Community packages?

[joepal]

I've been thinking about making monthly zips with all the assets in the user contrib repo, structured in a manner that makes it appropriate to just unzip it all in one's personal MakeHuman directory. I'd host this somewhere off-site so that it won't matter if it gets bashed by thousands of downloads.

Questions:

• Does any of the current authors think this is a bad idea? I plan to automatically create a text file along each asset containing license info, author info and description
• Do you think it's necessary to make separate zips with CC0 and CC-BY assets? (In the case any assets ever appear as AGPL, I'd have to make a separate zip of that since it's viral)

[learning]

The only problem at hand in this is that most/all contrib assets are pretty bad right now. We are simply not artists, and artists are not prepared to struggle with quirky Make* scripts. Other that that, I'm all for this idea. What can go wrong?

[joepal]

The only problem at hand in this is that most/all contrib assets are pretty bad right now. We are simply not artists, and artists are not prepared to struggle with quirky Make* scripts. Other that that, I'm all for this idea. What can go wrong?

I see that as an inspiration for people to do better.

Honestly though, I didn't expect any heroic efforts of high end design the first few months. I'm rather pleased with the occasional good pieces that have shown up.

[duststorm]

With a few vertex group templates for MakeClothes you'd already go a long way. With that I mean some base topologies of the most common clothes types, with a tweaked vertex group assignment (together with vertgroups on the basemesh) that works well with MakeClothes.
The only thing you'd have to do then is pick the template that approximates your clothes best, and weight transfer those vertex groups onto your own clothes, then tweak a bit if required.
But it would make things a lot easier and faster already.

I'll try to find some time in the near future to get back to asset production. I'd start with 3D scans of real clothes to get some very high quality results.

[brkurt]

The only problem at hand in this is that most/all contrib assets are pretty bad right now. We are simply not artists, and artists are not prepared to struggle with quirky Make* scripts. Other that that, I'm all for this idea. What can go wrong?

I see that as an inspiration for people to do better.

Honestly though, I didn't expect any heroic efforts of high end design the first few months. I'm rather pleased with the occasional good pieces that have shown up.

What we need--especially for MakeClothes--are some guidelines for mesh modeling, since our clothes are not identical to the base mesh.
For example: we could set a rule that no single vertex has more than six connections to other polygons. Thomas' rule is eight, but that gets tricky in the armpit and the crotch.

Most of the best work I've done has been accidental, because I don't really understand the MakeClothes parameters for draping and folding.
My best work--and I'm happy to compare it with anyone else's--is my Sari. There are three separate garments--halter, skirt, and wrap--none of which intersect each other. I can't however, control the folding process in MakeClothes, so as much as that one was a success, I couldn't control how the leatherwork of my warrior woman would illegally intersect with the body mesh.

If we wish for vast improvements in the quality of the clothes, hair, accessories, etc. then a clear set of guidelines needs to be created.

[joepal]

If we wish for vast improvements in the quality of the clothes, hair, accessories, etc. then a clear set of guidelines needs to be created.

I'm utterly incapable of giving such guidelines. I just usually model using the accidental approach too.

[learning]

I'd pretty much subscribe under everything said here. We need tutorials, guides, and generally making everything needed for creating assets more accessible. That was the topic of my first post on this forum, and now I'm still of the same opinion.

[brkurt]

I'd pretty much subscribe under everything said here. We need tutorials, guides, and generally making everything needed for creating assets more accessible. That was the topic of my first post on this forum, and now I'm still of the same opinion.

And now to the core of the problem: whose digital clothing are we to measure ourselves against?

If we decide to compare ourselves to the best of Marvelous Designer users, such as RosemaryR, the problem becomes that no matter how detailed the clothing model is, MD creates a unique mesh for that particular animation, and even then does not produce a pure quad result. MD did (or does not intend) to create universal animatable meshes, and the goal of MakeClothes is just that.

So, MakeClothes is intended to be a replacement for Marvelous Designer?

That brings the discussion to a major crossroads:

When you criticize the contributions, are you focusing on the lack of realistic detail for a stationary model, or the problems in illegal intersections and/or folding during an animation?

This has been discussed in detail. With respect to folding, here is the present state of things:

http://www.makehumancommunity.org/forum/viewtopic.php?f=7&t=11326&p=26621&hilit=marvelous+designer#p26621

[badwolf]

if some of y'all could help with the Project Valentina svg to mesh pipeline then lots and lots of clothes will come from there (since that is what PV is designed for)


does anybody know the tricks for Blender Cloth Sewing??

[joepal]

does anybody know the tricks for Blender Cloth Sewing??

There was a contributed tutorial on clothing that used it a while back, but apart from that I never even saw it.

See video 2 and 3 under clothing: http://www.makehumancommunity.org/wiki/ ... nder_tools