metadata

datasets:
  - Lightricks/Squish-Dataset
base_model:
  - Lightricks/LTX-Video-0.9.5
  - Lightricks/LTX-Video

Squish Effect LoRA for LTX Video

This repository contains a LoRA model trained on top of LTX Video v0.9.5 to create "squish"-style video effects.

Usage

For best results, use prompts with the "SQUISH" trigger word in the following format:

SQUISH two hands squeezing a squeezable object that is shaped like [your object]

A reference ComfyUI workflow is provided in assets/ltxv-i2v-lora.json. You can import this workflow directly into ComfyUI to get started.

Example input images that work well with this model can be found in the assets/images directory.

This LoRA is designed to work with LTX Video v0.9.5. Make sure you have the base model loaded in ComfyUI before using this LoRA.

The model was trained using the Squish Dataset, which is publicly available on Hugging Face.

Please refer to the license of the base LTX Video model for usage terms and conditions.