AI Manga Cleaner
Erase dialogue, narration, and hand-drawn SFX from manga pages while the AI redraws the artwork underneath. Built for scanlators and indie translators.

Erase dialogue, narration, and hand-drawn SFX from manga pages while the AI redraws the artwork underneath. Built for scanlators and indie translators.


Add colour to black-and-white manga, comics and line art while aiming to preserve the original lines.

Upload a photo and choose from 1 of 12 manga styles. The built-in prompt updates automatically, while the original subject and scene remain recognisable.

Translate manga by detecting speech bubbles, translating their text and re-rendering it in a similar style.

Upload a portrait, selfie, pet photo or group picture, then choose an anime style. The result aims to retain the features that make the original recognisable.

Text removal on manga pages calls for care: speech bubble outlines, screentones and artwork can sit close to the lettering. Use the controls to specify whether to clear dialogue, narration, hand-drawn sound effects or signs. The selected model redraws the edited areas, and results can vary, particularly when lettering overlaps detailed artwork.
Hand-drawn sound effects can overlap speed lines, action effects and screentones. Ask the model to remove the lettering and redraw what is underneath. Review the result closely: lines or textures may not match the original exactly, and complex areas can need manual touch-ups.


Colour pages and vertical-scroll layouts may include gradients, soft shading and transitions between panels. You can ask the selected model to remove text and redraw those areas to blend with the surrounding image. Results depend on the page and model, so check gradients, lighting and signboards before adding new lettering.
Use the controls to choose which dialogue bubbles or narration boxes to clean. The selected model processes the uploaded page; results may need review and touch-ups.
Ask the selected model to remove hand-drawn sound effects and redraw the area beneath. Check the result for speed lines, action effects and scenery that may need touch-ups.
The model can attempt to match screentones, colour gradients and lighting in cleared areas. Check how well they blend with the surrounding artwork.
Ask the model to preserve speech bubble shapes, tails and outlines while clearing the text inside. Review the result, as the outlines may not be preserved exactly.
Page cleanup can be one step in a scanlation workflow. For example, a complex page with overlapping sound effects may need careful editing in an image editor. Use the model to try an initial cleanup, then review the result and make any needed touch-ups. Time and output quality depend on the page and selected model.

It can remove dialogue in speech bubbles, narration in caption boxes, hand-drawn sound effects and background signs you want cleared. The AI redraws the areas beneath removed text to blend with the surrounding image. Results can vary, especially where text overlaps detailed artwork; check the page and make any needed touch-ups.
It can attempt to remove sound effects that overlap the artwork and redraw the area underneath. Results depend on the image, and intricate effects woven into detailed scenery may need manual touch-ups.
The tool is designed for manga, manhwa, manhua, webtoons and doujinshi pages. Results depend on the page and selected model. Choose the closest page type in the form to guide the requested cleanup.
The tool can be instructed to preserve speech bubble outlines, tails and shapes while clearing text inside them. Check the result, as preservation may vary with the page. The cleared space can then be used for translated lettering.
Choose the “Bubbles only” preset or select “Dialogue Bubbles” under “What to Clean”. This asks the model to leave sound effects, narration boxes and signs untouched; check the result afterward.
Use a clear, legible scan where possible. Very small or low-resolution text may be harder for the selected model to handle, and fine details around removed sound effects may need touch-ups. Check the selected model’s upload and output options before generating.
You need to sign in before generating, and generation uses account credits. Check your balance, selected model and displayed credit cost before you start. Plan features and output options vary.
This tool can be part of a scanlation workflow. For example, a team might use it for an initial cleanup, then review the page and make any needed touch-ups in an image editor. Results and time saved depend on the source page and selected model.

GPT Image 2 is an OpenAI image model for generating and editing images. It can render text in multiple languages and ...

Generate and edit images with Nano Banana Pro. Sign-in and account credits are required; review the displayed cost be...
Create videos from text prompts, animate an image or use video input in supported modes. Choose a generation mode and...

Create videos from text prompts or images with Sora 2. Choose from the available options, then review the displayed c...
ByteDance's AI video model combines text, images, video clips and audio in one prompt. Use its @-reference system to ...