Pixelcut

What is Pixelcut?

Pixelcut is an AI photo editor built by Pixelcut Inc., a company registered at 490 43rd St, Suite 210 in Oakland, California. Where a general-purpose design suite tries to serve every creative discipline, Pixelcut concentrates on a much narrower job: turning ordinary photographs of physical products into listing images and social creatives that look like they came out of a rented studio. That focus shows up everywhere in the product, from a background remover that markets sub-three-second cutouts to a bulk editor sized for an entire catalogue rather than a single hero shot.

The company sits in an unusual competitive position. TechCrunch, writing in January 2024 about the wider AI photo editing market, noted that Pixelcut had raised only $250k according to PitchBook — a fraction of what its better-known rivals had banked. Yet the product reached genuine scale on the back of mobile distribution rather than venture money. That history explains a lot about how the tool feels: pragmatic, aggressively free at the entry point, and optimised for someone photographing inventory on a phone rather than an art director working from a colour-calibrated monitor.

Functionally, Pixelcut spans three overlapping layers. The first is corrective editing — removing backgrounds, erasing unwanted objects, sharpening low-resolution files. The second is generative — inventing new backgrounds, extending a cropped frame, placing garments on synthetic models. The third is operational — running any of those operations across thousands of files at once, or calling them from your own code. Most competitors do one or two of these well. Pixelcut's argument is that a small seller needs all three and cannot afford to buy them separately.

Core Features

• One-click background removal: The background remover detects the subject, maps its edges and returns a cutout in what the product page describes as less than three seconds, with no sign-up required for a basic export. You can download a transparent PNG or swap straight to a white background, and a Refine brush lets you repair edges by hand where the model struggles. The marketing leans hard on difficult edges — flyaway hair, jewellery filigree, thin spectacle frames and fur — because those are exactly where cheap cutout tools leave halos.
• Image upscaling to 16K: The upscaler pushes images to 4K, 8K or even 16K, and its feature page states plainly that uploaded images are removed after processing. Two modes are offered: a Fast Upscale model that runs automatically on upload, and a Creative Upscale that lets you trade fidelity against invention by tuning creativity and resemblance levels. The first is right for restoring a usable listing photo; the second is closer to a reinterpretation and needs checking before it touches a product page.
• Magic Eraser and Pro Eraser: The cleanup tools remove objects, people, text and watermarks by brushing over them. Pixelcut positions this for photobombers and background clutter, but the more commercially valuable use is stripping price stickers, reflections and stray cables out of product shots that would otherwise need reshooting.
• Generative Fill and Uncrop: Generative Fill works by selecting a region and describing what should replace it, so you can add objects, swap elements or rebuild a background from a prompt. Uncrop handles the opposite problem — extending the canvas in any direction to hit a marketplace's required aspect ratio, with preset ratios and a regenerate button to cycle through alternative completions rather than accepting the first guess.
• AI product scenes and AI Actors: Product Showcase takes a product photo, removes its background automatically, and generates photorealistic environments from a text description such as a marble countertop in morning light. AI Actors extends this to people, generating consistent recurring personas that appear with the same face and styling across a campaign, alongside Ghost Mannequin for the floating-apparel look and Virtual Try-On for placing garments on generated models.
• Bulk editing at catalogue scale: The Bulk Image Editor processes up to 10,000 images in a single batch, pulls entire folders straight from Dropbox, and returns the finished set as one ZIP download. Batch tools cover background removal, resizing, upscaling, watermarks, logo overlays, shadows, cropping, expansion and background swapping, and parameters are set once then applied across every file.
• Developer API and an MCP server: Pixelcut exposes background removal, upscaling, background generation and try-on as REST endpoints. Separately, Pixelcut publishes a hosted MCP server at mcp.pixelcut.ai that plugs its remove background, generate image, upscale, generate video, erase object and asset library tools directly into Claude, Cowork and Claude Code, so an assistant can run the edits conversationally without you wiring up an SDK.

Use Cases

1. Marketplace listing preparation for a small seller: A seller photographing stock at home ends up with inconsistent backgrounds, mixed lighting and varying aspect ratios, which looks amateurish beside competitors on the same results page. The workflow here is upload, remove background, place on white or a generated scene, uncrop to the marketplace's required ratio, then upscale anything shot at low resolution. Because the free tier exports without a watermark, a seller can validate the output quality on a handful of items before paying for anything.
2. Catalogue migration across storefronts: Selling the same inventory on multiple channels means the same photograph needs several different treatments — square crops for one marketplace, white backgrounds for another, horizontal layouts for social. Doing that by hand is the definition of unbillable work. Running it through the bulk editor turns a multi-day task into one upload, one parameter set and one ZIP file, which is where the 10,000-image ceiling actually earns its keep.
3. Social and ad creative on a weekly cadence: Content teams shipping Reels, Shorts and paid social need volume more than they need perfection on any single asset. Generative Fill and AI Backgrounds let one product photograph become a dozen seasonal variants without a reshoot, and the Video Background Remover extends the same logic to motion clips.
4. Apparel merchandising without a model booking: Fashion sellers face a specific and expensive problem — garments photograph poorly flat, and hiring models for every SKU is out of reach for small brands. Ghost Mannequin produces the floating-apparel look, while Virtual Try-On places a garment on a generated person from a single image of each. Both are best treated as merchandising aids that still need a human eye on fit and drape before publication.
5. Embedding image editing into another product: A marketplace, CRM or listing tool can offer background removal to its own users by calling the API rather than building and hosting a segmentation model. Because billing is per credit with no seat commitment, the cost scales with actual usage, which suits a feature whose adoption you cannot forecast.
6. Assistant-driven creative pipelines: With the MCP server connected, a request phrased in plain language — generate lifestyle shots, remove the backgrounds, upscale the results — can be chained into an automated pipeline inside Claude. This is a genuinely different interaction model from opening an editor, and it suits repetitive production work where the steps are known and only the inputs change.

How to use Pixelcut

1. Open pixelcut.ai and pick a tool directly from the navigation — background remover, upscaler, Magic Eraser, Generative Fill or Uncrop. Basic background removal and upscaling run without an account, so you can test output quality before registering.
2. Upload your image by dragging it onto the page or using the upload button. Every uploader states the same envelope: JPG, PNG and HEIC files up to 100MB and 8000 by 8000 pixels. Files outside that range need resizing or converting first.
3. Let the AI run, then inspect the result at full size rather than in the thumbnail. For cutouts, use the Refine brush on hair, fur and transparent materials, which is where automated segmentation is most likely to leave artefacts.
4. Chain further edits without re-uploading — swap the background, extend the canvas with Uncrop, then upscale. Keeping everything in one session avoids the quality loss that comes from repeatedly exporting and re-importing compressed files.
5. For volume work, switch to the Bulk Image Editor, import from your desktop or Dropbox, choose a batch tool, set the parameters once and let it run across the whole set before downloading the ZIP.
6. Export at the resolution your channel requires. Free downloads are available without a watermark, but higher-resolution outputs — 4K video from the background remover, for example — sit behind the Pro plan.

Tips & Best Practices

• Shoot with the AI in mind. A subject with clear edge separation from its background gives markedly better cutouts than one photographed against a similarly coloured wall. Two minutes of attention at capture time saves more than any amount of refinement afterwards.
• Treat Fast Upscale and Creative Upscale as different tools with different risks. Fast Upscale restores detail and is safe for listings. Creative Upscale invents plausible detail that was never in the original, which is fine for artwork and dangerous for anything a customer will compare against the physical item they receive.
• Write scene prompts in concrete nouns rather than adjectives. Marble countertop with morning light from the left produces a more usable result than something premium and elegant, because the model has specific surfaces and lighting directions to work with.
• Prove your batch on a sample first. Run ten representative images through the settings you intend to use, check the edge cases, and only then commit the full catalogue — otherwise a wrong parameter multiplies across thousands of files.
• Keep your originals. Every operation here is destructive in the sense that you get a new file, not a reversible edit history, so archive the source photographs somewhere outside Pixelcut before you start replacing them.
• Watch your credit consumption per model. Generation routes through third-party models with very different costs, so the same monthly allowance stretches far further on a lightweight model than on a flagship one.
• If you are building a pipeline, decide early between the REST API and the MCP server. The API suits deterministic, server-side batch jobs; the MCP route suits exploratory work where a person is still in the loop making judgement calls.

Who is Pixelcut for?

• Independent e-commerce sellers: The core audience. Someone listing on a marketplace who needs studio-looking images from phone photographs, without a designer, a lightbox or a monthly software budget in the hundreds.
• Small brand and marketing teams: Teams that need creative volume across channels and value throughput over pixel-perfect craft on any single asset. Seat allocation matters here, since plans bundle team members rather than charging per seat.
• Social media creators and content managers: Anyone producing a steady stream of posts who benefits from fast background swaps, object removal and canvas extension on a phone.
• Apparel and accessory sellers: The Ghost Mannequin and Virtual Try-On features address a merchandising problem specific to clothing that generic editors do not solve.
• Developers integrating image editing: Teams that want background removal or upscaling inside their own product and would rather pay per operation than build and host segmentation infrastructure.
• Dropshippers and catalogue resellers: Users who receive supplier photographs of inconsistent quality and need to normalise them into a coherent-looking storefront quickly.
• Occasional users with a one-off job: Because core tools run free without registration, someone with a single image to fix can get it done without creating an account at all.

Platforms

• Web: The full product runs in a desktop or mobile browser, including the bulk editor, which is the natural home for large batches.
• iOS: Pixelcut AI Photo Editor has been on the App Store since November 2020, is free to download, and requires iOS 15.0 or later. The iOS app carries roughly 4.74 stars across about 217,806 ratings and has been on the App Store since November 2020.
• Android: The Google Play listing, published under the bundle identifier com.circular.pixels, has passed 10 million downloads with an average of roughly 4.21 stars from about 239,195 ratings, and carries an Everyone content rating.
• API: A REST API covers background removal, upscaling, background generation, try-on and outpaint, with keys issued from the dashboard.
• MCP server: A hosted connector at mcp.pixelcut.ai exposes the toolkit to Claude, Cowork and Claude Code through a single custom connector URL, with no API keys to manage separately.
• Interface languages: The website offers English, Spanish, French, Italian, Portuguese (Brazil), Indonesian, Thai, German, Hindi, Japanese, Korean, Turkish, Russian and Vietnamese.

Pricing & Plans

The free tier is unusually generous at the point of entry and unusually vague about its ceiling. It provides limited background removal and limited upscaling, and — importantly for anyone evaluating cheaper rivals — free export without a watermark. The word doing the heavy lifting is "limited": the comparison table marks the free allowances for background removal, Magic Eraser and Fast Upscaler as limited without publishing the daily number, so plan on the free tier being adequate for evaluation and occasional jobs rather than production.

Paid plans are billed monthly or annually, with an annual subscription saving 20 percent. Pro costs $10 per month and adds 600 AI credits per month, unlimited background removal, unlimited upscaling, 1,000 batch exports per month, a three-person team and a commercial license. Business costs $30 per month and raises that to 3,600 monthly credits, 2,000 batch exports and a ten-person team. Seats are bundled rather than billed individually, which is a meaningful cost difference for a small team compared with per-seat competitors. Generative features draw on credits, and the pricing page publishes a long conversion table showing how far a month's allowance goes on each underlying model, with video usage quoted per five seconds of output.

The API is priced separately and transparently. API credits are billed at $0.01 each in monthly bundles that run from 1,000 credits for $10 up to 1,000,000 credits for $10,000, with background removal costing 5 credits per operation and upscaling, background generation, try-on and outpaint each costing 10. Volume beyond a million credits a month is handled through a sales conversation. Checkout runs through Stripe and subscriptions can be cancelled at any time.

Alternatives

• Photoroom: The closest direct competitor, also focused on e-commerce product imagery with strong mobile apps. It is the better-funded of the two, and the choice between them usually comes down to output preference on your specific product category rather than feature checklists.
• Adobe Express and Photoshop: Adobe offers deeper, more controllable editing and mature colour management, at the cost of a steeper learning curve and higher price. Appropriate when you have a designer; overkill when you need 400 listing images by Friday.
• Canva: Stronger as a template-driven layout and brand system tool, weaker as a precision cutout engine. Many teams use Canva for composition and something like Pixelcut for the cutouts that feed into it.
• remove.bg: A narrower, single-purpose cutout service with a well-established API. Worth considering if background removal is genuinely all you need and you value that focus.
• Running open-source segmentation yourself: Self-hosting a segmentation model removes per-image costs and keeps data in-house, but you take on inference infrastructure, model updates and quality tuning. Only sensible at high sustained volume with engineering capacity to match.

Limitations & Considerations

• The customer-satisfaction picture is genuinely split, and you should know it before subscribing. The app stores are positive, with 4.74 stars on iOS and 4.21 on Android across hundreds of thousands of ratings. Trustpilot tells a very different story: the pixelcut.ai profile sits at 1.7 out of 5 across 70 reviews, with 82 percent of them one-star and the complaint tags clustering around subscription, refund, cancellation and payment. Note the sample sizes differ by three orders of magnitude and the Trustpilot profile is unclaimed with no review invitations, so it skews toward people motivated to complain. Read it as a billing-experience warning rather than a verdict on output quality: subscribe deliberately, check the renewal terms, and keep your own record of what you signed up for.
• Hard input limits apply everywhere. Every uploader states the same envelope: JPG, PNG and HEIC files up to 100MB and 8000 by 8000 pixels. RAW files, TIFFs and oversized scans need converting first, and there is no published support for layered formats.
• Web and mobile batch capacity are not the same. The mobile Batch Editor is capped at 100 photos at once, an order of magnitude below the 10,000 the web bulk editor accepts. If batch volume is your reason for subscribing, plan to work in a browser.
• Free-tier ceilings are not published as numbers. "Limited" appears in place of a figure for the free background remover, Magic Eraser and Fast Upscaler, so you cannot budget against it in advance.
• Generative output needs review before it reaches a customer. AI-generated backgrounds, extended canvases and try-on renders can misrepresent scale, texture or fit. For regulated categories, or anywhere a buyer will compare the image against the delivered item, treat generated scenes as illustrative and check them against marketplace image policies.
• Rights are clear on your own content but restricted on model training. You keep copyright of the images you process through the API and may use them commercially, but you may not use the API to build training data for your own models. The Terms say that by providing content you grant Pixelcut a non-exclusive, fully paid-up and royalty-free license to process it solely for the operation and maintenance of the service.
• Image retention beyond the upscaler is not specified. The upscaler page states that uploaded images are removed after processing, but the privacy policy itself is a general template that keeps data for as long as the collection purpose requires, without publishing a specific retention window for uploaded images or clarifying whether they are used to train models. If you handle client or unreleased product imagery under contract, ask support in writing before uploading. (To be verified)
• The service is restricted to adults. The Terms state that to access and use Pixelcut the user must be an adult under applicable law, and where the app was obtained through the Apple App Store, Apple may enforce those Terms as a third-party beneficiary.

FAQ

Q1. Is Pixelcut free to use?

Yes, there is a genuine free tier with limited background removal and limited upscaling, and free exports carry no watermark. The word "limited" is not quantified on the pricing page, so treat the free tier as suitable for evaluation and light use rather than production volume.

Q2. What does Pro actually cost and what changes?

Pro costs $10 per month and adds 600 AI credits per month, unlimited background removal, unlimited upscaling, 1,000 batch exports per month, a three-person team and a commercial license. Business is $30 per month with 3,600 credits, 2,000 batch exports and ten seats, and annual billing saves 20 percent.

Q3. Can I use the output commercially?

The Pro and Business plans include a commercial license, and for API users Pixelcut confirms you retain copyright of images you process and may use them commercially. The one explicit prohibition is using the API to create training data for your own models.

Q4. What file types and sizes can I upload?

JPG, PNG and HEIC, up to 100MB and 8000 by 8000 pixels. Anything larger, or in a format like RAW or TIFF, has to be converted or resized before upload.

Q5. How many images can I edit at once?

The web Bulk Image Editor handles up to 10,000 images in a single batch, imports folders from Dropbox and returns a ZIP. The mobile app's batch editor is limited to 100 photos at a time, so large jobs belong in a browser.

Q6. Is there an API, and how is it priced?

Yes. API credits are billed at $0.01 each in monthly bundles that run from 1,000 credits for $10 up to 1,000,000 credits for $10,000, with background removal costing 5 credits per operation and upscaling, background generation, try-on and outpaint each costing 10. Beyond a million credits a month, volume pricing is arranged directly.

Q7. Can I use Pixelcut inside Claude?

Yes. Pixelcut publishes a hosted MCP server at mcp.pixelcut.ai that plugs its remove background, generate image, upscale, generate video, erase object and asset library tools directly into Claude, Cowork and Claude Code. You add it once as a custom connector; there are no separate API keys to manage.

Q8. What happens to the images I upload?

The upscaler page states that uploaded images are removed after processing. The broader privacy policy is a general template that does not publish a specific retention period for uploaded images or address model training directly, so for sensitive or client-owned material it is worth confirming the specifics with support first. (To be verified)

Q9. Why are the app store ratings so much better than the Trustpilot score?

They measure different things and different populations. The app stores aggregate hundreds of thousands of ratings from people judging the editing experience, while the unclaimed Trustpilot profile collects 70 self-selected reviews dominated by billing, refund and cancellation disputes. Both are real signals: the product generally satisfies on output, and the subscription flow is where complaints concentrate.

Q10. Does Pixelcut handle video as well as images?

Partly. A Video Background Remover launched on January 8, 2026 and exports WebM with real transparency or MP4 on a solid colour, with free downloads capped at HD 720p and 4K reserved for Pro. There is also an AI video generator, but the product's depth remains in still imagery.