Samsa offers different model types to give you precise control over your generated images. Understanding when to use each type will dramatically improve your results.
Style models define the visual aesthetic of your images.
What they do:
Apply consistent visual styles across generations
Control photography style, illustration type, or artistic direction
Perfect for maintaining brand consistency
Best for:
Brand photography styles
Illustration aesthetics
Artistic directions
Visual identity consistency
Person models capture and reproduce specific people or characters.
What they do:
Generate consistent representations of people
Maintain facial features and characteristics
Create images of the same person in different contexts
Best for:
Brand ambassadors and mascots
Character consistency in campaigns
Personalized content creation
Object models learn specific products or objects.
What they do:
Accurately reproduce product details
Maintain product consistency across images
Generate products in various settings
Best for:
Product photography
E-commerce visuals
Marketing materials featuring specific products
Scene models capture environments and locations.
What they do:
Recreate specific environments
Maintain location consistency
Generate variations of places
Best for:
Location-based campaigns
Environmental branding
Consistent backgrounds
For maximum control, combine different model types:
Style + Person: Generate consistent characters in a specific visual style
Style + Object: Product shots with brand-consistent aesthetics
Style + Scene: Location imagery with unified visual treatment
To use a trained model:
Click the model selector in the prompt editor
Choose your model from the dropdown
The model will be applied to your generation
You can also @mention models directly in your prompt for quick access.
