View Full Version : SubD problem with Chamfered edge

12-11-2014, 03:37 PM
I'm doing a Chilton Webb tutorial on Sub D. He's chamfering a corner of 3D rectangle and then putting it in sub D and it's still a solid object. When I do it it's a mess. Pics included. I make the rectangle with the box tool. Select a point. I activate the Chamfer button and drag the point to chamfer the edge. I hit tab and I get a mess, he isn't. What am I doing wrong?
Thanks Bill

Oedo 808
12-11-2014, 04:01 PM
Providing there doesn't appear to be any other steps I would guess he's using Catmull-Clark SubDs.

12-11-2014, 04:12 PM
Providing there doesn't appear to be any other steps I would guess he's using Catmull-Clark SubDs.

Thats it. You'll have to change over to Catmull-Clark from the default subpatch. I personally don't recommend modeling like that though, using CC is a sloppy method and can come back and bite you in the butt.

12-11-2014, 04:36 PM
Well there has to be additional steps because even with Catmull clark subdivision, you wouldn't end up with something like what he shows in his tutorial which I assume is from this series here. https://www.youtube.com/watch?v=SFEiwHH9JO4
To create that same object in subdivision surfaces without Catmull Clark, you would need to cut in additional edges something like this:


Its all a matter of understanding how LightWave (and other 3D apps) subdivide polygons. In this case, simply chamfering a single point anywhere on a mesh is going to result in ngons (polygons with more then 4 points) which will not work with subdivision surfaces. Catmull Clark Subdivision can be used but even then, you have to consider how those polygons are going to divide over the surface of a mesh and relying on CC's to work where subpatches don't isn't always an ideal solution. Any type of operation which results in ngons is going to require some clean up to a degree. Ngons with CC's can be perfectly acceptable on flatter areas of a mesh but any time they're going across a corner or over a curved surface, they're going to be problematic.

12-12-2014, 09:55 AM
Wow. I don't remember exactly what I did there, but I'll go check it out and report back.

I was smarter back then :D


12-13-2014, 11:20 AM

Okay, I dug up the original files and this makes a little more sense now. That's not the only step I took. The end result looks like this.


And it's not perfect, but the imperfect parts are on flat planes, so they don't show up. This is all hard surface modeling, and is not designed to be deformed. If you were going to do this for a deformable surface, of course, you'd want to do a little more cleanup and get rid of any tris and ngons.