AI Image Edit
Edits existing images with text instructions using AI. Restyle, combine references, or turn sketches into art.
AI Image Edit takes one or more input images and applies text-based edits using AI. The flagship model is Nano Banana 2, which can restyle images, combine multiple references into one, turn rough sketches into polished art, and much more.
Four model tiers are available: FLUX.2 Turbo (fast, affordable edits), FLUX.2 Pro (professional quality), FLUX.2 Max (highest fidelity), and Nano Banana 2 (the most capable for complex creative edits and prompt following). All four tiers accept up to three reference images on the ref_1, ref_2, and ref_3 pins for multi-image fusion.
Parameters
Section titled “Parameters”Prompt
Text instruction for how to edit the image
Model
0=Fast, 1=Standard, 2=Quality, 3=Advanced
Generation
Content filter (Fast/Standard/Quality only)
Advanced tier aspect ratio
Advanced tier: 0.5K/1K/2K/4K
Advanced tier content filter strictness
Allow web search (+1 credit, Advanced only)
Random seed; -1=random each time
Tier-dependent visibility: Fast/Standard/Quality show Safety Tolerance. Advanced shows Nb2 * params. All tiers accept multi-image input via ref_1/2/3 pins.
Usage Tips
Section titled “Usage Tips”- Use Advanced tier (Nano Banana 2) for the best results on complex edits.
- Wire multiple images as input for reference-driven generation; the AI combines them.
- Be specific in your edit instruction: “make the background a sunset” is better than “change background.”
- Use Enable Web Search for reference-based edits that need real-world knowledge.