Looking for MH developer for Valentina 3D

[duststorm]

In any case, quadrangulation can only be done sensibly AFTER stitching the planes together (as you want continuity across seams), which at this moment still happens in Blender. So quadrangulating in Blender is the obvious option.

[brkurt]

In any case, quadrangulation can only be done sensibly AFTER stitching the planes together (as you want continuity across seams), which at this moment still happens in Blender. So quadrangulating in Blender is the obvious option.

I'm just a contributing artist--not a programmer--but I'm thinking the logic would be:

(1) Valentina interface solves valency and 'least-mean-squares' (evenly spread out quadrangles) issues;
(2) Final quadrangulation is a python script added as a Valentina add-on to Blender (as in the manner of mhx2).

[learning]

In any case, quadrangulation can only be done sensibly AFTER stitching the planes together (as you want continuity across seams)

Is it really a requirement? Because it makes the problem much more complex. Wouldn't making reasonably topologically sound parts and then fitting them onto a manequen and stitching them together (similarly to how it's done IRL) work as well?

[duststorm]

Is it really a requirement? Because it makes the problem much more complex. Wouldn't making reasonably topologically sound parts and then fitting them onto a manequen and stitching them together (similarly to how it's done IRL) work as well?

It would work. But unless your quads divider code run from Valentina knows how seam areas are going to fit (be stitched) onto each other, the lines of your quads will not match up. If the goal is to get a single connected mesh (where possible), then you will have to connect/merge the vertices at the seams together. This is more or less what marvelous does.
I believe MD applies a quad algorithm on the 3D model, after it was stitched together.

But in any case, having quads will already be an improvement.

[slspencer]

Hey this is the type of information we are looking for.
Anybody who knows how to solve this problem want to join the Valentina team?

Is it really a requirement? Because it makes the problem much more complex. Wouldn't making reasonably topologically sound parts and then fitting them onto a manequen and stitching them together (similarly to how it's done IRL) work as well?

It would work. But unless your quads divider code run from Valentina knows how seam areas are going to fit (be stitched) onto each other, the lines of your quads will not match up. If the goal is to get a single connected mesh (where possible), then you will have to connect/merge the vertices at the seams together. This is more or less what marvelous does.
I believe MD applies a quad algorithm on the 3D model, after it was stitched together.

But in any case, having quads will already be an improvement.

[learning]

Hey this is the type of information we are looking for.
Anybody who knows how to solve this problem want to join the Valentina team?

Is it really a requirement? Because it makes the problem much more complex. Wouldn't making reasonably topologically sound parts and then fitting them onto a manequen and stitching them together (similarly to how it's done IRL) work as well?

It would work. But unless your quads divider code run from Valentina knows how seam areas are going to fit (be stitched) onto each other, the lines of your quads will not match up. If the goal is to get a single connected mesh (where possible), then you will have to connect/merge the vertices at the seams together. This is more or less what marvelous does.
I believe MD applies a quad algorithm on the 3D model, after it was stitched together.

But in any case, having quads will already be an improvement.

I only know about 2d quadrungulation (the algorithm in the paper linked in one of my previous posts) and have some foggy ideas as to how to make the mesh better (again in 2d). Those are just theoretical constructions, I never tried to actually code them. This said, with a bit better code documentation I can try to ponder a little and see if I can help (I read Russian and Ukrainian btw, so documentation in those languages would do).

[slspencer]

Hi, learning!
Why don't you look into generating a valentina pattern as 3D wavefront .obj using quads?
It doesn't matter whether or not you come up with a complete solution right away, all of our breakthroughs have been incremental (and isn't this always the way of things? )
Please look at bitbucket.org/dismine/valentina
It's written in C++ using Qt, since Qt makes working with vector graphics so very easy.

I only know about 2d quadrungulation (the algorithm in the paper linked in one of my previous posts) and have some foggy ideas as to how to make the mesh better (again in 2d). Those are just theoretical constructions, I never tried to actually code them. This said, with a bit better code documentation I can try to ponder a little and see if I can help (I read Russian and Ukrainian btw, so documentation in those languages would do).