Skip to content

Docker Image Version

LTX 2.3 inference with ComfyUI

A streamlined and automated environment for running ComfyUI with LTX-2.3 video models, optimized for use on RunPod

Example included workflows

i2v

Select number frames, size , prompt and offload

iv2v pose transfer DWPose

Select number frames, size , prompt and offload

iv2v pose transfer SDPose (Body Ratio Mapper)

Select number frames, size , prompt and offload

🔧 Features

  • Automatic model and LoRA provisioning via environment variables.
  • Kajai models no checkpoint
  • Included workflows for video generation and enhancement using pre-installed custom nodes.
  • Compatible with high-performance NVIDIA GPUs (CUDA 12.8).
  • Compiled attentions and GPU accelerations.
  • Automatic selecting bf16 or fp8 models/workflows.
  • Latent preview enabled for both samplers.
  • Included worflows based on RuneXX
  • Lora manager

🔧 Built-in authentication

  • ComfyUI
  • Code Server
  • Hugging Face API
  • CivitAI API

📦 Deployment on runpod

📘 Tutorial

Pod running on L40S bf16

Pod running on L40S ComfyUI

Pod running on RTX 6000 Ada bf16

Pod running on RTX 6000Ada ComfyUI

Pod running on RTX A5000 fp8

Pod running on RTX A5000 ComfyUI

Pod running on RTX 4090 fp8

Pod running on RTX 4090 ComfyUI