Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

Hi, quick feedback: the demo is extremely short, so I can't really say much. Please generate more complicated scenes and, most importantly, inspect the wireframe. From what I could glance from the demo, the generated models are tri-based instead of quads, which would be a showstopper for me.


Just curious: why do you prefer/have a requirement of quad-based meshes?


Because traditionally, Blender modeling works best on a clean quad-based mesh. Just look at any modeling tutorial for Blender and one of the first things you learn is to always keep a clean, quad-based topology, and avoid triangles and n-gons as much as possible, as it will make further work on the model more painful, if not impossible. That starts with simple stuff like doing a loop cut to things like uv-unwrapping and using the sculpting tools. It's also better for subdivision surface modeling. You can of course use tri-based models, but if you want to refine them manually, it's often a pain. Usually, for me it's pretty much a "take as-is or leave it" situation for tri-based meshes, and since I see these AI-created models more as a starting point rather than the finished product, having a clean quad-based topology would be very important for me.


Is this true even if you do only or mostly sculpting?


No. But for animation meshes, it's the norm to use only quads. Mainly because of topology/retopology issues.


Sometimes texture artists like this a lot more.


Yes, because uv-unwrapping is much more predictable with quads, and you can place seams along edge loops. I'm by no means an expert here, maybe there are tools which make this similarly easy with non-quad topology, but at least from what I've learnt, the clean grids you get form quad-meshes are simply much easier to deal with when doing texturing.


On it, thank you for feedback




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: