VGGT-1B-Commercial (Daub build)

A repackaged copy of Meta's VGGT-1B-Commercial (VGGT: Visual Geometry Grounded Transformer, Wang et al., CVPR 2025), prepared for the Daub desktop app.

Changes from the original: the aggregator weights are stored in bfloat16, and the point and track heads are removed. The camera and depth heads are unchanged. Nothing was retrained.

File	Size	SHA-256
`daub-vggt-1b-commercial.safetensors`	2.81 GB	`1ae4259a19311e4372573b036e813ca15f166187ec1dc0d4da39b99393939a2f`

Licence

This model is distributed under the VGGT License (LICENSE.txt in this repository), the same licence as the original. It permits commercial use and includes an Acceptable Use Policy that prohibits, among other things, military, warfare and weapons uses. By downloading or using these weights you agree to it.

Limitations

Camera positions and depth estimated by this model can be wrong, especially with few, blurry or featureless photos. Results are not suitable for measurement, surveying or any safety-critical use.

VGGT, the original weights and code are by Meta Platforms, Inc. This repository is not affiliated with or endorsed by Meta. Original code: https://github.com/facebookresearch/vggt

Downloads last month: -; Downloads are not tracked for this model. How to track

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for another-nate/daub-vggt

Base model

facebook/VGGT-1B-Commercial

Finetuned

(1)

this model