LTX-2.3 Foley

LTX-2.3 Foley is a video-to-audio LoRA for LTX-2.3. It adds realistic, visually synchronized Foley and sound effects to video without adding a music overlay.

This LoRA is especially useful when LTX-2.3 adds background music, score, or rhythmic soundtrack material but the desired output is audible real-world sound effects.

Tutorial

Watch the tutorial: using the LTX-2.3 Foley LoRA

Watch the tutorial on YouTube

Demo with and without LoRA

Compatibility

This LoRA is intended for use with LTX-2.3 base or distilled video-to-audio models.

License

This model is released under the LTX Community License configured on the Hugging Face repository.

Files

  • ltx-2.3-foley-400-steps.safetensors: 400-step Foley LoRA weights

Recommended Usage

Use a LoRA multiplier between 1 and 3.

Start around 1. If music or score still appears, increase the multiplier toward 2 or 3.

Prompt with a short description of the visible action in the video, followed by:

No speech is present. No music is present.

Example prompt:

A barista uses an espresso machine to steam milk. No speech is present. No music is present

Recommended negative prompt:

music, melody, song, singing, vocals, score, soundtrack, beat, rhythm bed, instrumental backing, tinny, thin, harsh, clipped, distorted, low bitrate

Examples

Example videos generated with the LoRA:

Door close

Pineapple slicing

Race car

Squash

Comparisons showing LTX-2.3 outputs with and without the Foley LoRA:

Barista comparison

Tennis comparison

ComfyUI Workflow

Use the companion ComfyUI workflow here: FuzzPuppy/LTX-2.3-Foley-Workflow

Downloads last month

-

Downloads are not tracked for this model. How to track
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for FuzzPuppy/LTX-2.3-Foley-LoRA

Adapter
(69)
this model