# LoRA Palmer

> The smartest LoRA trainer for Flux models with AI-powered auto-captioning and one-click training.

## What is LoRA Palmer?

LoRA Palmer is a web-based tool for preparing and training Flux LoRA (Low-Rank Adaptation) models. It streamlines the entire training workflow from image upload to trained model download.

## Key Features

- **AI-Powered Auto-Captioning**: Automatically generates detailed, training-optimized captions using Gemini or Claude AI
- **3D Rotation Visualization**: Interactive cube for setting precise camera angles and spatial relationships
- **Multi-Provider Training**: Choose between Replicate (ostris/flux-dev-lora-trainer) or fal.ai (flux-2-trainer)
- **Product Category Optimization**: Specialized captioning rules for general products, apparel, and food photography
- **One-Click Training**: Streamlined workflow from upload to trained LoRA download

## Product Categories

### General Products
Focus on surface properties (metal, plastic, wood), machining details, parting lines, and micro-scratches.

### Apparel
Specialized for fabric analysis: weight, weave patterns, drape characteristics, and stitching details.

### Food Photography
Optimized for organic textures, moisture, freshness indicators, and appetizing presentation.

## Pricing

Credit-based system:
- **Training**: 100 credits per LoRA training
- **Inference Testing**: 2 credits per image generation

Credit Packages:
- Starter: $10 for 500 credits (5 trainings)
- Pro Creator: $25 for 1,500 credits (15 trainings)
- Studio: $100 for 7,000 credits (70 trainings)

## Technical Specifications

- **Supported Models**: Flux LoRA via Replicate or fal.ai
- **Image Requirements**: Automatically normalized to 1024x1024 with white background padding
- **Output Format**: Standard LoRA safetensors compatible with Flux inference
- **Training Parameters**: Configurable learning rate, LoRA rank (default 32), and dynamic step calculation

## Workflow

1. Upload product images
2. Set camera rotation angles using 3D cube
3. Configure trigger token and product category
4. Run AI auto-captioning
5. Review and refine captions
6. Start training
7. Download trained LoRA model

## Contact

Website: https://lorapalmer.com