Request access to the World Tracing object model
These checkpoints are released for research and product experimentation under the MIT license. Please share a few details below so we can keep a light audit trail of how the weights are used in the wild. Requests are reviewed manually, typically within 1-3 business days.
Log in or Sign Up to review the conditions and access this model content.
World Tracing — Object Model (6-layer, r75b)
Access
The checkpoints in this repo are released under the MIT license, but downloads are gated so we can keep a light audit trail of how the model is used. To download:
- Scroll up and fill in the "Submit access request" form (basic contact info + a short note on intended use).
- We review every request manually, usually within 1-3 business days. You will receive an email from Hugging Face once your request is approved.
- After approval, log in with
huggingface-cli login(or setHF_TOKEN) and run any of the inference examples from the GitHub repo — thewtpackage picks the token up automatically and--ckpt r75b/r69e/r76triggers a normalhf_hub_download.
Note: this is a manual review flow, not an auto-approve click-through. We read every request individually, so please give a one-line description of what you plan to use the weights for.
EMA-only release weights for the r75b object model from World Tracing: Generative Pixel-Aligned Geometry Beyond the Visible.
- Repo: https://github.com/haoz19/world-tracing
- Project page: https://haoz19.github.io/world-tracing-page/
- Config name:
r75b - Architecture:
MultilayerXYZModel(DINOv2-vit-L encoder + 6-layer diffusion head), 1.7 B params - Input: 504 × 504 RGBA, alpha-matted single object
- Output: per-layer XYZ in camera space, 6 stacked depth maps (visible surface + 5 occluded layers behind it)
- Training data: Objaverse renders + curated public 3D-asset corpora
Files
| File | Size | Format |
|---|---|---|
model.pt |
6.21 GB | bare state_dict, float32 |
This release contains the EMA weights only (no optimizer / config / gradients) so the download is ~26 % of the original training checkpoint.
Usage
git clone https://github.com/haoz19/world-tracing
cd world-tracing
pip install -e ".[viz,bg]"
python examples/infer_rgba.py \
--image examples/test_images/object/obj014_leather_briefcase.png \
--ckpt r75b \
--config r75b \
--out /tmp/wt_obj.rrd
Bare --ckpt r75b triggers huggingface_hub.hf_hub_download against
this repo and caches the weights under ~/.cache/huggingface/hub/.
First run downloads 6.21 GB; subsequent runs are instant.
Citation
@misc{zhang2026worldtracing,
title = {World Tracing: Generative Pixel-Aligned Geometry Beyond the Visible},
author = {Hao Zhang and Mohamed El Banani and Jen-Hao Cheng and Paul Zhang
and Yi Hua and Ben Mildenhall and Christoph Lassner
and Narendra Ahuja and Gengshan Yang},
year = {2026},
eprint = {TODO},
archivePrefix = {arXiv},
primaryClass = {cs.CV}
}
License
MIT — see the GitHub repo.