LTX-2.3: ComfyUI Workflow Examples

Learn how to use LTX-2.3 in ComfyUI with native workflows for text-to-video, image-to-video, FLF2V, audio-driven video, IC-LoRA control, and ID-LoRA personalization.

LTX-2.3 is the latest evolution of Lightricks’ open-source audio-video generation model, now natively supported in ComfyUI. Building on LTX-2, this release delivers major quality improvements across fine details, portrait video, audio quality, image-to-video consistency, prompt understanding, and text rendering.

ComfyUI ships with six native workflows for LTX-2.3, each designed for a specific generation mode:

• Text-to-Video (T2V): Generate videos from text prompts
• Image-to-Video (I2V): Generate videos from an input image
• FLF2V: Interpolate between a start and end image
• Image-Audio-to-Video (IA2V): Generate lip-synced videos from an image and audio
• IC-LoRA Union Control: Control video generation with depth, pose, or edge guidance
• ID-LoRA: Generate personalized videos with synchronized audio from a reference image and audio clip

Local users

Make sure your ComfyUI is updated.

• Download ComfyUI
• Update Guide

Workflows in this guide can be found in the Workflow Templates. If you can’t find them in the template, your ComfyUI may be outdated.

If nodes are missing when loading a workflow, possible reasons:

1. You are not using the latest ComfyUI version (Nightly version)
2. Some nodes failed to import at startup

Cloud users

• Cloud will update after ComfyUI stable release.

So, if you find any core node missing in this document, it might be because the new core nodes have not yet been released in the latest stable version. Please wait for the next stable release.

Key features

• Finer details: New latent space and updated VAE for sharper textures, cleaner edges, and more precise visuals
• 9:16 Portrait support: Greatly improved quality for vertical portrait videos
• Better audio: Cleaner sound with reduced noise and enhanced dialogue
• Improved image-to-video: More consistent motion and fewer glitches
• Smarter prompt understanding: Improved text encoder for more accurate interpretation
• Native ComfyUI support: All workflows are built-in, no custom nodes required

Getting started

LTX-2.3 is natively supported in ComfyUI. To get started:

1. Update ComfyUI to the latest version
2. Go to Template Library > Video > choose any LTX-2.3 workflow
3. Follow the pop-up to download models and run the workflow

ComfyUI Native Workflows

LTX-2.3 Text to Video (T2V)

Generate videos from text prompts with improved prompt understanding and text rendering.

LTX-2.3 Text to Video

Generate videos from text prompts with improved prompt understanding and text rendering.

Run on Comfy Cloud

Open in Comfy Cloud

Download Workflow

Download JSON or search “LTX-2.3 T2V” in Template Library

Model downloads

Checkpoint: ltx-2.3-22b-dev-fp8

Place in ComfyUI/models/checkpoints/

LoRA: distilled 1.1

Place in ComfyUI/models/loras/

LoRA: gemma abliterated

Place in ComfyUI/models/loras/

Text Encoder: gemma 12B

Place in ComfyUI/models/text_encoders/

Upscaler: spatial x2

Place in ComfyUI/models/latent_upscale_models/

Model storage

ComfyUI/
├── 📂 models/
│   ├── 📂 checkpoints/
│   │   └── ltx-2.3-22b-dev-fp8.safetensors
│   ├── 📂 loras/
│   │   ├── ltx_2.3_22b_distilled_1.1_lora_dynamic_fro09_avg_rank_111_bf16.safetensors
│   │   └── gemma-3-12b-it-abliterated_lora_rank64_bf16.safetensors
│   ├── 📂 text_encoders/
│   │   └── gemma_3_12B_it_fp4_mixed.safetensors
│   └── 📂 latent_upscale_models/
│       └── ltx-2.3-spatial-upscaler-x2-1.1.safetensors

Prompting tips

1. Core Actions: Describe events and actions as they occur over time
2. Visual Details: Describe all visual details you want to appear in the video
3. Audio: Describe sounds and dialogue needed for the scene

---

LTX-2.3 Image to Video (I2V)

Generate videos from an input image with more consistent motion and smoother animations.

LTX-2.3 Image to Video

Generate videos from an input image with improved motion consistency.

Run on Comfy Cloud

Open in Comfy Cloud

Download Workflow

Download JSON or search “LTX-2.3 I2V” in Template Library

Input Image: egyptian_queen.png

Download the default input image, or use your own image.

Model downloads

The I2V workflow uses the same model set as Text-to-Video.

Checkpoint: ltx-2.3-22b-dev-fp8

Place in ComfyUI/models/checkpoints/

LoRA: distilled 1.1

Place in ComfyUI/models/loras/

LoRA: gemma abliterated

Place in ComfyUI/models/loras/

Text Encoder: gemma 12B

Place in ComfyUI/models/text_encoders/

Upscaler: spatial x2

Place in ComfyUI/models/latent_upscale_models/

Model storage

ComfyUI/
├── 📂 models/
│   ├── 📂 checkpoints/
│   │   └── ltx-2.3-22b-dev-fp8.safetensors
│   ├── 📂 loras/
│   │   ├── ltx_2.3_22b_distilled_1.1_lora_dynamic_fro09_avg_rank_111_bf16.safetensors
│   │   └── gemma-3-12b-it-abliterated_lora_rank64_bf16.safetensors
│   ├── 📂 text_encoders/
│   │   └── gemma_3_12B_it_fp4_mixed.safetensors
│   └── 📂 latent_upscale_models/
│       └── ltx-2.3-spatial-upscaler-x2-1.1.safetensors

---

LTX-2.3 FLF2V

Interpolate between a start image and an end image to generate a smooth video transition.

LTX-2.3 First-Last Frame to Video

Interpolate between a start and end image.

Run on Comfy Cloud

Open in Comfy Cloud

Download Workflow

Download JSON or search “LTX-2.3 FLF2V” in Template Library

Start Image: high_view_classic_car.png

First frame of the video.

End Image: low_view_classic_car.png

Last frame of the video.

Model downloads

The FLF2V workflow uses the distilled checkpoint instead of the dev checkpoint.

Checkpoint: ltx-2.3-22b-distilled-fp8

Place in ComfyUI/models/checkpoints/

Text Encoder: gemma 12B

Place in ComfyUI/models/text_encoders/

Model storage

ComfyUI/
├── 📂 models/
│   ├── 📂 checkpoints/
│   │   └── ltx-2.3-22b-distilled-fp8.safetensors
│   └── 📂 text_encoders/
│       └── gemma_3_12B_it_fp4_mixed.safetensors

---

LTX-2.3 Image Audio to Video (IA2V)

Upload an image and an audio file to generate a high-quality video with synchronized lip movements.

LTX-2.3 Image and Audio to Video

Generate lip-synced videos from an image and audio.

Run on Comfy Cloud

Open in Comfy Cloud

Download Workflow

Download JSON or search “LTX-2.3 IA2V” in Template Library

Input Image: cactus_man.png

Reference image for the speaking character.

Reference Audio: ltx_23_audio.mp3

Audio track for lip synchronization.

Model downloads

The IA2V workflow uses the same model set as Text-to-Video.

Checkpoint: ltx-2.3-22b-dev-fp8

Place in ComfyUI/models/checkpoints/

LoRA: distilled 1.1

Place in ComfyUI/models/loras/

LoRA: gemma abliterated

Place in ComfyUI/models/loras/

Text Encoder: gemma 12B

Place in ComfyUI/models/text_encoders/

Upscaler: spatial x2

Place in ComfyUI/models/latent_upscale_models/

Model storage

ComfyUI/
├── 📂 models/
│   ├── 📂 checkpoints/
│   │   └── ltx-2.3-22b-dev-fp8.safetensors
│   ├── 📂 loras/
│   │   ├── ltx_2.3_22b_distilled_1.1_lora_dynamic_fro09_avg_rank_111_bf16.safetensors
│   │   └── gemma-3-12b-it-abliterated_lora_rank64_bf16.safetensors
│   ├── 📂 text_encoders/
│   │   └── gemma_3_12B_it_fp4_mixed.safetensors
│   └── 📂 latent_upscale_models/
│       └── ltx-2.3-spatial-upscaler-x2-1.1.safetensors

---

LTX-2.3 IC-LoRA Union Control

Generate LTX-2.3 videos with IC-LoRA using aligned control inputs like depth, pose, or edges. IC-LoRA is an in-context LoRA trained on top of LTX-2.3 that applies structural guidance from a reference video. It accepts control signals from various preprocessors including depth maps, Canny edges, and pose skeletons. This workflow uses a subgraph for modular processing.

Run on Comfy Cloud

Open in Comfy Cloud

Download Workflow

Download JSON or search “LTX-2.3 IC-LoRA” in Template Library

Learn about Subgraph

This workflow uses a Subgraph node for modular processing. Check out the Subgraph documentation to learn how to customize and extend the workflow.

Input materials

Control Video: stone_ruins.mp4

Driving video for control signal extraction.

Reference Image: the_forgotten_gate.png

Reference image for style and content.

Model downloads

Checkpoint: ltx-2.3-22b-distilled-fp8

Place in ComfyUI/models/checkpoints/

LoRA: IC-LoRA Union Control

Place in ComfyUI/models/loras/

LoRA: gemma abliterated

Place in ComfyUI/models/loras/

Text Encoder: gemma 12B

Place in ComfyUI/models/text_encoders/

Geometry: MoGe Normal

Place in ComfyUI/models/geometry_estimation/

Model storage

ComfyUI/
├── 📂 models/
│   ├── 📂 checkpoints/
│   │   └── ltx-2.3-22b-distilled-fp8.safetensors
│   ├── 📂 loras/
│   │   ├── ltx-2.3-22b-ic-lora-union-control-ref0.5.safetensors
│   │   └── gemma-3-12b-it-abliterated_lora_rank64_bf16.safetensors
│   ├── 📂 text_encoders/
│   │   └── gemma_3_12B_it_fp4_mixed.safetensors
│   └── 📂 geometry_estimation/
│       └── moge_2_vitl_normal_fp16.safetensors

---

LTX-2.3 ID-LoRA

Generate personalized videos with synchronized audio from a text prompt, reference image, and short audio clip. Uses ID-LoRA to adapt a person’s appearance and voice in a single generative model.

LTX-2.3 ID LoRA

Generate personalized videos with synchronized audio from a reference image and audio clip.

Run on Comfy Cloud

Open in Comfy Cloud

Download Workflow

Download JSON or search “LTX-2.3 ID-LoRA” in Template Library

Reference Image: vintage_thinker.png

Reference image for the character appearance.

Reference Audio: ltx23_reference_audio.mp3

Audio clip for voice cloning and lip sync.

Model downloads

Checkpoint: ltx-2.3-22b-dev-fp8

Place in ComfyUI/models/checkpoints/

LoRA: distilled 1.1

Place in ComfyUI/models/loras/

LoRA: ID-LoRA TalkVid

Place in ComfyUI/models/loras/

Text Encoder: gemma 12B

Place in ComfyUI/models/text_encoders/

Upscaler: spatial x2

Place in ComfyUI/models/latent_upscale_models/

Model storage

ComfyUI/
├── 📂 models/
│   ├── 📂 checkpoints/
│   │   └── ltx-2.3-22b-dev-fp8.safetensors
│   ├── 📂 loras/
│   │   ├── ltx_2.3_22b_distilled_1.1_lora_dynamic_fro09_avg_rank_111_bf16.safetensors
│   │   └── ltx-2.3-id-lora-talkvid-3k.safetensors
│   ├── 📂 text_encoders/
│   │   └── gemma_3_12B_it_fp4_mixed.safetensors
│   └── 📂 latent_upscale_models/
│       └── ltx-2.3-spatial-upscaler-x2-1.1.safetensors

Prompting tips

When writing prompts for LTX-2.3, focus on detailed, chronological descriptions of actions and scenes:

1. Core Actions: Describe events and actions as they occur over time
2. Visual Details: Describe all visual details you want to appear in the video
3. Audio: Describe sounds and dialogue needed for the scene

Resources

• Hugging Face Model
• GitHub Repository
• ComfyUI-LTXVideo
• IC-LoRA Model