LTX-Video 2.3 — Simple Workflow with Metadata
A clean and user-friendly ComfyUI workflow for LTX-Video 2.3, designed for image-to-video generation with audio while keeping the workflow simple and easy to use.
The workflow gives you control over the most important settings without turning the graph into a complicated mess.
✨ Features
🎬 Image → Video + Audio generation
📐 Automatic aspect-ratio handling — simply set the long side of the video and the other dimension is calculated automatically
↔️ Works with both horizontal and vertical videos
⏱️ Easy video duration control in seconds
🔊 Audio volume control
0— original volumeNegative values — quieter
Positive values — louder
🪜 Adjustable sampling steps
🎲 Full seed control
💾 Automatic metadata embedding into the final video
🔍 First-pass preview for quickly checking the result
🚀 Final upscaled output
🎨 Full LoRA support through Lora Manager
🌐 Models and LoRAs can be loaded directly from Civitai
🤖 Model
This workflow was tested with the distilled INT8 10EROS 1.4 model:
However, the workflow is not tied to this model.
You can use:
Other LTX-Video models
Distilled models
🧩 Required Custom Nodes
Most required custom nodes can be installed through ComfyUI Manager.
There is one exception:
⚠️ ComfyUI Image Saver
If you get an error related to:
Image Saver — Add Metadata to VHS Video
you may need to install ComfyUI-Image-Saver manually.
Open a terminal in your ComfyUI installation and run:
cd custom_nodes
git clone [email protected]:alexopus/ComfyUI-Image-Saver.git
cd ComfyUI-Image-Saver
pip install -r requirements.txt
Then restart ComfyUI.
🎨 LoRA Manager
The workflow fully supports Lora Manager, providing a workflow experience similar to Automatic1111.
You can browse your LoRA library, select a LoRA and adjust its strength directly from the workflow.
Quick Lora Manager setup
Install Lora Manager.
Restart ComfyUI.
A new Lora Manager icon will appear in ComfyUI.
Open the manager in its separate tab.
Press Refresh to scan your LoRA folders.
Your available LoRAs should then appear in the manager.
If you use LoRAs from Civitai, click the Fetch icon in Lora Manager to retrieve and update the available LoRA information from Civitai.
To add a LoRA from the Lora Manager window, simply click the paper airplane icon, then select node #1045.
You can also add a LoRA directly inside the node itself — just start typing the LoRA name or folder name, and select it from the list.
This makes it easy to experiment with different LoRAs without manually editing paths inside the workflow.
💾 Metadata + Civitai
One of the main features of this workflow is metadata preservation.
The final video can contain information such as:
Prompt
Negative prompt
Seed
Model (from Civitai)
LoRA information (from Civitai)
Generation parameters
Other workflow metadata
This makes it much easier to reproduce and share your generations.
Civitai Video Metadata Assistant
To make use of the embedded metadata on Civitai, install the following browser extension:
https://github.com/diodiogod/Civitai-Video-Metadata-Assistant
After installation, the assistant will appear on the Civitai video upload page.
It can read the metadata embedded in your generated video and automatically detect relevant information such as your LoRAs and generation data, making uploads much easier.
🚀 How to use
The workflow is intentionally simple:
1. Select your model
2. Select your LoRA(s), if needed
3. Load your source image
4. Write your prompt
5. Set the video length
6. Set the long-side resolution
7. Adjust steps and audio volume
8. Generate
That's it.
The workflow handles the rest, including aspect-ratio calculation, preview generation, upscaling, audio and metadata.
❤️ Why this workflow?
LTX-Video workflows can quickly become complicated, especially when combining image-to-video, audio, upscaling, LoRAs and metadata.
This workflow is designed around a simple idea:
Keep the interface simple, while keeping the important controls accessible.
You don't need to dig through dozens of nodes just to change the resolution, duration, seed or LoRA strength.
Set the parameters you actually care about and generate.
📌 Notes
This workflow is provided as a general-purpose LTX-Video 2.3 workflow.
Your results will depend on the model, LoRAs, source image, prompt, resolution and sampling settings.
If you use a different model or LoRA setup, some parameters may require adjustment.