microsoft / microsoft/TRELLIS.2
Black and reflective spots on light images
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 11.3k
- Forks
- 1.4k
- PR merge metrics
- No merged PRs in 30d
Description
I'm trying to make a simple fastapi service around Trellis and it's mostly working fine, but for some cases it generates odd results. Take for example the attached "labubu"
This yields a mostly correct topology, but there are weird black spots on the texture.
I used the following settings:
# nvdiffrast simplify
simplify_target: 2000000
# o_voxel postprocess to_glb
decimation_target: 50000
texture_size: 1024
remesh: True
remesh_band: 2
remesh_project: 0
Github doesn't let me upload a .glb, so here is a drive link: https://drive.google.com/file/d/1cwhNmLm_R7y_2k2tE2dDM-s7alWRUCx3/view?usp=sharing
You can use https://sandbox.babylonjs.com/ or other similar tools to view the glb.
I tried adjusting the levels, contrast, saturation, brightness and such and nearly all of them reduce the black surface area, but still leave little dots.
Contributor guide
No contributing guide indexed for this repository
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Reproduce the labubu case using the listed nvdiffrast and o_voxel settings, then inspect the generated GLB in Babylon.js or another viewer. Compare the texture output with and without remesh and the stated postprocessing values; done means identifying a reproducible cause and eliminating the black or reflective spots without relying on manual image adjustments.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- fastapi, python
- Domain
- computer-graphics, machine-learning
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Quiet
- Clarity
- Needs clarification
- Newbie friendliness
- 35/100