How is image to image different from text to image?

Text to image starts from a prompt. Image to image starts from uploaded references, making it better for edits, preservation, background changes, and design extensions.

How many reference images can I upload?

The workspace supports up to 14 reference images for subject, style, layout, material, or color direction.

How do I keep the result closer to the original?

Clearly state what should stay unchanged, such as face, product shape, pose, camera angle, clothing, logo, or main composition.

Nano Banana Pro image-to-image workspace

Upload references, describe the edit, and let Nano Banana Pro preserve useful details while changing the requested parts.

Image to imagegoogle/nano-banana-pro

Upload references, describe the edit, and let Nano Banana Pro preserve useful details while changing the requested parts.

Reference images0/14

Reference images(0/14)

Prompt

Tell the model what should change and what must stay untouched.0/2500

Aspect ratio

Resolution

Output format

Safety filter

Allow fallback model

If the official model is temporarily unavailable, allow a lower-cost fallback; if it is used, the credit difference is returned.

Cost 18 credits

2Kmatch_input_imageJPG

Generated images

Ready to generate

Edited images will appear here once the task completes.

Open text-to-image workspace

How to use Nano Banana Pro image to image

The image-to-image workspace uses existing visuals as a foundation for restyling, background changes, reference blending, and detail enhancement.

Reference-guided editing

Upload subject, style, or composition references to guide the generated result toward a clearer direction.

Preserve important details

Use the prompt to explain which people, products, poses, camera angles, and brand elements should stay stable.

Multi-image reference blending

Use up to 14 reference images for style blending, subject composition, and more complex visual exploration.

Nano Banana Pro image to image FAQ

Understand how references, edit prompts, and output settings affect the final image.