Edit and manipulate an image from a natural-language instruction prompt.
Public URL of the image to edit. The image must be between 300 and 4096 pixels in both width and height, and its width-to-height ratio must be between 0.4 and 2.5.
JPG, PNG, WEBP · 1 image · 20 MB · 300–4096 px per side · width/height ≥ 0.4 · width/height ≤ 2.5
Image model. Supported values: BANANA, BANANA_2, BANANA_PRO, WAN2_7, QWEN_IMAGE_3_PRO, SEEDREAM5, GPT_2_IMAGE, or AIDRAW_EDIT.
1K (fixed)
Generation route. Supported values: eco or stable. Defaults to stable.
Whether the prompt should be optimized upstream.
Optional Nexscope editing template. Supported values: white-background, product-main-image-retouching, scene-generation, poster-design, add-person-or-object, change-animal, flat-lay-apparel, 3d-apparel, mannequin-to-model, cartoon-illustration, body-reshape, change-style, remove-element, remove-text, remove-watermark, add-watermark.
Optional traffic or product source code.
GPT Image quality only. Supported values: low, medium, or high. Defaults to medium.
API keys require an active subscription. Trial credits don’t include API access. View plans.
Final credits are shown with your result.
Your image will appear here
Saved in this browser for 24 hours
Your generations will appear here.
