Mesh generation w/deep learning for babylonjs

Hi There,

Is anyone working on a pipeline (and/or have they integrated) a deep learning solution for mesh generation? I’m doing a lot of research into different kinds of machine learning integration for babylon, came across this PyTorch3D · A library for deep learning with 3D data and am beginning to see people doing this kind of work. At this point, I’m trying to identify people in the community with similar interests. (I’ve done a bit of fine-tuning on GPT2, so I can feed it text and it will spit out similar text; I’m now trying to do the same thing for meshes).

Also, update on reinforcement learning, I’ve found a library for reinforcement learning in javascript, and am working to build some 3D and VR environments for agent training.

this reply is a year late. I’m interested! I’m just now dipping my toes into deep learning. Got myself the deep learning with javascript which is about tensorflow.js but only in the first few chapters. I’m interested in mesh generations from photos and also animation of characters with reactions to physics. But don’t have a clue where to start yet.

don’t worry about it

apparently the library for this is this: GitHub - NVIDIAGameWorks/kaolin: A PyTorch Library for Accelerating 3D Deep Learning Research</tit

it looks incredible dope

also, apparently, there is a no code? (low code?) gui: NVIDIA Kaolin | NVIDIA Developer

if you get this running, would you let me know?

it’s on my to do list… but I’ve had some trouble with it.

Here are some searches to play with

Style transfer is self explanatory
Gan can be used to upscale, also try “waifu”

I’ve been wanting to do something with materials and style transfer for a while. Kinda like this GitHub - reiinakano/arbitrary-image-stylization-tfjs: Arbitrary style transfer using TensorFlow.js

1 Like