LIA-DiTella/DiffUDF

Repo for the CVPR 2024 paper: "DUDF: Differentiable Unsigned Distance Fields with Hyperbolic Scaling"

30
/ 100
Emerging

This tool helps 3D graphics professionals and researchers create highly detailed and smooth 3D object representations from initial mesh models or point clouds. It takes a 3D mesh or a sampled point cloud as input and outputs a neural network that represents the object's shape as a Differentiable Unsigned Distance Field (UDF). Users can then render the object with various lighting and curvature options, or extract dense, oriented point clouds for further processing.

No commits in the last 6 months.

Use this if you need to accurately reconstruct complex 3D shapes from existing meshes or sparse point clouds and require fine-grained control over rendering and surface properties.

Not ideal if you need a quick, low-detail 3D model for immediate use without deep reconstruction, or if you lack a Linux machine with a CUDA-enabled GPU.

3D-reconstruction computer-vision 3D-graphics shape-representation rendering
Stale 6m No Package No Dependents
Maintenance 0 / 25
Adoption 7 / 25
Maturity 16 / 25
Community 7 / 25

How are scores calculated?

Stars

27

Forks

2

Language

Python

License

MIT

Last pushed

Oct 15, 2024

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/diffusion/LIA-DiTella/DiffUDF"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.