CivArchive
    LTXV2.3 GGUF I2V/T2V/FFLF+IDaudioLora 12gb Vram sageattn - v1.5
    NSFW

    A simple, color-coded GGUF workflow for running LTXV-2 in ComfyUI.
    Designed to be straightforward and lightweight.

    Common Issues

    Error: LTXVEmptyLatentAudio – AttributeError

    Update Kijai Node Pack to v1.2.8

    Checkerboard / patterned noise output

    Update all nodes + ComfyUI
    (Outdated GGUF loader or ComfyUI build)


    Required Node Packs

    ComfyUI-GGUF
    https://github.com/city96/ComfyUI-GGUF

    ComfyUI-KJNodes
    https://github.com/kijai/ComfyUI-KJNodes (install & update it to nightly)

    additional install via manager


    📦 Downloads

    🧩 Kijai’s Repo (ALL files except diffusion models — choose ONE LoRA)

    https://huggingface.co/Kijai/LTX2.3_comfy/tree/main

    Includes:

    • Distilled LoRA (REQUIRED)

    • Text Projection (embedding connector)


    🧠 Diffusion Models (GGUF)

    https://huggingface.co/unsloth/LTX-2.3-GGUF/tree/main

    Choose based on VRAM:

    • Q4_K_M → lower VRAM (recommended)

    • Q8_0 → higher quality, more VRAM

    ⚠️ Very low quants = basically unusable

    Includes:

    • VAE (audio + video)

    • GGUF models


    ✍️ Text Encoders (REQUIRED)

    Option 1 – Gemma (recommended for low VRAM)

    https://huggingface.co/Comfy-Org/ltx-2/tree/main/split_files/text_encoders

    Option 2 – Full version (higher VRAM, better quality)

    ➡️ You still ALSO need:

    • Text Projection (from Kijai’s repo)


    🔍 Latent Upscaler

    https://huggingface.co/Lightricks/LTX-2/tree/main/

    Files:

    • ltx-2-spatial-upscaler-x2-1.0.safetensors

    • ltx-2-temporal-upscaler-x2-1.0.safetensors


    ⚠️ Important Notes

    • Use DEV GGUF diffusion models (no distilled base model yet)

    • Distilled LoRA is still REQUIRED

    • Most issues = outdated nodes or ComfyUI

    .


    Description

    FAQ

    Comments (11)

    tomastomy987155Feb 25, 2026· 6 reactions
    CivitAI

    im getting LTXVSeparateAVLatent

    tuple index out of range

    when im trying to run it. How do i fix this?

    homoludensFeb 25, 2026

    Me too. I updated ComfyUI and took an EmptyAudioLatent node from the workflow advertased there
    https://www.kombitz.com/2026/01/12/how-to-use-ltx-2-image-to-video-with-gguf-in-comfyui/

    error went away, but the workflow here produces the garbage (completely unreadable video) on the RTX 3060. I was able to run this workflow in FP8 mode of the ComfyUI. Probably this is the issue or its because I used the FP4 text encoder.
    Tried to run this workflow in FP16 mode, but machine hung out.

    All in all my investigation is still in progress...

    The_frizzy1
    Author
    Feb 26, 2026

    @homoludens I had the same issues while building the workflow i fixed it by just updating the nodes I thought they were up to date they were not please check that again thats why i put everywehre to update the nodes!! and for the garabge video I have put if youre gettign checkerboard pattern etc its because the nodes are not updated!!! Please make sure to read through the model page before attempting it!

    homoludensFeb 26, 2026· 1 reaction

    @The_frizzy1 The only thing that resolved the basic problem is to go away from GGUF (unet and clip). I tried Q2, Q4 and Q6 GGUFs for unet. I replaced GGUF unets with an LTX-2 Dev FP8 safetensors (at least works but quality needs to be improved by tuning of the workflow). Same goes to CLIP. All gemma 12b GGUF I've used generate error like "unexpected model name gemma3". Probably it will be resolved in future or I need to install a cutting edge transformers or smth else from github, but gemma 3 12b FP8 just works for me. It is slow on rtx 3060 with 12Gb of VRAM, but works. Torch Compile nodes don't work at all on rtx 3060 with LTX 2 or produce too noisy picture.

    homoludensFeb 26, 2026· 1 reaction

    PS: I suspect there is nothing to fix in this workflow. Problem is that on my side GGUF unet tells me FU no matter what I am trying to do [SHRUGZ ASCII]

    The_frizzy1
    Author
    Feb 26, 2026

    @homoludens personally I dont recommend running the gguf versions of Clip or anything except the model I usulaly get errors when using that.

    The_frizzy1
    Author
    Feb 26, 2026

    Are you on the portable version or desktop version if you can try the desktop version.

    PirateGirlMar 2, 2026

    v1.5. Very interesting workflow, but I have the same problem. Updated all nodes and comfy too. I even tried 4 different gguf loaders, but none worked. Non gguf version is not a solution for low vram. So, I wait until fix is found and try to incorporate low vram patches in these workflows to some other workflow in the meantime. I have 3060 12Gb and use portable comfy.

    The_frizzy1
    Author
    Apr 20, 2026

    In the new version I fixed it some things were not connected correctly but because things have already updated so much different text projection etc I deceided to update the version. I will make a new video on it!

    colorgbMar 2, 2026
    CivitAI

    got error on text encoder
    ***
    DualCLIPLoader
    Error while deserializing header: incomplete metadata, file not fully covered
    ***
    gemma_3_12B_it_fp4_mixed and connector fully downloaded

    The_frizzy1
    Author
    Mar 25, 2026

    Im not sure how to help cause im not getting the issue what i can say try to use a gguf version and use the gguf clip loader or try the full gemma if possible not the fp versions.

    Workflows
    LTXV2

    Details

    Downloads
    712
    Platform
    CivitAI
    Platform Status
    Available
    Created
    2/23/2026
    Updated
    7/27/2026
    Deleted
    -