# huanngzh/MV-Adapter

[ICCV 2025] Official impl. of "MV-Adapter: Multi-view Consistent Image Generation Made Easy"

Repository: https://github.com/huanngzh/MV-Adapter
Canonical: https://ross.abutalabs.com/products/mv-adapter
Homepage: https://huanngzh.github.io/MV-Adapter-Page/
Language: Python
License: Apache-2.0
License Family: permissive
Topics: 3d-generation, image-generation, multiview, texture-generation, multiview-generation
Last push: 2025-06-26T12:38:14+00:00

## Health v2 (maintenance only)
Score: 34/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 28, release rhythm 35, longevity 46
- inputs: {"age_days": 644, "days_push": 433, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1285, forks 95 (observed 2026-08-28T04:04:14.770143+00:00)

## What it is
MV-Adapter is a plug-and-play adapter that turns pre-trained text-to-image diffusion models (e.g., SDXL, SD2.1) into multi-view consistent image generators, supporting text-, image-, and geometry-conditioned generation as well as 3D texture generation. It is the official PyTorch implementation of the ICCV 2025 paper, with model weights, Gradio demos, training code, and a ComfyUI integration.

## Use cases
- generate multi-view consistent images from a single image
- generate multi-view images from text prompts
- generate textures for 3D models from text or images
- adapt personalized or distilled T2I models like DreamShaper or LCM to multi-view generation
- use ControlNet with multi-view generation
- train a custom multi-view generation adapter on my own dataset
- generate multi-view images for 3D scene generation pipelines

## When to choose
- you need multi-view consistent image generation built on existing Stable Diffusion models
- you want to preserve the quality and priors of a base T2I model instead of full fine-tuning
- you need geometry-guided texture generation for 3D assets
- you want a ComfyUI or Gradio workflow for multi-view generation

## When to avoid
- you need a standalone text-to-image model rather than a multi-view adapter
- you lack a GPU or need low-VRAM inference for the SDXL-based pipelines
- you need production-grade 3D reconstruction rather than multi-view images
- you need a non-Python or non-diffusers-based integration

## Facets
- artifact type: library
- maturity: active
- function: image-processing, machine-learning, deep-learning, llm-inference
- domain: artificial-intelligence, computer-vision, image-processing, graphics
- platform: python, cross-platform
- tags: diffusion-models, stable-diffusion, multi-view-generation, texture-generation, 3d, sdxl, comfyui, iccv-2025, research, 3d-generation, gpu

## Member repositories
- huanngzh/MV-Adapter (main) score 34

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:14.770143+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T04:56:39.691010+00:00, confidence not recorded.
  - readme: https://github.com/huanngzh/MV-Adapter (fetched 2026-08-28T04:04:14.770143+00:00, sha afe55d6c6b9a)
  - homepage: https://huanngzh.github.io/MV-Adapter-Page/ (fetched 2026-08-29T12:12:32.005982+00:00, sha 897b9b6fabfe)
- Data as of 2026-08-30T08:39:29.467469+00:00.