When you train a model in Atlas, you’re teaching our AI how your specific product looks so it can generate new, realistic, and brand-aligned images. This model learns from the photos you upload, allowing it to recreate your product in new styles, backgrounds, and scenarios.
The more images you upload — and the more varied they are — the smarter your model gets.
How to Train Your AI Model
Go to the AI Product Photos section in the Atlas app
Click "Train Model"
Upload product images of your item
Use multiple angles and lighting conditions
Include close-ups and lifestyle shots
The more images you add, the better your results
Select your training duration
Fast Training: Quicker results, good for tests or rough drafts
Extended Training (3 hours): Strongly recommended for high-quality, photo-realistic results
Click "Start Training" and wait for the process to complete. You’ll receive a notification once your model is ready
Training Tips
Use these guidelines to get the most accurate and high-quality results:
Use high-resolution, original product images (avoid screenshots or blurry photos)
Show your product from all sides — front, back, top, angled
Vary the background slightly (white, wood, lifestyle scene)
Include any close-ups that showcase important texture or product details
Understanding Models and Generation
Product-Specific Models: Each trained model represents ONE specific product
Training: Teaching the AI to understand a particular product's details
Generating: Creating new photos using a trained product model
You must first train a model for each product before you can generate images