Downloads · 30 days
0
dbz-fight-style-lora
Bisre/dbz-fight-style-lora
dbz-fight-style-lora is a machine learning model from Bisre. Use it for the machine learning task on the model card, and read the license before you ship it in a product.
This repository contains a 60 epoch lora training to adapt the fight scene motion of the famous dragon ball anime series. The Lora was trained on 60, 5 second videos from a recent dragon ball movie, using the WAN2.1 1…
Access
Public
Updated Sep 16, 2025
Repo size
737 MB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors359 MB · 95%
At a glance
- Access
- Public
- Created
- Jul 30, 2025
- Updated
- Sep 16, 2025
- SHA
- 2cfa9658
Model card
From the Hugging Face model README
Dragon Ball Fight Scene Adaptation.
This repository contains a 60 epoch lora training to adapt the fight scene motion of the famous dragon ball anime series. The Lora was trained on 60, 5 second videos from a recent dragon ball movie, using the WAN2.1 14B I2V model as a base.
How to Use with
- Download the models you want and load into comfyUI.
- Load the Wan 2.1 14B I2v model.
- Genereate videos using the dbz_fight_style, Trigger word and then writing your prompt.
Training Details
Base Model: The LoRAs were trained on top of the WAN2.1 14B parameter I2V model. Training Library: The training was implemented using the diffusionpipe library. Hyperparameters: For a detailed look at the training configuration and hyperparameters, please visit the training repository here.
Dataset
You can find the dataset used here
Demo video
<img src="https://huggingface.co/Bisre/dbz-fight-style-lora/resolve/main/1_with_lora.webp" width="512" height="512">- More videos can be found here.
- Created
- Jul 30, 2025
- Updated
- Sep 16, 2025