LTX-2 Text to Video (T2V) in ComfyUI: AI Video Generation

About This Workflow

1. Generate AI Videos Directly from Text

LTX-2 T2V turns simple text prompts into full AI videos—no images or footage required.

2. Create Realistic Human Motion and Lip-Sync

Generate talking, singing, or expressive characters with accurate lip-sync, facial expressions, and natural body movement.

3. Support for Cinematic Camera Control

Use camera prompts and LoRAs to add smooth pan, dolly, zoom, and cinematic motion to your text-generated videos.

4. Works for Many Video Styles

Perfect for cartoons, ASMR scenes, realistic people, 3D toy renders, and cinematic storytelling—all from text.

5. Easy Local Workflow in ComfyUI

Runs fully inside ComfyUI with local models, LoRA support, upscaling, and no cloud limits—ideal for creators and AI filmmakers.

Required Models

Model links

checkpoints

• ltx-2-19b-dev.safetensors
• ltx-2-19b-dev-fp8.safetensors

text_encoders

• gemma_3_12B_it.safetensors

loras

• ltx-2-19b-distilled-lora-384.safetensors
• ltx-2-19b-lora-camera-control-dolly-left.safetensors

latent_upscale_models

• ltx-2-spatial-upscaler-x2-1.0.safetensors

Model Storage Location

📂 ComfyUI/
├── 📂 models/
│   ├── 📂 checkpoints/
│   │      ├── ltx-2-19b-dev.safetensors
│   │      └── ltx-2-19b-dev-fp8.safetensors
│   ├── 📂 text_encoders/
│   │      └── gemma_3_12B_it.safetensors
│   ├── 📂 loras/
│   │      ├── ltx-2-19b-distilled-lora-384.safetensors
│   │      └── ltx-2-19b-lora-camera-control-dolly-left.safetensors
│   └── 📂 latent_upscale_models/
│          └── ltx-2-spatial-upscaler-x2-1.0.safetensors

Workflow Graph