---
title: HY-OmniWeaving_repackaged
canonical_url: "https://www.modelscope.cn/models/AI-ModelScope/HY-OmniWeaving_repackaged"
md_url: "https://www.modelscope.cn/models/AI-ModelScope/HY-OmniWeaving_repackaged.md"
repository: AI-ModelScope/HY-OmniWeaving_repackaged
last_updated: 2026-04-08
base_model:
  - tencent/HY-OmniWeaving
base_model_relation: finetune
parameters: 18.0B
tensor_type:
  - F32
  - BF16
library_name:
  - safetensors
  - pytorch
frameworks:
  - pytorch
downloads: 96
stars: 1
tags:
  - diffusion-single-file
  - comfyui
---

# HY-OmniWeaving_repackaged

> HY-OmniWeaving_repackaged - AI-ModelScope 在 ModelScope 开源的模型。ComfyUI branch with OmniWeaving support: https://github.com/Comfy-Org/ComfyUI/pull/13289

AI-ModelScope/HY-OmniWeaving_repackaged 是 ModelScope 魔搭社区上的 18.0B 参数机器学习模型，基于 tencent/HY-OmniWeaving 构建。

- **Repository**: AI-ModelScope/HY-OmniWeaving_repackaged
- **Parameters**: 18.0B
- **Base model**: tencent/HY-OmniWeaving
- **Tags**: diffusion-single-file, comfyui
- **Downloads**: 96
- **Stars**: 1
- **Last updated**: 2026-04-08

Source: https://www.modelscope.cn/models/AI-ModelScope/HY-OmniWeaving_repackaged

---

ComfyUI branch with OmniWeaving support: https://github.com/Comfy-Org/ComfyUI/pull/13289

Workflow: drag the animation from PR into ComfyUI

Workflow has two new nodes - HunyuanVideo 15 Omni Conditioning and Text Encode HunyuanVideo 15 Omni, which let you link images and videos as references.

Important setup rule: use the same task on both Text Encode HunyuanVideo 15 Omni and HunyuanVideo 15 Omni Conditioning. The text node changes the system prompt for the selected task, while the conditioning node changes how image/video latents are injected.

It supports the same tasks as shown in their Github - text2vid, img2vid, FFLF, video editing, multi-image references, image+video references (tiv2v) https://github.com/Tencent-Hunyuan/OmniWeaving

Video references are meant to be converted into frames using GetVideoComponents, then linked to Conditioning.