CPU3DAI tools for 3D, video and audio

EMERGE-Policy: video generator — what it does and what you need to run it

EMERGE-Policy is an open source video generator that solves the image-to-3D task. The developer is Emerge. The tool turns an image into a three-dimensional result and is positioned by the authors as a system in which "the robot's mind goes beyond a single policy."

What it does

As described by the author, EMERGE-Policy is a project at the intersection of agentic AI, robotics, VLA models and a world model. The main stated task is image-to-3D: an image is fed in, and a three-dimensional scene or object comes out.

It is built on the VGGT-1B model hosted on Hugging Face. The authors do not provide detailed figures on generation quality, speed, or supported input image formats.

What you need to run it

The code is written in Python. The repository was created on August 28, 2026, and the last code change was on October 1, 2026. The latest release is v1.0.0.

All weight files take up 4.7 GB, and the largest weight file is also 4.7 GB. The weights are distributed under the cc-by-nc-4.0 license. This means the model may only be used for non-commercial purposes, with attribution.

The authors do not specify requirements for system RAM, GPU, or disk subsystem.

Who it suits

The tool will be useful to those researching the combination of generative models and robotics: the repository tags include agentic-ai, robotics, vla and world-model. The image-to-3D task makes the project interesting for prototyping three-dimensional scenes from single shots.

EMERGE-Policy is not suitable for commercial use: the cc-by-nc-4.0 weights license permits non-commercial use only. The project is also unlikely to suit those looking for a ready-made tool with detailed documentation on system requirements — the authors do not publish data on the required amount of memory or GPU type.

EMERGE-Policy is an open research project with a single v1.0.0 release and the VGGT-1B model at its core. It covers the image-to-3D task and is aimed at non-commercial scenarios in robotics and agentic AI.

EMERGE-Policy — video generator pipeline From image to 3D scene via an agent model Image input frame single shot Encoder VGGT model scene features Agent policy robotic mind decision making World model environment simulation dynamics forecast Video 3D scene camera animation frame generation Specs open source Python cc-by-nc license weights 4.7 GB metadata
How the EMERGE-Policy pipeline works. The diagram is drawn from the tool’s fact sheet.

Fact sheet

Taskimage-to-3d source
Weights licensecc-by-nc-4.0 source
Largest weights file4.7 GB source
All weights files4.7 GB source
LanguagePython source
Last code change2026-10-01 source
Repository created2026-08-28 source
Changes often — as of 2026-10-21
Latest releasev1.0.0 source
Release date2026-09-30 source
GitHub stars443 source
Forks6 source
Downloads per month532640 source
Model updated2025-03-17 source

Values are collected automatically from official sources and were checked on 2026-10-21. Each one links to its source, and values taken from the developer’s pages also carry a verbatim quote — hover over the note. Pricing and versions are shown as of the check date and change most often; verify on the vendor’s site before buying.

See also