README.md
4.8 KB · 100 lines · markdown Raw
1 ---
2 license: apache-2.0
3 language:
4 - en
5 - zh
6 pipeline_tag: image-to-video
7 tags:
8 - video-generation
9 - text-to-video
10 - image-to-video
11 - video-to-video
12 - reference-to-video
13 - minimax-h3
14 - comfyui
15 - fine-tuned
16 - hdr
17 - singularity
18 base_model:
19 - MiniMaxAI/MiniMax-H3
20 ---
21
22 # Minimax-h3_Singularity
23
24 <p align="center">
25 <a href="https://www.runninghub.ai/post/2096339589492432897/?inviteCode=rh-v1559">
26 <img src="https://img.shields.io/badge/🚀_Online_Demo-RunningHub-blue.svg" alt="Online Demo">
27 </a>
28 <a href="https://youtube.com/@AIGC-Singularity">
29 <img src="https://img.shields.io/badge/YouTube-AIGC--Singularity-red.svg" alt="YouTube">
30 </a>
31 <a href="https://space.bilibili.com">
32 <img src="https://img.shields.io/badge/Bilibili-AIGC--Singularity_Space-00a1d6.svg" alt="Bilibili">
33 </a>
34 </p>
35
36 ## 📖 Model Overview
37
38 **Minimax-h3_Singularity** is a comprehensive fine-tuned fusion model specialized in enhancing the capabilities of **MiniMax-H3**. Designed as a versatile **multimodal video generation model**, it natively supports **Text-to-Video (T2V)**, **Image-to-Video (I2V)**, **Reference-to-Video (Ref2V)**, and **Video-to-Video (V2V)** workflows within **ComfyUI**.
39
40 Built upon a strategic fusion of key checkpoints (including `ref`, `fl`, `b25-49`, etc.), this model underwent deep high-step fine-tuning. To preserve the original model's foundational strengths and broad generalization while solving artifacts introduced by high-step training, we spent **3 full days on precise model pruning and weight optimization**. The result is a clean, sharp, and highly dynamic video generation model.
41
42 ---
43
44 ## ✨ Key Improvements & Features
45
46 * 🎬 **HDR Image Quality & Blur Reduction**: Fine-tuned on high-dynamic-range (HDR) video datasets to significantly enhance visual clarity and eliminate motion blur during high-speed action.
47 * 👤 **Distant Face Restoration**: Drastically reduces facial distortion, blurriness, and collapsing in medium-to-long shots.
48 * 🎨 **Clean & De-Oiled Aesthetic**: Removes heavy, unnatural skin shine and glossy textures, rendering natural lighting and photorealistic materials.
49 * ⚔️ **Enhanced Dynamic Motion**: Boosts motion fluidity and physical impact, excels in complex action sequences such as **sword fighting and martial arts/melee combat**.
50 * 🌌 **VFX & Fantasy Effects**: Specifically optimized for fantasy spellcasting, particle aura, and magical combat visual effects.
51 * 🎭 **Expressive Facial Dynamics**: Captures subtle facial expressions and emotional nuances more vividly.
52 * 📹 **Cinematography & Camera Control**: Strengthens responsiveness to camera movements (pan, tilt, zoom, tracking shots) for cinematic storytelling.
53 * 🛡️ **Full Base Capability Retention**: 100% preserves MiniMax-H3's original prompt adherence, style adaptability, and base multimodal generation strength.
54
55 ---
56
57 ## 🎬 Showcase
58
59 <video src="https://huggingface.co/WarmBloodAban/Minimax-h3_Singularity/resolve/main/video/1.mp4" controls autoplay loop muted width="20%"></video>
60 <video src="https://huggingface.co/WarmBloodAban/Minimax-h3_Singularity/resolve/main/video/3.mp4" controls autoplay loop muted width="20%"></video>
61 <video src="https://huggingface.co/WarmBloodAban/Minimax-h3_Singularity/resolve/main/video/AIGCTYD2_00001_p84-audio_ganvc_1788634594%20(1).mp4" controls autoplay loop muted width="20%"></video>
62 <video src="https://huggingface.co/WarmBloodAban/Minimax-h3_Singularity/resolve/main/video/AIGCTYD2_00010_p87-audio_alipa_1788652671.mp4" controls autoplay loop muted width="20%"></video>
63
64 ---
65
66 ## 💡 Usage Guide
67
68 ### Multimodal Pipeline Support
69 This model is fully compatible with **ComfyUI** and supports:
70 * **Text-to-Video (T2V)**
71 * **Image-to-Video (I2V)**
72 * **Reference-to-Video (Ref2V)**
73 * **Video-to-Video (V2V)**
74
75 ### 🚀 Recommended Acceleration LoRA
76 For high-speed generation with minimal quality loss, we strongly recommend pairing with:
77 * **`minimax_h3_ref2v_turbo_4step_v0.1`** (Enables 4-step fast inference)
78
79 ### 🌐 Online Interactive Demo
80 Test the model directly in your browser without local GPU setup:
81 👉 [**Try it on RunningHub Workflows**](https://www.runninghub.ai/post/2096339589492432897/?inviteCode=rh-v1559)
82
83 ---
84
85 ## 🙏 Acknowledgements
86
87 Special thanks to the **MiniMax** open-source team for creating and releasing the powerful `MiniMax-H3` multimodal video model, providing a solid foundation for the open-source community! 🤝
88
89 ---
90
91 ## 🤝 Community & Commercial Inquiries
92
93 Feel free to connect for tutorials, community discussions, workflow sharing, or commercial collaborations:
94
95 * **YouTube Channel**: [AIGC-Singularity](https://youtube.com/@AIGC-Singularity)
96 * **Bilibili Channel**: AIGC-Singularity Space
97 * **QQ Group 1**: `1058747239` (Request to join)
98 * **QQ Group 2**: `1072010342` (Request to join)
99 * **Business Inquiries (WeChat)**: `aigctyd`
100 * **Email**: `a592991299@gmail.com`