# Tencent-Hunyuan/HunyuanCustom

HunyuanCustom: A Multimodal-Driven Architecture for Customized Video Generation

Repository: https://github.com/Tencent-Hunyuan/HunyuanCustom
Canonical: https://ross.abutalabs.com/products/hunyuancustom
Homepage: https://hunyuancustom.github.io/
Language: Python
License: NOASSERTION
License Family: other
Topics: audio-driven, diffusion-models, image-to-video, image-to-video-generation, video-editing, video-generation
Last push: 2025-10-15T06:54:23+00:00

## Health v2 (maintenance only)
Score: 40/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 47, release rhythm 35, longevity 34
- inputs: {"age_days": 483, "days_push": 322, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1227, forks 110 (observed 2026-08-28T04:04:03.257316+00:00)

## What it is
HunyuanCustom is a multimodal-driven customized video generation framework built on HunyuanVideo, supporting image, text, audio, and video conditions with subject identity consistency. It ships inference code, model checkpoints, and ComfyUI integration for single-subject, audio-driven, and video-driven video customization.

## Use cases
- generate videos featuring a specific person or subject from a reference image
- create talking-head videos driven by an audio track
- edit or restyle videos while keeping subject identity consistent
- generate multi-subject videos with consistent characters
- run customized video generation locally with low VRAM
- use customized video generation inside ComfyUI workflows

## When to choose
- you need identity-consistent video generation conditioned on images, text, audio, or video
- you want an open model with checkpoints and ComfyUI support for video customization
- you have a GPU (or as little as 8GB VRAM via WanGP) and want local inference

## When to avoid
- you need fast real-time video generation rather than offline diffusion-based rendering
- you need training/fine-tuning code rather than inference
- you cannot run GPU workloads locally or in the cloud

## Facets
- artifact type: library
- maturity: active
- function: video-processing, machine-learning, deep-learning, image-processing, audio-processing
- domain: artificial-intelligence, deep-learning, media
- platform: python
- tags: video-generation, diffusion-models, subject-consistency, audio-driven, image-to-video, hunyuanvideo, customized-video, multimodal, video, linux, gpu

## Member repositories
- Tencent-Hunyuan/HunyuanCustom (main) score 40

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:03.257316+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 2026-08-30T06:15:14.101138+00:00, confidence not recorded.
  - readme: https://github.com/Tencent-Hunyuan/HunyuanCustom (fetched 2026-08-28T04:04:03.257316+00:00, sha a72801180b5c)
  - homepage: https://hunyuancustom.github.io/ (fetched 2026-08-29T12:23:14.164539+00:00, sha e6c497e655f7)
- Data as of 2026-08-30T08:39:29.467469+00:00.