Gun-Shooting LoRA Model
| Property | Value |
|---|---|
| Base Model | Wan2.1 14B I2V 480p |
| Author | Remade-AI |
| Model Type | LoRA for Image-to-Video |
| Training Epochs | 20 |
| Model URL | https://huggingface.co/Remade-AI/Gun-Shooting |
What is Gun-Shooting?
Gun-Shooting is a specialized LoRA (Low-Rank Adaptation) model trained on the Wan2.1 14B I2V 480p base model. It's designed to transform static images into dynamic videos where the subject draws and fires a gun. The model demonstrates remarkable versatility, capable of applying this effect to various subjects, from people to animals, while maintaining consistent quality.
Implementation Details
The model was trained on a carefully curated dataset consisting of 1 minute of video, broken down into 10 distinct clips. Each clip was individually captioned to ensure precise training. The implementation uses a modified version of Kijai's Wan Video Wrapper workflow in ComfyUI, with specific optimizations for gun-shooting animations.
- LoRA Strength: Recommended setting of 1.0
- Embedded Guidance Scale: 6.0
- Flow Shift: 5.0
- Format: Safetensors
Core Capabilities
- Transforms static images into dynamic gun-shooting sequences
- Supports diverse subjects (humans, animals, characters)
- Camera zoom-out effect integration
- Consistent trigger phrase implementation
- 480p resolution output
- ComfyUI workflow compatibility
Frequently Asked Questions
Q: What makes this model unique?
This model stands out for its ability to consistently generate gun-shooting animations from any subject while maintaining visual quality and realistic motion. It's specifically optimized for the Wan2.1 14B I2V framework and includes camera movement effects.
Q: What are the recommended use cases?
The model is designed for creating gun-shooting animations from static images. It works best when following the specific prompt template: 'A [object] looks intently off-camera and the camera zooms out as the [object] p5lls g4un pulls a gun and starts shooting.' Users should replace [object] with their desired subject.
