Image Editing and Cropping

Nano Banana Pro InpaintingAPI

Powerful inpainting model run by Nano Banana Pro. Uses image, mask and optional reference image.

Test in Dashboard View Docs ← All Tools

Three ways to use this tool

REST API

One POST request. Get the result back directly — as an image, video, or JSON depending on the tool.

terminal
curl -X POST https://apiai.me/api/workflow/nano-banana-pro-inpainting \
     -H "X-API-Key: YOUR_API_KEY" \
     -F "image=@input.jpg" \
     -F "aspect_ratio=1:1" \
     -F "blend_strength=1.0" \
     -F "image_mask=@image.jpg" \
     -F "image_reference=@image.jpg" \
     -F "image_size=1K" \
     -F "max_output_tokens=VALUE" \
     -F "negative_prompt=VALUE" \
     -F "prompt=Add prompt here" \
     -F "safety_filter_level=BLOCK_ONLY_HIGH" \
     -F "temperature=0.3" \
     -F "top_k=64" \
     -F "top_p=0.95"
     --output result.png

Dashboard Playground

Test this tool visually before writing any code. Upload an image, set parameters, and see the result live. When it looks right, copy the auto-generated curl command and paste it into your app.

  1. Open the dashboard → API Toolbox
  2. Find Nano Banana Pro Inpainting and click it
  3. Upload your input and adjust parameters
  4. Copy the curl command and ship
Open Dashboard →

Batch Processing

Process hundreds of images at once without writing a loop. Upload a CSV with one row per item, set your parameters, and download the results as a ZIP when they're done.

  1. Open the dashboard → Batches
  2. Select Nano Banana Pro Inpainting
  3. Add your content and start the batch
  4. Download results ZIP when complete
Go to Batches →

Parameters

Name Required Description Default / Options
aspect_ratio optional Output aspect ratio. Options: 1:1, 3:4, 4:3, 9:16, 16:9. The mask is padded (no crop) to this aspect before the Gemini call, so the final image has exactly this shape. 1:13:44:39:1616:9
blend_strength optional How strongly the AI output replaces the subject inside the mask. 1.0 = full AI; <1.0 mixes in some subject. 1.0
image_mask required Grayscale mask. White = where AI output should appear. Black = ignored area.
image_reference optional Optional reference image (color sample, material to apply etc.). If provided, Gemini receives it as a second image alongside the main image. If skipped, the prompt alone drives the inpainting.
image_size optional Output resolution. Options: 1K, 2K, 4K. 1K
max_output_tokens optional Max output tokens (up to 32768). Leave empty to use the model default.
negative_prompt optional Things to exclude. If set, appended to the prompt as a 'Avoid: ...
prompt required Instruction for Gemini. Describes how to apply the reference to the subject. Add prompt here
safety_filter_level optional BLOCK_LOW_AND_ABOVE, BLOCK_MEDIUM_AND_ABOVE, BLOCK_ONLY_HIGH, BLOCK_NONE. Applied to all safety categories. BLOCK_ONLY_HIGH
temperature optional 0 = deterministic, 1 = creative . Default 0.3 keeps the AI close to the subject. 0.3
top_k optional Max 64. Lower values prioritize structural coherence; higher values allow for more intricate and varied details. 64
top_p optional Max 1. Lower values keep the output focused and consistent; higher values allow for more variety and creative nuance. 0.95

Ready to integrate?

Get an API key, test Nano Banana Pro Inpainting in the dashboard, and copy the curl command.

Get API Key Full API Reference