I think I solved it with a slight redo of the upper section, kept the rim and the phaser arrays, well the big ones, and the dorsal connector so it wasn’t near a total redo. Not perfect still a few little pinches but drastically reduced.
I want to re-iterate, this is not a problem with blender. You will have this issue in every 3d package because the issue is bad topology. Reading through the topology guides @Lumpengnom linked will help. As would watching a few video tutorials on hard surface modeling.
To solve this particular problem you may want to use creases instead of beveling.
Or adding control edges instead of beveling.
Or using boolean brushes instead of beveling.
Triangles and Ngons will only ever look good in a subdivided surface if they are completely flat. Otherwise you’ll likely get the pinching you are experiencing.
I’ve tried creases but have never heard of booean brushes. Have to look into that, thanks.
Honestly, this problem can’t seem to be solved by any reasonable solution. I’m just going to have to live with ninor distortions as some if the solutions just make worse problems or are too difficult to implement they would take an unacceptabble amount of work to carry out.
Some problems just aren’t worth the effort it would take to solve them.
It can be solved but the solution is a more carefully planed topology.
From the screenshot above it all looks a bit “boxy” and doesn’t really flow with the layout. I think you’re going a bit overboard with the bevels if you’re adding subsurf. Just 2-3 “control” edges should be more than enough when you subdivide the mesh.
The Topology guide from Lumpengnom above is pretty good but here’s also a great thread from Polycount with proper ways to model complex shapes.
This isn’t something specific to Blender, you’d run into these unclean bevels in pretty much any equivalent 3d software. What it all comes down to is the topology and using the edge flow to control the subdivisions.
Edit: Here’s another Polycount thread, from someone with a similar problem with the solution.
It can take some work to plan and figure out the proper way to puzzle things together and if you don’t want to remake this asset, that’s fine, just be more mindful of these things from the start in your next asset.
Look, people said “start over” and having put in waaay too much work on parts attached to the lander that wasn;t an option as i’d have to redo them all. No. I did eliminate the enter upper half and redo it with more edges. Same problem.
Should I usut dump subsurf altogether and try it without it?
More edges doesn’t solve anything if they’re not correctly placed. If you’re going to subsurf, very few edges are actually needed, the subdivisions will take care of the rest. As others have said here, clever use of poles and proper edgeflow is the way, not adding needless amount of edges in grid patterns.
Since your screenshots above are probably outdated, a newer one without subsurface enabled would help, but based on that screenshot your edgeflow around the corners is far from optimal and inefficient for a subdivision workflow.
You don’t have to start from scratch, you could just as easily fix things with the knife tool, removing edges and joining others with the j key, it could be a few hours work, sure, but modeling in general is a time consuming process.
Just take clear screenshots of what you have, paste them to imgur or similar site (takes 20 seconds) and that would help us point out how you could improve the mesh.
Did you take a look at the polycount threads? I know 170 pages of information is a lot but going through just a handful of pages has helped me figuring out stuff in the past. There is a lot of solutions there that would be very applicable to a star trek ship.
Look at these corners. These have proper support edges, no pinching artifacts. Quite different from the edgeflow on your screenshots above.
I know I’m reiterating what others have said, but that’s because they’re right. Did you apply some of the stuff from Lumpengnoms topology guide link?
I think you should start by learning proper topology. It’s a lot to learn, but without that you will find yourself with “unsolvable” problems pretty often.
I’ve reduced the problem with knife cuts to bisect ngons and taking a done of skruittol which keeps me from being so upset over a few ugly edges here and there.
But if I have to make a duplicate mesh for every piece I want to have openings in, subdivide it like 2 levels more than the regular mesh or vice versa, Hellllllllllll no!
Again, I’ve looked it over in blender 2.8 and 2.79 and it seems that splitting ngons into pairs of quads killed 90% of my uglies. There are very faint ones that are much smaller and less noticeable now, i’m fairly pleased.
From what ive seen your trying to add detail to a very basic/coarse mesh which will create pinches and bumps.
The mesh your detailing should be, in my opinion, atleast triple the polygons to support the detail your trying to add so you avoid unwanted irregularities on such a smooth surface.
Also if you look how similar surfaces are manafactured its not one continuous surface but many different panels. Jets or airplanes are fine examples of this
I would recommend rebuilding the mesh so theres enough resolution to support these details. But from the sound of it you’re not really interested in rebuilding the mesh so your stuck with these irregularities.
Guys, I appreciate that you think starting over is the solution, but I’m just not doing it. Dammit I’ve been on blender for years and all i ever do is start over. and over and over and over and over. I never get finished because the only solution to anything is to start over. I may not get ahead at it but i’m not going back anymore. Going back never gets me ahead.
Ok this might be a bit harsh but you gotta understand it.
You are trying to something, but you lack the ability and knowledge to do so.
Because you are still not getting the most important thing here: YOU NEED TO LEARN PROPER TOPOLOGY.
Doesn’t matter how many times you start over, without this knowledge you will NEVER produce excellent results. You shouldn’t even try to model something if you don’t know how. You are not qualified - why waste your time?
It only is an unsolvable problem because you never made any serious effort in finding the actual solution.
Go back to “school” and learn proper SubD modelling. It’s not particularly difficult, but it isn’t obvious either, you will not find a solution by stumbling in the dark trying to find your way alone.
After Reading this post I am going to study topology.
I’m just ignoring the minor uglies, there, problem solved. Please close this thread.