OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
Image & Video AI

Mvedit

Mvedit is the official PyTorch implementation of 3D-Adapter, a research project for geometry-consistent 3D generation using multi-view diffusion. The repository also presents a controlled multi-view editing method for generic 3D diffusion models and provides a Gradio interface and APIs for inference.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Mvedit — OpenViST project profile in Image & Video AI.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use Mvedit when you need the codebase for local inference or research into multi-view diffusion for 3D generation, and can meet its GPU and software environment requirements.

Who it is for

  • Researchers exploring multi-view diffusion and geometry-consistent 3D generation.
  • Developers who want to run or adapt the PyTorch implementation.
  • Users with a suitable GPU who want to try the local Gradio inference interface.

Documented uses

  • Run inference through the Gradio Web UI.
  • Access the interface's automatically generated API documentation and use the project's Python API examples.
  • Study or build on an implementation of controlled multi-view editing for 3D generation.
Before you choose

Requirements & limits

The repository is public and specifies an MIT license. Its local Web UI requires a GPU with at least 24 GB of VRAM.

  • The recommended local inference interface has a stated minimum of 24 GB of GPU VRAM.
  • The documented Linux environment uses CUDA Toolkit 11.8 or later and PyTorch 2.1 or later.
  • First-run model downloads require network access to GitHub, Google Drive, and Hugging Face.
  • The README says Windows can work if configured correctly, but some packages may need installation adjustments.
  • The locally generated Gradio API documentation may show inaccurate data types or default values; the README recommends using the Web UI defaults as reference.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Set up the local inference interface

Install the documented environment and start Mvedit's Gradio Web UI for local inference.

Before you start

  • Linux, CUDA Toolkit 11.8 or later, and PyTorch 2.1 or later.
  • A GPU with at least 24 GB of VRAM for the Web UI.
  • Conda, Python 3.10, and network access to GitHub, Google Drive, and Hugging Face for model downloads.

Steps

  1. Create and activate a Conda environment with Python 3.10, NumPy 1.26, and Ninja.

  2. Install PyTorch and its CUDA package for a suitable CUDA version.

  3. Clone the repository and install dependencies with the command below.

    Documented command
    pip install -r requirements.txt

  4. Start the interface with the command below.

    Documented command
    python app.py --unload-models

Expected outcome

The README says the Web UI is available locally at `http://localhost:7860`; it loads models on demand.

Guide limitations

  • The first run may take a long time while models download.
  • FFmpeg and x264 are optional and are specified for exporting videos.
  • Windows may require installation adjustments for some packages.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
17 October 2026
Original website title
Lakonik/MVEdit · Official homepage title · retained in its source language
How we review sourcesSuggest a correction