Training

Guide: Train Your First FLUX LoRA in Under 6 Hours

LoRA (Low-Rank Adaptation) lets you fine-tune a model on your own images without training from scratch. This guide covers building a character or style LoRA for FLUX Dev using Kohya SS — the most reliable training tool available.

#Hardware Requirements

GPU	VRAM	Training Time (10 epochs)
RTX 5080	16GB	~2–3 hours
RTX 4090	24GB	~1.5–2 hours
RTX 3080 16GB	16GB	~4–5 hours
RTX 3080 10GB	10GB	Requires gradient checkpointing

Minimum: 10GB VRAM with 8-bit Adam optimizer.

Hardware Partner

Running these workflows? ComputeAtlas.ai helps you find the right GPU

Optimization is only half the battle. Get precise VRAM benchmarks and hardware recommendations tailored for ComfyUI.

Check GPU Prices →

[AD VISUAL: GPU BENCHMARKS]

#Step 1 — Dataset Preparation

Quality of training data is 80% of your result.

Image Requirements

→Quantity: 15–50 images (character LoRA), 50–200 (style LoRA)
→Resolution: 1024x1024 minimum, consistent aspect ratio
→Variety: Different angles, lighting, expressions (character), or diverse examples (style)
→Quality: Sharp, no compression artifacts, no watermarks

Folder Structure

comfyui-workflow.json

my_dataset/
├── img/
│   └── 10_my_character/     # "10" = repeat count
│       ├── image_001.jpg
│       ├── image_001.txt    # caption file
│       ├── image_002.jpg
│       ├── image_002.txt
│       └── ...
├── log/
└── model/

The number before the underscore (10_) is the repeat count — how many times each image is shown per epoch. For small datasets (15–20 images), use 10–15. For larger datasets (50+), use 3–5.

#Step 2 — Auto-Captioning

Every image needs a text caption. Use WD14 tagger for automatic captioning.

comfyui-workflow.json

# Install tagger
pip install wd14-tagger

# Run on your dataset
python wd14_tagger.py \
  --input_dir ./my_dataset/img/10_my_character \
  --output_dir ./my_dataset/img/10_my_character \
  --threshold 0.35 \
  --caption_extension .txt

Edit the captions — Add your trigger word to the start of every caption:

comfyui-workflow.json

# Before:
woman, blonde hair, blue eyes, outdoor, sunlight

# After:
mycharcterv1, woman, blonde hair, blue eyes, outdoor, sunlight

Your trigger word (mycharcterv1) is what you type in prompts to activate the LoRA later.

#Step 3 — Install Kohya SS

comfyui-workflow.json

git clone https://github.com/kohya-ss/sd-scripts
cd sd-scripts
python -m venv venv
.\venv\Scripts\activate
pip install torch torchvision --index-url https://download.pytorch.org/whl/cu121
pip install -r requirements.txt

#Step 4 — Training Configuration

Create flux_lora_config.toml:

comfyui-workflow.json

[model_arguments]
pretrained_model_name_or_path = "C:/models/flux1-dev-fp8.safetensors"
clip_l = "C:/models/clip_l.safetensors"
t5xxl = "C:/models/t5xxl_fp16.safetensors"
ae = "C:/models/ae.safetensors"

[dataset_arguments]
train_data_dir = "C:/my_dataset/img"
resolution = "1024,1024"
batch_size = 1
cache_latents = true
cache_latents_to_disk = true

[training_arguments]
output_dir = "C:/my_dataset/model"
output_name = "my_character_lora_v1"
save_every_n_epochs = 2
max_train_epochs = 10
learning_rate = 1e-4
optimizer_type = "AdamW8bit"
lr_scheduler = "cosine_with_restarts"
lr_warmup_steps = 100
gradient_checkpointing = true
mixed_precision = "bf16"
save_precision = "bf16"
seed = 42

[network_arguments]
network_module = "networks.lora_flux"
network_dim = 32
network_alpha = 16

Key Parameters Explained

network_dim (rank) — Controls LoRA size. Higher = more capacity, more VRAM.

→Character LoRA: 32
→Style LoRA: 16–64
→Concept LoRA: 8–16

network_alpha — Usually set to half of dim. Controls learning rate scaling.

learning_rate — Start at 1e-4. If results are too strong, lower to 5e-5.

max_train_epochs — 8–12 for character, 5–8 for style.

#Step 5 — Run Training

comfyui-workflow.json

# Activate venv
.\venv\Scripts\activate

# Start training
python flux_train_network.py --config_file flux_lora_config.toml

RTX 5080 Performance

comfyui-workflow.json

Epoch 1/10: ~18 minutes
Total (10 epochs): ~3 hours
VRAM usage: ~14.2GB peak
Loss start: ~0.18
Loss target: ~0.01–0.03

Watch the loss curve — it should decrease steadily. If it plateaus early, lower the learning rate.

#Step 6 — Evaluate Your LoRA

Test your LoRA in ComfyUI after each saved checkpoint:

comfyui-workflow.json

# Test prompt
mycharcterv1, portrait photo, professional lighting, sharp focus

# LoRA weight in ComfyUI
Start: 0.8
Adjust: 0.6 (subtle) — 1.0 (strong)

Good signs:

→Subject is recognizable at weight 0.7–0.9
→Prompt still controls other elements (background, lighting)
→No artifacts or distortion

Bad signs:

→Only activates at weight 1.0+ (undertrained)
→Breaks non-subject elements (overtrained)
→Flickering or artifacts (learning rate too high)

#Step 7 — Export and Use

Your trained LoRA is saved as a .safetensors file in your output directory.

comfyui-workflow.json

# Place in ComfyUI
ComfyUI/models/loras/my_character_lora_v1.safetensors

# Load in workflow
Use "Load LoRA" node
Set strength_model: 0.8
Set strength_clip: 0.8

#Common Issues

OOM at start — Enable gradient_checkpointing = true and reduce batch_size to 1.

Loss not decreasing — Check your captions are correct and trigger word is consistent.

Subject not activating — Increase LoRA weight, or train more epochs.

Overfit (everything looks like subject) — Reduce epochs or increase dataset variety.

Slow training — Ensure cache_latents = true and mixed_precision = "bf16".

Hardware Partner

Running these workflows? ComputeAtlas.ai helps you find the right GPU

Optimization is only half the battle. Get precise VRAM benchmarks and hardware recommendations tailored for ComfyUI.

Check GPU Prices →

[AD VISUAL: GPU BENCHMARKS]