Face to Many Kontext

Frequently Asked Questions

What types of images work best as input?

Clear, well-lit photos of faces work best. The model supports JPEG, PNG, GIF, and WebP formats. Make sure the face is clearly visible and not heavily obscured.

Can I generate multiple variations at once?

Yes, you can generate up to 10 images at once by adjusting the 'Number of Images' setting. This is useful for exploring different variations of the same style and persona combination.

What's the difference between Style and Persona?

Style refers to the artistic rendering (anime, cartoon, pixel art, etc.), while Persona refers to the character type (ninja, robot, vampire, etc.). You can combine any style with any persona for unique results.

What does 'Random' do for Style and Persona?

When set to 'Random', the model will automatically select different styles or personas for each generated image, giving you a variety of unexpected and creative results.

Can I preserve elements from the original image?

Yes, you can choose to preserve the background and/or outfit from the original image using the respective checkboxes. This helps maintain consistency while transforming the face.

What is Safety Tolerance?

Safety Tolerance controls how strictly the model filters content. Level 0 is most strict, Level 2 is most permissive. Higher levels allow more creative freedom but may produce content that needs manual review.

How do I get reproducible results?

Set a specific seed value to get the same results with the same inputs. Leave the seed empty for random results each time.