QuadMesh prototype command available in latest WIP


(David Eränen) #21

hifred,

This is what I’m aiming for. It is very tricky, but there is some good scientific literature out there which goes into how to find the base complex of a surface. At least one such technique should be “easy” to implement as part of my current algorithm. But this is not a high priority at the moment - I see it more as the long term goal.


#22

Sounds good to me!


#23

Your very welcome. It was just a first pass on playing with it and I plan to explore more as schedule permits.

s


#24

oohhh… this:

would be so awesome! but I agree that at least having a way to have some lower density that is more editable in SubD that a high-poly quad mesh would also be a great start.


#25

Some nice progress in the latest wip: V5 geometry accepted now, lower polycounts by default. Seems that symmetry evaluation still needs some tweaking, but perhaps this falls into the realm of Rhino’s shell function too, as the command failed creating a symmetrical offset of the input polysurface. Maybe something for @chuck to look into?

I think a globally available object property “symmetry” (which all commands had to respect) would help Rhino in just countless areas – it’s the only 3D app I use, where this feature is practically non-existent. Attached is another test file, some play with axis and radial symmetry and also the effects of seams being moved around, while not actually changing the shape of the underlying Nurbs model. In an ideal world the mesher would understand, that it can savely disregard the orientation of seams of some ring shaped surfaces and instead prioritize a symmetrical and orderly outcome :o)

radial-symmetry-srf-seam.zip (577.1 KB)


#26

Latest build feels quicker and the results are looking better.


#27

Here’s some simple testing I did on this week’s WIP. There’s a notes panel with some comments.

V6Test01 002.3dm (2.3 MB)


(David Eränen) #28

After a few months break from QuadMesh I’ve now gotten back to improving and expanding it. The latest WIP has a new version - go ahead and try it out! This version uses soft locking of the direction field, which, in practice, means less singularity points.

All feedback is welcome!

-David


#29

Just tried Quadmesh on the latest WIP. Worked very badly at quad size 1, better at 0.5 (example in file). But then it refused to work at all saying there were 0 sized triangles.
Nick
quadmesh.3dm (3.6 MB)


#30

This is excellent, I love the progress I’m seeing. I am going to start running this on some of the past projects I modeled in Nurbs. Hope to post soon!


#31

Ok, so I’ve ran some test. Had some successes and some failures. I moved onto a Comlicated Lego Piece.3dm (1.3 MB) I modeled. It had some issues with non-manifold edges. So I fixed that (Atleast according to Rhino 6) and then tried to quadmesh. It says the object has non-manifold edges. But when I run a showedges it says it doesn’t have any non-manifold or naked edges. So one of the two commands seems to show a bug. But maybe not?


#32

Hi Ryan,

I found two issues with the model:

(This applies to the other side as well)

HTH
-Willem


#33

Good eye Willem, I didn’t even think of that. I didn’t look too far into the model, I guess I let the technology do too much work for me.


(David Eränen) #34

Nick,

Thanks for testing. I’ve just fixed the issue you were having with zero-area triangles. It should be available in next week’s WIP.

Here is the output with QuadSize = 0.5

There are some uneven quads at the top and bottom which is something I’m working on fixing currently.


#35

Not much progress on these issues as of this week’s WIP.


#36

@DavidEranen

Here’s some additional testing on the Sept 22 WIP using the same objects. Please read the revised notes.

V6Test01 003.3dm (2.4 MB)

I’d also be grateful if you could give a description of what each of these options and reported items are:

<a class=“attachment”

(You know you’ll need to do it eventually anyway; a draft right now would help us testers :smile:)

Edited 9/24/15 10:30 pm EDT to correct link to attachment .3dm


What is the down arrow symbol pointing at a horizontal bar?
#37

This looks very promising!
And do you think it would be possible to make a “reducemesh-quad-remesher” tool based on some of the work you are doing?


(David Eränen) #38

AIW,

I was going to respond to your previous post last week as a new WIP went out, but because of build issues it didn’t go out as planned. I will get back to you once the latest version of QuadMesh is up. Things have improved in some areas and the model in your post works better now.


(David Eränen) #39

I’ve been thinking about having a post-processing step after you’ve run QuadMesh where you can selectively increase or decrease the density of any part of the quad mesh. This could also happen automatically until a certain tolerance criteria with the original surface is met.

But I’m guessing you want the reduction to happen first, and then quad mesh the result?


#40

Hi David, that sounds great.
I just thought it would be great with a (low poly) quad remesher that could be used on scan data and other meshes, that’s all, so what happens first and last doesn’t really matter to me I think.