created when I unwelded & merged.
11 November 2002, 06:08 AM
quantisize, rewelding points, tried them. only thing that works most of the time is to rebuild them, and sometimes even that doesnt work. id have to move the point position a little.
01 January 2006, 10:00 PM
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.