---
id: capture/vggt
about:
  depiction: capture/vggt.jpg
  label: VGGT-Ω
  description: >-
    Feed-forward 3D vision model from Meta AI and Oxford's Visual Geometry Group
    that predicts camera poses, depth and point clouds from video; used as a
    reconstruction backend in DeepReefMap for coral reef mapping.
  tags:
    - 3d reconstruction
    - computer vision
    - depth
    - camera pose
    - coral reef
    - video
tool:
  form: research model
  group: 3D capture and reconstruction
  homepage: 'https://github.com/facebookresearch/vggt-omega'
  licence: noncommercial research licence
  cost: free
open:
  source: partly
  data: unknown
  api: unknown
  participation: unknown
  licence: FAIR Noncommercial Research License
provenance:
  origin: agent
  sources:
    - title: 'GitHub - facebookresearch/vggt-omega: [CVPR 2026 Oral] VGGT Omega'
      url: 'https://github.com/facebookresearch/vggt-omega'
    - title: >-
        GitHub - eceo-epfl/deepreefmap: Tools for reconstructing and mapping
        coral reef scenes from underwater images
      url: 'https://github.com/eceo-epfl/deepreefmap'
  created: '2026-10-17'
  reviewed: '2026-10-17'
  reviewed_by: 'Claude, from the sources listed'
  status: draft
import:
  source: openai-linkedin-scrape 2026-10-17
  id: DT042
  kind: enabling_model
  family: enabling_technology_and_adjacent_work
  priority: supporting
  verified: true
  posts:
    - P107
  posters:
    - Jonathan Sauder
domains:
  - places
---
VGGT-Ω (VGGT-Omega) is a family of one-billion-parameter models, presented as an oral paper at CVPR 2026 (arXiv 2605.15195).[^1] Feed-forward means the model computes its answer in one pass through the network.[^2] Given a set of frames, it returns camera poses and intrinsics (lens settings such as focal length), per-pixel depth with confidence, which can be unprojected into a point cloud, and camera tokens.[^1] Variants are trained at 512, 416 and 256 pixel resolution. One checkpoint adds a text-alignment embedding.[^1]

DeepReefMap, from EPFL and MIT, builds a semantic 3D point cloud, an ortho-mosaic and benthic cover statistics for a reef from handheld underwater video.[^2] An ortho-mosaic is a stitched overhead image with perspective removed. Benthic cover is what covers the seabed, such as coral or sand.[^2] VGGT-Ω is one of DeepReefMap's four reconstruction backends, and its README rates it high quality.[^2] Unlike the streaming alternative, it looks at the whole clip at once, so GPU memory caps the length of video it can handle.[^2] The README quotes upstream figures of roughly 6 GB for the model plus about 75 MB per frame on an A100.[^1,2]

The code is on GitHub.[^1] The weights are at facebook/VGGT-Omega on Hugging Face, behind an access request that is reviewed automatically.[^1,2] DeepReefMap names the licence as the FAIR Noncommercial Research License and keeps the backend as an opt-in extra for that reason.[^2] The VGGT-Ω repository points readers to its LICENSE file and lists updates from September 2026.[^1] The model runs on NVIDIA, AMD and Apple Silicon GPUs.[^2]