MagicRemover

What is MagicRemover?

MagicRemover is a browser-based AI photo editing tool for erasing unwanted elements from images. It handles people, objects, text, watermarks, logos, stickers, and similar overlays. The site is organised as a set of task-specific tools rather than one generic editor, because brush size, margin guidance, and realistic success rates differ by job. Users upload a JPG or PNG, brush a red mask over the region to remove, and run the job. No installation is required, no account is needed to start, and outputs carry no badge or watermark. The service returns a PNG of about 1 megapixel; a 4000×3000 photo comes back at roughly 1155×866, a size suited to screens and social posts rather than large prints.

How MagicRemover works

The method is generative inpainting. The uploaded image and mask are sent to a diffusion model trained to predict what plausibly belongs in a hidden region given the surrounding pixels. It does not copy pixels from elsewhere in the photo, and it does not reveal what was actually behind the object. It generates a new surface and blends it into the existing image. This distinction matters: over walls, sky, sand, water, or smooth gradients the prediction is often close to invisible, while over fine repeating detail such as foliage, brickwork, crowds, or patterned fabric the model makes an educated guess and a careful viewer may notice. Image jobs typically return in 15 to 30 seconds. Image processing runs on hosted inference providers, currently RunningHub with Replicate as a fallback path; the video subtitle tool runs on apimodels.app; results are stored on Cloudflare R2 and served over HTTPS.

Main features

• Background people remover: clears photobombers, tourists, pedestrians, and crowds while aiming to keep the main subject crisp.
• AI object remover: erases street clutter, cables, signs, trash, parked vehicles, traffic cones, and similar unwanted objects.
• Text removal: erases date stamps, subtitles, captions, meme text, and burned-in labels without reading or translating them.
• Watermark, logo, and sticker removal: scrubs © stamps, overlays, signatures, brand logos, broadcast bugs, sponsor badges, stickers, and emoji from images the user owns.
• AI image generator: creates an original image from a text description, with optional reference photos, at 3 credits per image and no free allowance.
• No-watermark output and task-specific tool pages with guidance for each removal type.

Pricing

The free tier gives guests 2 removals per rolling 24 hours per device. Signing in with Google adds a separate allowance of 2 per day per account, so a signed-in visitor can use up to 4 in a day. Both counters reset at midnight UTC and are shared across every tool on the site. No credit card is required to start, there is no quietly expiring trial, and outputs are not watermarked. Past the daily allowance, signed-in users can buy credits: a removal costs 1 credit, while the image generator and edit-text tool cost 3 credits per image and have no free allowance. The supplied pages do not list currency prices for credits.

Common use cases

MagicRemover is aimed at quick clean-up work on existing photos and screenshots. Typical uses include clearing tourists or passers-by from travel shots, venue photography, and candid portraits; removing street clutter, cables, construction debris, or parked vehicles from scene photos; cleaning date stamps, captions, subtitles, and meme text from screenshots and scanned material; and removing watermarks, logos, signatures, stickers, or emoji from images the user owns. It can also be used to generate a new image when a suitable picture is not available. The task-specific pages are intended to match the tool to the removal job rather than force every edit through one generic brush workflow.

Limitations

MagicRemover does not recover hidden information. A rebuilt area is generated, not uncovered, so if a sign covered someone’s face the result is a plausible face-shaped guess rather than the actual face. The text tools are not OCR: they erase writing but never read or translate it. Busy texture is hard, and removals over foliage, crowds, brick, or text-on-text often leave soft residue that needs a second pass or manual touch-up. The web UI works one image at a time and deliberately has no batch mode. It does not defeat invisible watermarking; provenance signals such as SynthID or C2PA metadata are not what the brush touches, and the service makes no attempt to strip them. Very large images are downscaled to about 1 megapixel before processing and return at that size. Results are sized for screens and social posts rather than large prints.

Summary

MagicRemover is a free-to-start, browser-based AI object remover with a clear daily allowance and a credit system for additional use. Its generative inpainting approach works best on smooth backgrounds and screen-sized images, while fine detail, hidden-content recovery, OCR, batch processing, and large-print output remain outside its scope. Users who need a quick removal on a single JPG or PNG can work without installing software or adding a watermark, provided they accept the resolution limit and the generated nature of the fill.