Custom Model & LoRA Training

Train checkpoints and LoRAs for SDXL, Flux, Pony, and Illustrious based on your specific needs: characters, styles, products. Services are performed directly by Copax – the author of model series like Copax TimeLess and SDG-Realistic, with many self-trained models released on Civitai, Tensor.art, and Hugging Face.

What you can order

Character LoRA: maintain consistent facial features and appearance across multiple images – ideal for virtual characters, comics, and social media content.
Style LoRA: recreate a brand's unique drawing or photography style.
Product LoRA: integrate real products into AI-generated advertising photos.
Checkpoint: fine-tune base models when you need to change the overall style and capabilities.

Not sure whether to choose LoRA or a checkpoint? Read What is the difference between LoRA and Checkpoint?

Process

Contact us, describe your goals, desired base model, and reference images.
Consultation on data requirements (image quantity, quality) and pricing.
Data preparation, captioning, and training.
Testing, sending sample images for approval, and fine-tuning.
Delivery of the model file (.safetensors) along with usage instructions (trigger words, weights, recommended parameters).

Ready for API use

Trained models can be integrated with our AI Image Generation API to be used on your website or app.

Pricing

Note

You must have the rights to use the images for training. We do not accept training for real people's images without their explicit consent.

Custom Model & LoRA Training