search

samxuxiang / BrepGen

[SIGGRAPH 2024] Official PyTorch Implementation of "BrepGen: A B-rep Generative Diffusion Model with Structured Latent Geometry".
Other
227 stars 28 forks source link

Conditional Generation #8

Closed LaFeuilleMorte closed 4 months ago

LaFeuilleMorte commented 5 months ago

Thanks for your brilliant work. Can we use other condition to generate objects? If we train the b-rep model along with images/text, can we generate objects given image/text prompt? Another question about the dataset: Can we use meshes or pointclouds to train this model , as B-Rep model datasets are still very small.

samxuxiang commented 4 months ago

Hi @LaFeuilleMorte, you can modifify the network to take in other conditions. We did this for the furniture dataset using class embedding. The data representation between pointcloud and brep is very different. There are better model (e.g point-e) for pointcloud generation if you are interested.

LaFeuilleMorte commented 4 months ago

Hi @LaFeuilleMorte, you can modifify the network to take in other conditions. We did this for the furniture dataset using class embedding. The data representation between pointcloud and brep is very different. There are better model (e.g point-e) for pointcloud generation if you are interested.

Thanks for your kindly reply. May I ask another question regarding the data structure? Can I convert other 3D formats (obj, glb, ply ...) to brep representations? Thanks in advance for your reply.