Cleanup.pictures

Remove unwanted objects from photos with AI inpainting.

Image Generation

Overview

Cleanup.pictures is the AI object remover that popularized the "paint over it, and it's gone" workflow for the general public. Built by ClipDrop — the AR-first tooling company that was acquired by Stability AI in early 2023 and then spun back out to Jasper in mid-2024 — Cleanup.pictures made the LaMa inpainting model accessible to anyone with a browser, no signup required for the free tier, no software to install, no learning curve beyond "brush over what you want to remove." Seven-plus years after the underlying research landed, it remains the tool most people are thinking of when they say "just erase that person in the background."

The pitch is narrow on purpose. Cleanup.pictures does not want to be a full photo editor. It does not generate images, restyle them, or upscale them meaningfully. It removes unwanted objects — people, signs, watermarks, power lines, tourists, ex-partners — and reconstructs the pixels underneath. That is the entire product. In a market where nearly every AI image tool has ballooned into a "creative platform" trying to cross-sell you a stock library and a video generator, Cleanup.pictures still respects the specialist bet: one job, done well, with an interface a five-year-old could figure out in under thirty seconds.

The 2026 version has evolved technically without breaking that promise. The default inpainting engine is materially better than the original LaMa model — it handles textured backgrounds (grass, brick, water) and reconstructs plausible geometry in scenes with strong perspective, both of which used to be reliable failure modes. There is a "HD" mode on paid tiers that runs a heavier model with better texture continuation, and a batch endpoint for teams that need to strip watermarks or license plates across large image sets. But the free tier is still the calling card, and the reason most people know the product name.

Where Cleanup.pictures stops being enough is when object removal is one step in a longer workflow. Product photographers who need cutouts, scene composition, and batch resizing are better served by Photoroom. Anyone who needs full editorial retouching — skin, color grading, dodge and burn — belongs in Photoshop or Affinity Photo with a proper generative fill. Cleanup.pictures is the fastest way to erase something from a photo. Once you need anything after that, you are in a different tool.

Key Features

Cleanup.pictures ships a deliberately short feature list. The entire product is the eraser and the model behind it.

  • AI object removal via brush. Paint over any object, person, sign, or blemish and the model reconstructs the pixels underneath. Works in-browser, no signup for casual use, no installation.
  • Automatic subject detection (Pro). Instead of brushing, click the object you want gone. The model segments it and removes the whole shape. Speeds up complex removals — trees, vehicles, animals — dramatically.
  • HD resolution processing. The free tier caps output at ~720px on the long edge. Pro removes the cap and processes at full input resolution up to 25MP, using a heavier inpainting model.
  • Text and watermark removal. A specific mode tuned for logos, watermarks, and printed text on photos. Works notably better than generic inpainting on repetitive patterns.
  • Batch API for volume workloads. REST endpoint that accepts an image plus a mask and returns the inpainted output. Priced per call. Used by real-estate platforms to strip watermarks and by e-commerce catalogs to remove backgrounds from product shots at scale.
  • Undo/redo history with local editing. Multiple removal passes on the same image without leaving the canvas. Useful when the first pass doesn't quite get everything.
  • Mobile web that actually works. The interface holds up on phones — brush gesture, pinch to zoom, no mystery menu diving. A lot of real usage is thumb-driven on iPhone.
  • No login for one-off use. You can land, upload, brush, and download without creating an account on the free tier. Rare in 2026 and a real reason casual users trust the product.

Pricing

Cleanup.pictures runs a lightweight freemium model, with a small monthly subscription for HD output and an API tier for volume.

Plan Price What you get
Free $0 Web tool, 720px output, no login required, watermark on downloads
Pro $5/month or $36/year Full-resolution output up to 25MP, HD inpainting model, auto-detection, no watermark
Team $19/month per seat Pro features + shared workspace, presets, batch upload UI
API $0.02-$0.05 per call REST endpoint, priced by resolution and model tier, volume discounts

The Free tier is unusually usable. There is no daily quota, no signup gate, and no watermark on preview-resolution output. What holds it back is the resolution cap and the older model behind it — for a social share or a low-effort cleanup, it is genuinely fine, and users routinely never upgrade.

Pro at $5/month is the price point Cleanup.pictures has held for years, and it remains one of the cheapest paid tiers in the AI image tooling space. The Annual plan at $36/year works out to $3/month and is the honest recommendation for anyone using the tool more than twice a month.

Team at $19/seat is aimed at real-estate agencies, marketplace ops teams, and small marketing shops that need shared presets and a batch UI. It is fine, but most teams with real batch needs skip to the API.

API pricing is the most useful tier for developers. At $0.02-0.05 per call depending on resolution and model choice, Cleanup.pictures' inpainting endpoint is aggressively priced against ClipDrop's own generative APIs and Stability AI's inpainting endpoints — often the right pick when object removal is a fixed subroutine inside a larger pipeline.

Pros and Cons

Pros

  • Fastest and simplest object removal UX in the category. Brush, wait, download.
  • Free tier without login is unusual and materially useful for one-off cleanups.
  • Pro pricing at $5/month is one of the best value paid tiers in AI image tooling.
  • Auto-detect click-to-remove on Pro is a legitimate speed-up over the brush.
  • API is priced aggressively for pipeline use.
  • Handles textured backgrounds and repetitive patterns (walls, grass, watermarks) notably better than most generic generative fill tools.

Cons

  • One-trick tool. Removes objects and nothing else. No generative fill with prompts, no upscaling, no color work.
  • The 720px free-tier cap is limiting once you notice it — most real photos are much higher resolution.
  • Complex scenes with strong geometry (bookshelves, brick patterns, tiled floors) still produce visible seams occasionally.
  • No prompt control over what the model reconstructs. If the tool guesses wrong, your options are to reshape the mask or try again.
  • Web-first UI has no desktop app or Photoshop plugin — everything runs in a browser tab.
  • Sold and resold across three owners (ClipDrop, Stability AI, Jasper) in five years — the roadmap has been visibly bumpy.

Best Use Cases

  • Social-media and content creators stripping tourists, poles, or bystanders out of otherwise good phone shots. Free tier covers this end-to-end.
  • Real-estate agents and property listings removing watermarks from stock photography or cleaning up interior shots (personal items, cables, clutter) before listing.
  • E-commerce sellers running the API to strip watermarks or clean up product shots before uploading to marketplaces.
  • Journalists and editorial teams removing incidental branding or logos from photojournalism when licensing forbids visible marks.
  • Anyone with an image and one distracting thing in it. The specific reason Cleanup.pictures still owns the "erase" search intent — it just works, fast, without commitment.

Alternatives

  • Remove.bg. The specialist for background removal, not object removal. If your task is "keep only the subject" rather than "erase this specific thing," Remove.bg is the correct pick. See /tools/remove-bg.

  • Photoroom. Product-photo workflow tool that includes AI object removal as one of many features. Better choice if you need scenes, cutouts, batch, and templates around the removal step. See /tools/photoroom.

  • Upscayl. Open-source upscaler that runs locally. Different job (increasing resolution rather than removing objects), but often the second step in a workflow that starts with Cleanup.pictures. See /tools/upscayl.

  • Photoshop Generative Fill. Adobe's generative fill inside Photoshop 2026 is materially better than Cleanup.pictures on complex scenes and gives you prompt control over the reconstruction. It is also inside a $23/month subscription and requires a Photoshop workflow. For anyone already in Creative Cloud, generative fill is the daily driver and Cleanup.pictures is a browser fallback.

  • Google Magic Eraser (Photos / Pixel). Free for Google One subscribers and now available on any phone through Google Photos. Comparable quality on simple removals, less capable on complex scenes.

  • ClipDrop Cleanup and other Stability AI endpoints. Same underlying research, different packaging. Cleanup.pictures is the consumer-friendly wrapper; ClipDrop and Stability API expose the same class of models with more flexibility for developers.

Getting Started

  1. Test on a phone photo without signing up. Go to cleanup.pictures, drop in an image with something obvious you want gone — a bystander, a sign, a shadow — and brush over it. Free tier requires zero commitment and is the honest first evaluation.
  2. Judge Pro on the HD tier, not on the model quality. The free-tier model is already good enough that Pro is mostly about resolution and auto-detect. Upload a full-resolution photo and download the HD version to know whether the extra pixels matter to your work.
  3. Use the API when it becomes a subroutine, not a task. If you are removing watermarks from ten photos per week, use the web tool. If you are removing watermarks from ten thousand, request API pricing directly — the retail Pro subscription is not the volume path.

FAQ

Is Cleanup.pictures free? The web tool is free with a 720px resolution cap and no login required. Full-resolution output requires the Pro tier at $5/month or the API.

How does Cleanup.pictures compare to Photoshop's Generative Fill? Photoshop's fill is more capable on complex scenes and gives you prompt control. Cleanup.pictures is faster for simple erasures and does not require a Photoshop subscription or workflow. For pure object removal, they are close enough that either works; for editorial retouching around the removal, Photoshop wins.

Can I use Cleanup.pictures output commercially? Yes, on Pro and Team tiers, and via the API. The free tier is intended for personal and evaluation use.

Does Cleanup.pictures work on video? No. Single-image tool only. Frame-by-frame processing via the API is possible but not the intended use, and there is no temporal consistency between frames.

Who owns Cleanup.pictures in 2026? The product is currently under Jasper after being sold from Stability AI in 2024. The underlying inpainting research still traces back to LaMa and ClipDrop's iterations on top of it.

Is there a Photoshop plugin? No. Cleanup.pictures lives entirely in the browser. This is a real limitation for anyone whose retouching workflow is Photoshop-centric — use Adobe's generative fill instead.

Verdict

Cleanup.pictures is the tool you send your less-technical friend to when they want to know how to erase someone from a photo. In 2026 it is still the fastest and cheapest way to do exactly that, with a free tier generous enough that a lot of users never upgrade, and a Pro tier at $5/month that is one of the best-value paid subscriptions in the entire AI image tooling category. If your job involves occasional cleanup — a tourist here, a watermark there, a power line in an otherwise beautiful landscape — Cleanup.pictures is the correct default, and you do not need to look further.

Where it stops being the answer is when object removal is one step in a larger workflow. Product photographers, catalog operators, and real-estate teams processing thousands of images per week are usually better off in Photoroom (for the workflow around the removal) or on the API (for the pipeline). And retouchers whose day job lives in Photoshop are already using generative fill and rarely need Cleanup.pictures at all.

The honest recommendation: keep the URL bookmarked, pay $36/year if you use it more than twice a month, and switch to the API the moment removal becomes a routine step in an automated pipeline. For everything else in the image stack, pair it with Remove.bg for background cutouts, Photoroom for e-commerce workflow, and Upscayl for local upscaling — the four tools together cover most of the AI image editing surface without a Photoshop subscription.

Explore more in the Image Generation category, or compare it against the specialist workflow tools before committing to a paid subscription you may not need.