Hacker News Reader

Comment by astrashe2 | Hacker News Reader

Were any people who work for the garment industry involved in GNL's creation, or is it something that's coming entirely from tech people?

View on HN

Tech person - there's only one contributor, it's less than 48 hours old, and appears to be primarily vibe coded with the assistance of Claude Code. No mentions of types of stitches even though it's crucial to understanding how a garment is made. I wonder too if this grammar can represent a glove made from a single strand of yarn.

If I understand what you mean, that's more in the realm of knitting which does already have several rigorous notations in common use.

This is for pattern drafting, which assumes knit or woven fabric as the raw material for the garment construction, along with the pattern.

That said it still does not seem suitable for this task based on my experience sewing from and modifying patterns.

It looks like it's missing so much that you'd need even to hand-sew a pattern at home. There's no mention of interfaces or bindings.

This looks more like something for making clothing as digital content - e.g. Marvellous Designer. Possibly more straightforward even.

Edit: found interfacing. It calls it "interlining".

It's nowhere near Marvelous Designer. Marvelous Designer is for making 3D clothing for games, animation, and such. It's a limited version of Clo[1] , which is for making real-world clothing. Clo lets you design clothing, put it on an avatar, and watch it move and drape with clothing physics. It looks real. When you see good clothing in a game, it was probably created with Marvelous Designer.

Then Clo exports a file for fabric cutting compliant with the ASTM D6673-10 standard, Standard Practice for Sewn Pattern Data Interchange, which is used for the production of garment patterns. It's kind of clunky, being based on Autodesk DXF, AutoCAD's export format from the 1980s, but it's what the industry uses. You can bring such files into anything that reads DXF and view them. So a widely used formal descriptive language for fabric cutting already exists. You can send those files to a contract garment manufacturer and get garments back.

Marvelous Designer is just Clo minus the cutting pattern export feature.

[1] https://www.clo3d.com/en/

[2] https://www.normsplash.com/Samples/ASTM/191361149/ASTM-D6673...

Stitches are load-bearing, so specifying a bartack or a flatlock seems pretty important to unambiguously specifying a garment. Along the same lines, I don't see a way to specify hardware that isn't for closures, e.g. the rivets used to reinforce denim pockets.

I know, I make clothes too. Probably unlike the creator of this thing.

But the comment I was responding to seemed to be using "stitch" in the way knitters use it, not the way sewists use it. No pattern drafting system can represent the stitches necessary to create a panel of knit fabric, that's simply not the level of abstraction they work at.

This thing isn't good but not for the reason of being unable to represent a one-strand mitten or whatever, which is what I think they were getting at.

Well, I actually had two interrelated thoughts and because of proximity I think I confused things. I guess what I was thinking was "garments are constructed not of "panels" but of threads of a given material which can be abstractly thought of as being panels when woven or knitted, but ..." and from there I thought of failure modes, like the fact that this doesn't have a way of specifying straight vs zigzag stitches, which doesn't have a way of specifying things that are not joined together via stitching panels together, etc. Like, I don't think this can specify a pair of jeans, because the hem of a jean requires a chain stitch at the bottom, which isn't unambiguously defined. This project feels like it devalues the complexity of something that is one of the defining features of civilization.

Is this even able to specify patterns? Or is it just how to assemble the pieces of cut cloth?

It's Claude Code slop

Reply on news.ycombinator.com