CivArchive
    【AI video】The right way to use LTX2.3 text-to-video! - v1.0
    NSFW

    Here free to try online first:

     

    Workflow: LTX2.3 Creative Assistant – Automatic Prompt Collection – Text, Image, First-and-Last-Frame Video Generation  

    Experience URL: https://www.runninghub.ai/post/2033139139712917506/?inviteCode=rh-v1401  

     

    Workflow: LTX2.3 – Fully Automatic Prompts – Text-to-Video  

    Experience URL: https://www.runninghub.ai/post/2031218445026594817/?inviteCode=rh-v1401  

     

    Workflow: LTX2.3 – Fully Automatic Prompts – Image-to-Video  

    Experience URL: https://www.runninghub.ai/post/2031218459471777794/?inviteCode=rh-v1401  

     

    Workflow: Free Creation – LTX2.3 – Manual Prompts – Text-to-Video – Designed for use with the LTX2.3 Creative Assistant!  

    Experience URL: https://www.runninghub.ai/post/2034597884850606081/?inviteCode=rh-v1401

     

     

    Hey everyone, this is my LTX2.3 text-to-video workflow—available in both an auto-prompt version (powered by Gemini Flash) and a manual version, designed to help you get the most out of this impressive open-source model.

     

    LTX2.3 is a huge step up from 2.0: Asian faces look much better (the guy in the thumbnail was generated purely with description, no LoRA needed), and lip-sync is now stable—after over 100 tests, I’ve only seen garbled speech once in 15‑second clips. You can even use two character LoRAs to shoot short scenes, making it perfect for mini-dramas.

     

    At 1280×720 resolution on a 24GB GPU, it takes about 15 minutes to generate a 15‑second video, costing roughly $0.035—even factoring in a 50% success rate, it’s still incredibly cheap compared to models like SeeDance ($1/second).

     

    The auto-prompt workflow simplifies everything: just enter the duration, shot count, and a brief idea; Gemini Flash expands it into a high-quality prompt, and a built‑in enhancement node formats it for LTX2.3. If you prefer full control, the manual version lets you iterate freely.

     

    While LTX2.3 still struggles with medium‑to‑long shots, large motions, and lacks directorial reasoning, its character quality and lip‑sync stability make it an amazing tool for creative projects at almost no cost. Give it a try!

    Description

    text to video

    FAQ

    Workflows
    LTXV 2.3

    Details

    Downloads
    301
    Platform
    CivitAI
    Platform Status
    Available
    Created
    3/21/2026
    Updated
    5/27/2026
    Deleted
    -

    Files

    AIVideoTheRightWayToUse_v10.zip

    Mirrors

    HuggingFace (1 mirrors)