AI Image Editor AI

Two ways to edit. Prompt edit describes a change in words and lets Gemini's image-editing model do it. Remove (brush) paints over a watermark, logo, or object and erases it cleanly with our self-hosted inpainting model — no AI watermark added.

Server-Side Processing

Your image is sent to our backend, processed by Gemini, and the result is streamed back. The original and the edit are deleted after the response — we don't keep copies.

Drop an image here or click to upload

PNG, JPEG, or WebP — up to 20 MB

How It Works

1

Drop an image

PNG, JPEG, or WebP up to 20 MB. Larger images are scaled down to 2048px on the longest side before editing — preserves quality for screen use, keeps things fast.

2

Describe your edit

Plain English. The model follows specific instructions better than vague ones — say "remove the man in the red shirt", not "clean it up".

3

Review and download

See before/after side by side. If the edit isn't right, tweak the prompt and run again.

Frequently Asked Questions

What model is this using?

Google's Gemini image-editing model (informally "Nano Banana"). ChamPDF wraps the API and adds a focused UI for common edits.

How is this different from the PDF Watermark Remover?

The PDF Watermark Remover works on PDF pages. The AI Image Editor works on a single image and gives you two tools: a prompt mode (describe any change, Gemini applies it) and a brush "Remove" mode (paint over a watermark or object and our self-hosted LaMa model erases it cleanly). Use whichever fits the job.

Why does my edit sometimes refuse to run?

Gemini has safety filters and will refuse some prompts. If a request is declined, the model's reason is shown. Reword the prompt or pick a different image.

Are my images stored?

No. The original and the edit are streamed through our backend and deleted as soon as the response completes. We don't keep copies.