Hi Will,

> Yes, I agree that the surfaces are actually joined but perhaps
> a coined term is they're not aligned?

It is tough to create terminology for this stuff. But this problem here:



seems to be described pretty well by an "alignment" problem - in this case the vertices that were created are shifted in position from one another, in other words not lined up or not aligned...

I guess that one is a "watertight vertex alignment", and the other one that you are talking about is a "quad to quad alignment", where you're talking about a situation where only quads are created along a shared edge and that you want matching quads between the pieces instead of n-gons.


> What I typically do as in exactly the vertices you circled is to
> move them around and merge them together, add additional
> edges where required etc.,

Can you describe a bit more the reason why you would be moving these around or adding more of them in?

The more concrete things and actual example models that you could show would help me to understand what you are trying to do better and also give me some better test cases for trying to improve it.

- Michael

Hi Michael, yes absolutely I wanted to send you what I was working on and get some of your advice - I was going to email you but any info here will certainly be helpful to anyone doing similar work.

As you know I love to gather reference material and practice by modeling and thereby improving my skills.

I started working with this image (of course, all props to the creator of this gorgeous model!):


After various attempts creating the main shape, I got to this point:



Since the ultimate goal for the majority of my models is for a game I'm working on (think Wing Commander type game) I'll probably have to go back to direct polygon modeling but of course I can't help but love how MOI lets you get your ideas and concepts modeled really easy!

You can play around with meshing parameters for a good amount of time trying to "balance" the resultant mesh but this is what I settled on.
You can see how the fillet edges wonderfully matches the front edge on the "barrel" (however the main body's edges dont' line up at all):






But to show the kinda things that make what's left of my hair fall out, while the front of the barrel is perfect, this is what I get at the back:


You'd think that the mesher would continue around to the rear end and give results similar to the front!


Of course rendering the model in Cinema 4d gives smooth results:



However with my models ultimately targeting a game engine, ngons will most likely not be an option anyway requiring me to either clean up the mesh or retopologize the model - this is where I started merging points and added edges to better match in certain areas:


But that quickly turns into hours of work...


-Will

Hi Will, thanks for posting the example, that does help a lot.

> You'd think that the mesher would continue around to the rear end
> and give results similar to the front!

If you line each of those fillet pieces up with each other, you can see that they are different lengths:



Due to the curvature of the hull part, one of these pieces got chopped down to be smaller than the other side.

That difference in length will right now create a difference in the generated mesh as well - that's currently the reason why one side has a different mesh structure than the other.


But this is the kind of situation that I think I can improve, because it has a plane on the other side of it. It won't be exactly easy, but I should be able to come up with a method that will align quad structures better in this case.

When there are 2 just general curved surfaces on either side of the fillet, that case will be way more difficult yet to try and align because there will be kind of 2 different alignment pressures on the same surface. With a plane on one side, it can kind of "relieve" the pressure for alignment on that side, allowing me to focus on aligning to just one other structure (that's currently my theory anyway).

I'm not exactly sure when I will be able to accomplish this, it will require a fair amount of restructuring of the mesh processing since things will be more dependent on being meshed in a certain order as well.


re: Game engine - sorry, I didn't quite pick up that was the final goal for this before.

Of course, that is one of the most difficult types of meshes to create, typically those are done by someone using great care in the placement of every single point... It's pretty difficult to set up automatic processing that replicates that exactly.

But I think I should be able to get one notch closer to it, and the specific example that you mention here should be doable. Your model here will be helpful as a test case when I work on that, if you run across other examples that you wished created different mesh structures, please send those when you run into them.

- Michael

Thanks for the explaining what's happening Michael.

Of course I created the barrel and performed the fillet and mirror first before the boolean with the ship hull but yes, I see that the result did shorten the rear side as you have shown.

Thinking a bit, perhaps it might be easier if instead of attempting to create a complete model in MOI, I can use it to create subassemblies which I can piece together back in my poly app, hmmm...

I'd still have to be careful how I create the individual parts though!

While more work, hopefully it would be less work than trying to repair the whole model as I attempted before.

It's just looking at that model, it just seems like something you'd totally want to create in MOI! (And of course you can, it's just the resultant topology is not going to be game engine friendly!)

-Will

Just wanted to say I really like the look of your ship, sorry for off topic!
David W.