Install and Run FTX-Video on Local Computer with GPU – Free Text to Video Model

In this tutorial, we explain how to install and run the FTX-Video model on a local computer. FTX-Video is a powerful and free to use AI model for generating videos from text descriptions or images.

We are using ComfyUI to run FTX-Video model locally. We are running the model on a computer with NVIDIA 3090 GPU with 24 GB RAM, and with 48GB regular RAM memory, and the Intel i9 processor. It takes around 30-60 seconds to generate a video from a textual description on this computer. Most likely, the FTX-Video model can be executed on lower-end GPUs.

The YouTube tutorial is given below.

Installation of Perquisites

To properly install and use LTX-Video locally, we need a portable version of ComfyUI, 7-zip, Git, and Git support for large files.

Let us start with 7-zip. Go to this website:

https://www.7-zip.org

and download the recent version of 7-zip and install the program. You will need the 7-zip extractor to extract the ComfyUI folder.

To install the portable version of ComfyUI, go to this website

https://github.com/comfyanonymous/ComfyUI/releases

and download the recent 7-zip compressed folder. Decompress the ComfyUI folder and move it to the C drive.

The name of the folder is ComfyUI_windows_portable. Next, try to ComfyUI by navigating to the folder and by executing the file called run_nvidia_gpu.bat

After the file is executed, ComfyUI should be opened in the web browser

The next step is install Git and Git support for large files. To install Git go to this website

https://git-scm.com/downloads/win

and download and install the Git software

Next, download and install Git support for large files from this website

https://git-lfs.com

Installation of LTX-model Files in ComfyUI

First, install ComfyUI-Manager. To do that, download the bat file from this link

 scripts/install-manager-for-portable-version.bat 

and save this file to the base folder of ComfyUI

and execute this file. Next, open ComfyUI and make sure that you can open ComfyUI-Manager:

Next, install the ComfyUI-LTXVideo node. To do that, open the manager, and click on Custom Nodes Manager, and then search for ComfyUI-LTXVideo and click on install

and after that restart

Next install ComfyUI-VideoHelperSuite by repeating the procedure

and restart

Finally, we should have the following nodes:

The next step is to download and install the model checkpoints and files.

Download the file ltx-video-2b-v0.9.safetensors  from

https://huggingface.co/Lightricks/LTX-Video/blob/main/ltx-video-2b-v0.9.safetensors

and place it in the folder

ComfyUI/models/checkpoints

Then, download the file t5xxl_fp16.safetensors

From

https://huggingface.co/Comfy-Org/mochi_preview_repackaged/blob/main/split_files/text_encoders/t5xxl_fp16.safetensors

and put it in the folder called

ComfyUI/models/text_encoders/

Next, download the additional files. Open a command prompt and type

cd\
cd ComfyUI_windows_portable
dir
cd ComfyUI
cd models
cd text_encoders
git clone https://huggingface.co/PixArt-alpha/PixArt-XL-2-1024-MS

The final step is to download the configuration file. Download the json file from this linnk

https://comfyanonymous.github.io/ComfyUI_examples/ltxv/ltxv_text_to_video.json

and save it on your computer. Does not need to be in the ComfyUI folder. Finally, close all the previous versions of ComfyUI (both command prompt and browser), and start ComfyUI again.

Once ComfyUI is opened, click on Workflow and Open and then open the json file.

and then, click on Queue to start the generation of the video