ComfyUI Workflows
Free, ready-to-use workflows for Kontext LoRA training
Single Image to LoRA
The complete workflow for training character models from a single reference image. Includes automatic captioning, parameter optimization, and quality validation.
Download Latest Version View on GitHub →Basic LoRA Trainer
Simplified workflow perfect for beginners. Minimal setup required with automatic parameter selection.
Advanced Parameter Control
Full control over all training parameters including learning rate scheduling, noise offset curves, and validation checkpoints.
Anime Character Specialist
Optimized specifically for anime and cartoon character training with enhanced color vibrancy and style preservation.
Low VRAM Optimized
Specially designed for 8GB GPU users with memory optimization, FP8 precision, and efficient resource management.
Batch Training System
Train multiple characters sequentially with queue management, automatic model saving, and progress tracking.
Research & Experimentation
Full featured workflow with A/B testing, loss curve analysis, and experimental parameter exploration tools.
🔧 System Requirements
ComfyUI Setup
- ComfyUI latest version
- Custom nodes installed
- FLUX.1 Kontext model
- Python 3.10+ environment
Hardware Requirements
- NVIDIA GPU with 8GB+ VRAM
- 16GB+ system RAM
- 50GB free storage space
- CUDA 11.8 or 12.1
Required Models
- FLUX.1 Kontext [dev]
- T5 Text Encoder
- VAE Decoder
- Optional: FP8/GGUF versions
Custom Nodes
- ComfyUI-Manager
- rgthree-comfy
- ComfyUI-KJNodes
- efficiency-nodes-comfyui
🚀 Quick Installation Guide
Get up and running with Kontext LoRA workflows in just 5 minutes
Download Workflow
Choose and download the workflow that fits your needs
Install Dependencies
Use ComfyUI-Manager to install required custom nodes
Load Models
Download and place FLUX.1 Kontext models in the correct directories
Import Workflow
Drag and drop the JSON file into ComfyUI interface
Start Training
Upload your image and click "Queue Prompt" to begin training