U.S.S. Kelvin CG VFX Model
- Thread starter PixelMagic
- Start date
3d modelers, I got a question if you can help me figure it out. I have an area that needs holding edges to keep a sharp panel line (marked in red). On the other side, I have a large cylinder that needs to smooth properly, so it can only have one edge running out (green). My solution for edge loop termination (blue) produces pinching and folding of the geometry after subdivision, as you can see. Any suggestions?
3d modelers, I got a question if you can help me figure it out. I have an area that needs holding edges to keep a sharp panel line (marked in red). On the other side, I have a large cylinder that needs to smooth properly, so it can only have one edge running out (green). My solution for edge loop termination (blue) produces pinching and folding of the geometry after subdivision, as you can see. Any suggestions?
You've got a couple of tris sitting in the middle of that group, highlighted in yellow below. Triangles don't always subdivide nicely, so I'd reccomend deleting the edge between them, in purple, to start. See what effect that has.
The central vertex where the tris meet also has six edges converging there. That's gonna mess with your subdivs – my knowledge is a touch out of date on stuff like Catmull-Clark, but in the older types, vertecies were weighted by how many edges they shared.
I can't see what the polys that flow into the corners of the panels are, but they look like n-gons. You ideally want all of your faces to be quads in a subdivision mesh.
Get everything quads, then use Edges menu > Spin to start playing around with the orientation of the edges. That should let you find an edge flow that gives you better results.
Last edited:
You've got a couple of tris sitting in the middle of that group, highlighted in yellow below. Triangles don't always subdivide nicely, so I'd reccomend deleting the edge between them, in purple, to start. See what effect that has.
The central vertex where the tris meet also has six edges converging there. That's gonna mess with your subdivs – my knowledge is a touch out of date on stuff like Catmull-Clark, but in the older types, vertecies were weighted by how many edges they shared.
I can't see what the polys that flow into the corners of the panels are, but they look like n-gons. You ideally want all of your faces to be quads in a subdivision mesh.
Get everything quads, then use Edges menu > Spin to start playing around with the orientation of the edges. That should let you find an edge flow that gives you better results.
Thank you. I'll take a look at this tonight after work.
Finished modeling the back of the Kelvin nacelle.
Last edited:
