Crash-zoom-out Open-source Model - Free Implementation of the Zoom-out Effect, Bringing a Sense of Scale, Surprise, and Disorientation

Crash Zoom Out

Developed by Remade-AI

A LoRA trained on the Wan2.1 14B I2V 480p model to achieve a sudden zoom-out effect from the subject to reveal the surrounding scene, creating an abrupt sense of scale, surprise, or disorientation.

Video Processing EnglishOpen Source License:Apache-2.0 #Crash zoom out effect #Dramatic scene reveal #LoRA fine-tuned model

Downloads 152

Release Time : 5/24/2025

Model Overview

This LoRA model is specifically designed for image-to-video tasks, utilizing the crash zoom out effect to achieve dramatic or comedic scene reveals, suitable for video generation that requires quick presentation of the relationship between the environment and the subject.

Model Features

Rapid zoom out effect

Capable of quickly zooming out from a subject close-up to reveal the surrounding environment, creating dramatic effects

Multi-scene applicability

Consistent performance across different object types and scenes, suitable for various themes

Simple prompt structure

Clear and straightforward prompt structure, easy to adjust and optimize

Compatibility with base model

Trained on the Wan2.1 14B 480p I2V base model, maintaining high-quality output

Model Capabilities

Image-to-video generation

Special effects shot control

Rapid scene transition

Dramatic effect implementation

Use Cases

Film and TV special effects

Suspense scene reveal

Quickly zoom out from a character close-up to reveal a dangerous environment, enhancing tension

Example video shows a gunman appearing behind a tied-up man

Stunning environment display

Zoom out from a character close-up to reveal spectacular natural scenery

Example video shows a snow-covered seaside landscape

Creative expression

Artistic atmosphere creation

Create a mysterious artistic atmosphere through rapid shot transitions

Example video shows a tattooed woman in a candlelit room

🚀 Crash zoom out LoRA for Wan2.1 14B I2V 480p

This LoRA is designed for the Wan2.1 14B I2V 480p model, enabling an abrupt zoom - out effect to create a sense of scale, surprise, or disorientation.

🚀 Quick Start

Example Prompts and Outputs

Prompt 1: The video begins with a close - up on a man's face, his hands tied with rope, and an anxious expression. Then, a cr34sh crash zoom out effect reveals a dark and obscure room, the man is still tied up and two men wearing balaklavas and holding guns appear to be standing behind him.
- Output: example_videos/1.mp4
Prompt 2: The video begins with a close - up on the man's face, with ice covering his beard and eyelashes. He has a concerned or startled expression, his eyes are a vivid blue. A cr34sh crash zoom out effect rapidly pulls the camera back, revealing the man in a yellow jacket set in a icy landscape. The cr34sh crash zoom out effect shows his position: standing on the edge of the sea with icebergs in the background.
- Output: example_videos/2.mp4
Prompt 3: The video begins with a close - up shot of a woman's face with intricate black and white tribal markings on her face, neck, and chest. Her eyes are closed and she is wearing dark red eyeshadow and lipstick. The cr34sh crash zoom out effect then begins, quickly pulling back to reveal that the woman is in a dimly lit room, with candles all around her.
- Output: example_videos/3.mp4

✨ Features

Trained on Specific Model: Trained on the Wan2.1 14B 480p I2V base model.
Consistent Results: Provides consistent results across different object types.
Simple Prompt Structure: Has a simple prompt structure that's easy to adapt.

📦 Installation

📋 Download Links

crash_zoom_out.safetensors - LoRA Model File
wan_img2vid_lora_workflow.json - Wan I2V with LoRA Workflow for ComfyUI

💻 Usage Examples

Recommended Settings

LoRA Strength: 1.0
Embedded Guidance Scale: 6.0
Flow Shift: 5.0

Trigger Words

The key trigger phrase is: cr34sh crash zoom out effect

Prompt Template

For prompting, check out the example prompts; this way of prompting seems to work very well.

ComfyUI Workflow

This LoRA works with a modified version of Kijai's Wan Video Wrapper workflow. The main modification is adding a Wan LoRA node connected to the base model.

Workflow Screenshot

See the Downloads section above for the modified workflow.

📚 Documentation

Model Information

The model weights are available in Safetensors format. See the Downloads section above.

Training Details

Property	Details
Base Model	Wan2.1 14B I2V 480p
Training Data	Trained on 50 seconds of video comprised of 10 short clips (each clip captioned separately) of scenes that used the crash zoom out camera motion.
Epochs	25

Additional Information

Training was done using Diffusion Pipe for Training

Acknowledgments

Special thanks to Kijai for the ComfyUI Wan Video Wrapper and tdrussell for the training scripts!

🔧 Technical Details

The LoRA is trained on the Wan2.1 14B I2V 480p model. The training data consists of 50 - second video clips with the crash zoom - out camera motion. After 25 epochs of training, it can achieve consistent results with a simple prompt structure.

📄 License

This project is licensed under the Apache - 2.0 license.

Featured Recommended AI Models

Empowering the Future, Your AI Solution Knowledge Base

English 简体中文繁體中文にほんご