Editor's Note: ComfyUI-LTXVideo is the official custom node pack from Lightricks for running LTX-2.3 models inside ComfyUI. This guide covers installation via ComfyUI Manager, required models, available workflows, and advanced features like IC-LoRA control and HDR output — everything you need to get started with LTX 2.3 video generation.

ComfyUI-LTXVideo Nodes: Complete Installation and Usage Guide

A collection of powerful custom nodes that extend ComfyUI's capabilities for the LTX-2 video generation model.

LTX-2 is built into ComfyUI core, making it readily accessible to all ComfyUI users. This repository hosts additional nodes and workflows to help you get the most out of LTX-2's advanced features.

Prerequisites

Before you begin using an LTX-2 workflow in ComfyUI, make sure you have:

• ComfyUI installed
• CUDA-compatible GPU with 32GB+ VRAM
• 100GB+ free disk space for models and cache

Installation via ComfyUI Manager

The recommended way to install ComfyUI-LTXVideo is through ComfyUI Manager:

1. Open ComfyUI
2. Click the Manager button (or press Ctrl+M)
3. Select Install Custom Nodes
4. Search for "LTXVideo"
5. Click Install
6. Wait for installation to complete
7. Restart ComfyUI

The nodes will appear in your node menu under the LTXVideo category. Required models will be downloaded on first use.

Required Models

Download the following models to the appropriate folders in your ComfyUI installation:

LTX-2.3 Model Checkpoint

Choose one and place it in COMFYUI_ROOT_FOLDER/models/checkpoints:

• ltx-2.3-22b-dev.safetensors
• ltx-2.3-22b-distilled-1.1.safetensors

Spatial Upscaler

Required for two-stage pipeline workflows. Place in COMFYUI_ROOT_FOLDER/models/latent_upscale_models:

• ltx-2.3-spatial-upscaler-x2-1.1.safetensors
• ltx-2.3-spatial-upscaler-x1.5-1.0.safetensors

Temporal Upscaler

Required for two-stage pipeline workflows. Place in COMFYUI_ROOT_FOLDER/models/latent_upscale_models:

• ltx-2.3-temporal-upscaler-x2-1.0.safetensors

Distilled LoRA

Required for two-stage pipeline implementations. Place in COMFYUI_ROOT_FOLDER/models/loras:

• ltx-2.3-22b-distilled-lora-384-1.1.safetensors

Gemma Text Encoder

Download all files to COMFYUI_ROOT_FOLDER/models/text_encoders/gemma-3-12b-it-qat-q4_0-unquantized:

• Gemma 3 (google/gemma-3-12b-it-qat-q4_0-unquantized on HuggingFace)

IC-LoRA Models (Optional)

Place in COMFYUI_ROOT_FOLDER/models/loras:

• ltx-2.3-22b-ic-lora-union-control-ref0.5.safetensors — depth + edge control
• ltx-2.3-22b-ic-lora-motion-track-control-ref0.5.safetensors — motion tracking
• ltx-2.3-22b-ic-lora-hdr-0.9.safetensors — HDR output
• ltx-2-19b-ic-lora-detailer.safetensors — video detailing
• ltx-2-19b-ic-lora-pose-control.safetensors — human pose control

Camera Control LoRAs (Optional)

A set of camera movement LoRAs is available for dolly in/out/left/right, jib up/down, and static shots.

Example Workflows

After installation, example workflows are available at:

ComfyUI/custom_nodes/ComfyUI-LTXVideo/example_workflows/

LTX-2.3 Workflows

• Text/Image to Video (Single Stage) — Full and distilled model, single stage generation
• Text/Image to Video (Two Stage with Upsampling) — Distilled model with spatial/temporal upscaling
• IC-LoRA Union Control — Depth + human pose + edges control
• IC-LoRA Motion Tracking — I2V with motion tracking
• IC-LoRA HDR — High dynamic range video output

LTX-2.0 Workflows (Legacy)

• Text to video (full and distilled)
• Image to video (full and distilled)
• Video to video detailer
• IC-LoRA with depth, pose, and edge control

Union IC-LoRA

The Union IC-LoRA model combines depth and edge (canny) control conditions into a single unified LoRA.

Key features:

• Unified Control: One LoRA supports multiple control conditions (depth or edges)
• Downsampled Latent Processing: Operates on downsampled latent size, reducing memory usage and speeding up inference while maintaining quality

The union LoRA learns to parse multiple conditions and identify which control signals are present in the input, then processes them at reduced resolution for efficiency.

HDR IC-LoRA

The HDR IC-LoRA generates linear HDR video encoded in ARRI LogC3, enabling workflows that output high-dynamic-range content suitable for grading and EXR export.

Key features:

• Linear HDR output: Produces frames in LogC3-compressed space; the LTXVHDRDecodePostprocess node decodes these back to linear HDR values
• SDR preview + raw HDR: Outputs both a Reinhard-tonemapped SDR preview and the raw linear HDR tensor
• EXR export: Set OPENCV_IO_ENABLE_OPENEXR=1 in your environment before starting ComfyUI to enable EXR sequence export

Low VRAM Tips

For systems with limited VRAM:

• Use the model loader nodes from low_vram_loaders.py — these ensure correct execution order and perform model offloading to fit generation within 32GB VRAM
• Use the --reserve-vram ComfyUI parameter: python -m main --reserve-vram 5 (adjust the number in GB as needed)

Sources

• ComfyUI-LTXVideo — Official Repository
• LTX-2 Main Repository
• LTX-2.3 Models on HuggingFace
• Open Source Documentation