ltx-m-v2 / README.md
romanbwrk's picture
Implement LTX-2.3 image-to-video Gradio Space
e4e75f3
|
Raw
History Blame Contribute Delete
1.01 kB
---
title: Ltx M V2
emoji: 🔥
colorFrom: purple
colorTo: purple
sdk: gradio
sdk_version: 6.19.0
python_version: '3.12'
app_file: app.py
pinned: false
---
Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
# LTX-2.3 Image → Video
Upload an image and a prompt to generate ~5 seconds of video (with audio) using
the **LTX-2.3 22B distilled** model via `diffusers`. This is a native Gradio
reimplementation of the model stack used by the WhatDreamsCost "LTX Director 2"
ComfyUI workflow.
- **Model:** `diffusers/LTX-2.3-Distilled-Diffusers` (8 steps, CFG 1)
- **Output:** 121 frames @ 24 fps, with audio
- **2× upscale:** optional toggle, off by default (slower; may exceed the
ZeroGPU per-call time budget)
## Hardware
Requires a **ZeroGPU (H200)** Space — set this in the Space's *Settings →
Hardware*. The 22B model uses `enable_model_cpu_offload()` to fit in 70 GB.
Design doc: `docs/superpowers/specs/2026-06-25-ltx-image-to-video-space-design.md`