Here is one workflow:
- ltx2.3-it2v-gguf.json
use gguf and split models.
Reference output:
RTX4070 16GB vRAM + 32G RAM i2v(10Eros_v1.4; 15*24*768*960): Prompt executed in 441.41 seconds
RTX5080 laptop 16G vRAM + 64G RAM i2v(10Eros_v1.4; 15*24*768*960): Prompt executed in 338.79 seconds
Requirements:
Models:
- 10Eros_v1.4-Q5_K_M.gguf (GGUF), Put in: ComfyUI\models\unet
https://huggingface.co/vantagewithai/LTX2.3-10Eros-1.4-GGUF/resolve/main/10Eros_v1.4-Q5_K_M.gguf
- ltx-2.3-22b-distilled-lora-1.1_fro90_ceil72_condsafe.safetensors (distill lora) Put in: ComfyUI\models\lora\LTX
- ltx-2.3-spatial-upscaler-x2-1.1.safetensors (latent upscale) Put in: ComfyUI\models\latent_upscale_models
https://huggingface.co/Lightricks/LTX-2.3/resolve/main/ltx-2.3-spatial-upscaler-x2-1.1.safetensors
- gemma-3-12b-it-heretic-v2_fp8_e4m3fn.safetensors (CLIP) Put in: ComfyUI\models\text_encoders
- ltx-2.3_text_projection_bf16.safetensors (Split file.)
- Split VAE files (VAE) Put in: ComfyUI\models\vae\ltx
- LTX23_video_vae_bf16.safetensors
https://huggingface.co/Kijai/LTX2.3_comfy/resolve/main/vae/LTX23_video_vae_bf16.safetensors
- LTX23_audio_vae_bf16.safetensors
https://huggingface.co/Kijai/LTX2.3_comfy/resolve/main/vae/LTX23_audio_vae_bf16.safetensors
- taeltx2_3.safetensors
https://huggingface.co/Kijai/LTX2.3_comfy/resolve/main/vae/taeltx2_3.safetensors
ComfyUI Nodes:
- rgthree-comfy
- ComfyUI-KJNodes
- ComfyUI-VideoHelperSuite
- ComfyUI-GGUF
Description
use gguf and split models.
FAQ
Comments (6)
There are a lot of LTX gguf workflows available. What is the interest of yours?
thanks for respose. I just wanna make it easy to understand and use.
I just started using this workflow it's really good! Any tips for someone who's never used LTX, and is there any plans to add lipsync for audio files?
I'm glad the workflow has been useful to you. Since prompts directly impact the output, you can ask LLM AI models (such as GPT or Gemini) to consult the recommended prompting guidelines for the specific generative model you are using, and then have them assist you in creating effective prompts. As for audio file synchronization, I'll look into that when I have some free time.
@Chris_T Awesome! One other thing, do you know a way I can input a final frame as well? I know that was a thing in WAN but not sure if it can be done in ltx.
works really well out of the box. 8sec videos on a rtx 3060 with 12gb ram takes around 10min.