Every European AI image tool reviewed
Paris, France
Founded 2019
Free tier / Pro from about €12.99 per month
Free tier
Best for: Sellers and brands processing product photography at volume
PhotoRoom applies AI to the part of image work that is genuinely tedious rather than the part that is fun. Cutting a product out of its background, dropping it onto white or onto a generated scene, and doing it identically across an entire catalogue is the job that stands between a seller and a listing that converts — and it is the job nobody has time for.
Batch editing is what turns that from a feature into an argument. One photograph is two minutes; four hundred photographs is a week, which is why so many marketplace listings still show a product on a kitchen table.
Processing the set in one operation, with consistent framing and background, is the difference between a catalogue that looks like a shop and one that looks like a car boot sale. The API extends the same processing into an e-commerce platform or an import pipeline, so it happens on upload rather than as a task someone remembers.
PhotoRoom operates from Paris under GDPR, so product images — frequently unreleased when they are processed — stay in EU jurisdiction. There is a free tier with Pro from about €12.99 per month. Its scope is deliberately narrow: this is commercial product imaging, not a general editor and not a generative model. For retouching, compositing or illustration, the answer is in design tools.
What PhotoRoom does well
- Batch processing across a whole catalogue at once
- API puts the same editing inside an e-commerce pipeline
- Consistent framing and background across a product set
- French company, EU processing under GDPR
- Free tier, Pro from about €12.99/month
Where PhotoRoom falls short
- Narrow scope — product imaging, not general editing
- Not a generative model
- Batch editing needs the paid tier
- Complex retouching still needs a real editor
Standout feature. Four hundred product photographs cut out in one operation — the job that decides whether a catalogue ever gets done.
Freiburg, Germany
Founded 2024
API usage-based / open-weight models free to run yourself
Open-weight models
Best for: Teams needing state-of-the-art generation with open weights or local deployment
Black Forest Labs is the strongest argument against the idea that European AI is a follower. It was founded in Freiburg in 2024 by Robin Rombach, Andreas Blattmann and Patrick Esser — the core creators of Stable Diffusion — and its FLUX models became the open-weight benchmark for image generation almost immediately.
The part that matters commercially is licensing and deployment. Open weights on some FLUX variants mean the model can be downloaded and run on your own hardware: generation without a network connection, no record anywhere of what was asked for, and fine-tuning on your own material.
Clear commercial licensing on top of that answers the question every legal team asks about generated assets, which closed consumer services tend to leave vague. FLUX Kontext extends the models into in-context editing rather than generation from scratch.
Black Forest Labs GmbH operates from Freiburg under GDPR, with API access usage-based and the open-weight models free to run yourself. The considerations are those of a young research company: the product is models and an API rather than an application with an interface, using it well assumes technical capability, and a lab that raised at this pace is moving faster than its documentation.
What Black Forest Labs does well
- FLUX models are the open-weight benchmark for image generation
- Open weights on some variants — run locally, fine-tune on your own data
- Clear commercial licensing for generated assets
- FLUX Kontext for in-context editing as well as generation
- German company, EU jurisdiction under GDPR
Where Black Forest Labs falls short
- Models and an API, not a finished application
- Assumes technical capability to use well
- Self-hosting needs GPU capacity
- Young company moving faster than its documentation
Standout feature. The people who built Stable Diffusion left and founded their next lab in Freiburg — and FLUX is why that matters.
London, United Kingdom
Founded 2019
Models free for many uses / API usage-based / enterprise licensing above revenue thresholds
Downloadable models
Best for: Teams that want to fine-tune image models on their own data
Stability AI released Stable Diffusion, which is the single event that made image generation something anyone could run rather than something you queued for on someone else's service. The whole open image ecosystem — the tooling, the fine-tunes, the interfaces, the research built on top — exists because those weights were published.
That remains the reason to use it. Downloadable weights with community and enterprise licences mean a team can fine-tune on its own material and deploy the result on its own hardware: a product photography model trained on your catalogue, an illustration model trained on your house style, running where the images never leave. Its range extends past images into video and audio generation, with API access for teams that would rather not run infrastructure.
Stability AI Ltd operates from London, which is a UK adequacy position rather than EU establishment — relevant only if procurement names EU member states, and removable entirely by running the models locally. Models are free for many uses with enterprise licensing above revenue thresholds, so check where your use falls before building on it. FLUX has taken the quality lead on newer work; Stable Diffusion's advantage is the depth of ecosystem around it.
What Stability AI does well
- Downloadable weights with fine-tuning support
- By far the largest ecosystem of community tooling
- Video and audio generation alongside images
- Free for many uses, API available for hosted access
- Models run entirely on your own hardware
Where Stability AI falls short
- UK jurisdiction rather than EU establishment
- Enterprise licensing applies above revenue thresholds
- FLUX leads on newer model quality
- Fine-tuning needs GPU capacity and expertise
Standout feature. The weights that started the open image ecosystem — and still the deepest tooling around any model here.
London, United Kingdom
Founded 2022
Free tier / paid from about $12 per month
Free tier
Best for: Designers who need editable assets rather than flat images
Recraft generates real vectors, and that single difference decides whether the output is an asset or a picture. A generated PNG is finished: scale it and it blurs, change the brand colour and you regenerate and hope. Recraft produces editable SVG paths that open in Illustrator or Figma and get adjusted like anything a designer drew.
Custom brand styles are the second idea and they solve the problem that makes generated assets unusable in brand work. Define a style once and every subsequent generation matches it, so a set of thirty icons reads as a set rather than as thirty separate experiments in thirty adjacent aesthetics. Icon and illustration sets, an infinite canvas, mockups and an API round it out.
Recraft AI Limited was founded in London in 2022 by machine learning scientist Anna Veronika Dorogush, with data in the UK and EU — a UK adequacy position rather than EU establishment. Free tier with paid plans from about $12 per month. It is not the tool for photorealistic imagery, where FLUX and Stable Diffusion are far stronger, and it cannot be self-hosted.
What Recraft does well
- Generates genuine editable SVG vectors
- Custom brand styles keep a set visually consistent
- Icon and illustration sets, infinite canvas, mockups
- API for embedding generation in a workflow
- Free tier, paid from about $12/month
Where Recraft falls short
- Not aimed at photorealistic imagery
- UK jurisdiction rather than EU establishment
- No self-hosting
- Vector output needs a designer to be worth it
Standout feature. Editable SVG out of a prompt — the difference between an asset your designer adjusts and one they redraw.
Malaga, Spain
Founded 2010
Free tier with attribution / Premium from about €12 per month
Free tier
Best for: Marketing and design teams needing generated and stock assets together
Freepik AI answers a practical problem rather than a technical one: generation does not always produce the right thing, and when it does not, you need something already drawn. Pairing AI image and video generation with a very large library of licensed stock assets, vectors and templates under one subscription means the fallback is a click away rather than a separate purchase and a separate licence to read.
It runs multiple underlying models rather than one, so the generation itself is not tied to a single lab's aesthetic or a single lab's roadmap, and it adds the practical operations around images — background removal, upscaling — plus an API for teams putting it into a workflow.
Freepik Company S.L. has operated from Málaga since 2010, which makes it one of the more established companies in this category rather than a recent arrival, with EU processing under GDPR. There is a free tier with attribution and Premium from about €12 per month. The trade-off is depth: it is a platform assembling capability rather than a lab producing it, so on pure generation quality the model providers lead, and there is no self-hosting.
What Freepik AI does well
- Generation and a large licensed stock library in one subscription
- Multiple underlying models rather than one lab's output
- Background removal, upscaling and an API included
- Spanish company operating since 2010, EU processing
- Free tier with attribution, Premium from about €12/month
Where Freepik AI falls short
- A platform assembling capability rather than producing it
- Generation quality trails the model labs
- No self-hosting
- Free tier requires attribution
Standout feature. When the prompt does not work, something already drawn is one click away — under the same licence and the same bill.
Paris, France
Founded 2020
Free tier with limits / Pro from about €9 per month / API priced separately
Free tier
Best for: Practical image fixes rather than generating images from scratch
Clipdrop is a set of focused operations rather than a generative playground, and the list is the product: cleanup to remove an object convincingly, uncrop to extend an image past its frame, relight to change how a photograph is lit after the fact, background removal, and upscaling. Each does one thing that a photographer or marketer actually needs done, built on open generation models underneath.
That framing makes it complementary to the model labs rather than competitive with them. FLUX and Stable Diffusion make images; Clipdrop fixes the ones you already have. Batch processing, an API and mobile and desktop applications cover the different ways that work arrives.
The jurisdictional position needs stating plainly, because it is one of two caveats in this category: the team is French and based in Paris with EU processing under GDPR, and the company is owned by Jasper, an American business — that is neither a US-established processor nor a European-owned one, and which of those matters is a procurement question rather than a technical one.
The other caveat, Luminar Neo, is a Cyprus-registered company whose own terms of use run under New York law. Free tier with limits, Pro from about €9 per month, API priced separately.
What Clipdrop does well
- Focused operations: cleanup, uncrop, relight, background removal, upscaling
- Batch processing and an API for volume work
- Mobile and desktop apps alongside the web tools
- French team, EU processing under GDPR
- Free tier, Pro from about €9/month
Where Clipdrop falls short
- Owned by Jasper, a US company
- Not a generative model in its own right
- No self-hosting
- Free tier limits bite quickly on real work
Standout feature. Relight a photograph after it was taken — the operation on this list that no amount of prompting replaces.
Prague, Czech Republic
Free with advertising / paid tier removes ads
Free, no account required
Best for: Anyone who needs Photoshop for ten minutes and does not have it
Photopea is a full raster and vector editor that runs in a browser tab and opens PSD files — layers, masks, adjustment layers and smart objects intact. It is the answer to a situation almost everyone hits: a designer sends a PSD, you need one change, and buying Creative Cloud for that is absurd.
What makes it belong in an AI image category rather than only in design tools is the practical overlap. The generated and processed images from every other tool here still need cropping, layering, retouching and export at the right size, and Photopea is where that happens for anyone without an Adobe subscription. It also handles the file formats those tools produce and the ones clients send.
Its architecture is the privacy argument and it is unusually strong: the editor runs in your browser and the files stay on your machine rather than being uploaded to be processed. For a designer working on unreleased material, that is a categorically different position from every cloud editor.
Photopea is made in Prague, free with advertising and a paid tier that removes the ads. The limits: it is a browser application, so very large files strain it, there is no real-time collaboration, and the interface deliberately imitates Photoshop, which helps if you know Photoshop and not otherwise.
What Photopea does well
- Opens PSD files with layers and masks intact
- Runs in the browser — files are not uploaded anywhere
- Free with no account required
- Handles raster and vector in one editor
- Czech company, EU jurisdiction
Where Photopea falls short
- Browser application strains on very large files
- No real-time collaboration
- Free tier is ad-supported
- Interface assumes Photoshop familiarity
Standout feature. The file never leaves your machine — a full editor with no upload, which no cloud tool here can claim.
Milan, Italy
Free download with limited free enhancements; auto-renewing subscription (weekly, monthly or annual) via the App Store or Google Play, price shown at checkout and varies by region
Free download, limited free enhancements
Best for: Restoring old photos and sharpening low-quality faces and video
Remini does one job that the generation-focused tools in this category don't touch: taking a blurry, low-resolution or damaged photo and making the face in it look sharp again. Upload an old family photo with scratches and faded colour, or a screenshot-quality selfie, and it reconstructs detail that a generic upscaler guesses at badly — skin texture, eyes, hair — because the model is trained specifically on faces rather than images in general.
That focus is the whole product. Photo restoration, face enhancement and colourisation of old black-and-white photos sit alongside a matching video enhancer, so the same reconstruction applies to old home footage. It is a mobile-first app rather than a batch-processing platform for a catalogue, which puts it in a different use case from PhotoRoom: this is for the one photo that matters, not four hundred product shots.
Remini is built by AI Creativity S.r.l., a Bending Spoons company operating from Milan under GDPR — Bending Spoons S.p.A. is named as the data controller on its own privacy documentation. The app is free to download with limited free enhancements, then an auto-renewing subscription through the App Store or Google Play at a price that varies by region. For anyone restoring inherited photographs or old video, it does a job none of the generation-first tools here are built for.
What Remini does well
- Face-specific reconstruction outperforms generic upscalers
- Restores scratches, fading and colourises old photos
- Video enhancement uses the same technology
- Italian company, EU processing under GDPR
- Free download with limited free enhancements
Where Remini falls short
- Mobile app, not a batch or catalogue tool
- Subscription pricing not published on the website itself
- Not a generation or editing tool beyond restoration
- Heavier processing needs a paid plan
Standout feature. A blurry face from a twenty-year-old photo, reconstructed rather than just upscaled — the job the rest of this category doesn't attempt.
Białystok, Poland
Premium plan (single photo) about $13.95, per PhotoAiD's own comparison table; free to prepare, pay only once you accept the result
Free to prepare and preview; pay only if you accept the result
Best for: Passport, visa and ID photos that have to pass official requirements
PhotoAiD solves a narrow, specific problem: a passport or visa photo has hard requirements — head size, background colour, expression, lighting — and getting it wrong means a rejected application. It checks a phone photo against those official rules in real time, crops and resizes it automatically, and removes and replaces the background to the required colour, so the output is built to pass rather than just look presentable.
The commercial model matches the job: prepare and preview the photo for free, and pay only once the result is something you would actually submit. Its own comparison against other passport-photo apps lists a Premium plan at about $13.95, with a money-back guarantee if the photo is rejected by the authorities — a narrower promise than a general image editor makes, but a more specific one.
PhotoAiD S.A. operates from Białystok, Poland, registered in the National Court Register (KRS 0000941997). Its scope is deliberately narrow — this is not a photo editor or a generation tool, and outside ID and passport photography there is nothing else here for it to do. For the specific job of a compliant biometric photo without a trip to a photo booth, it is the tool built for exactly that.
What PhotoAiD does well
- Checks photos against official passport/visa/ID rules
- Free to prepare and preview before paying
- Money-back guarantee if a photo is rejected
- Polish company, EU jurisdiction
- 16M+ ID photos taken through the service
Where PhotoAiD falls short
- Narrow scope — ID and passport photos only
- Not a general photo editor or generation tool
- Pay-per-photo rather than a subscription
- Premium plan needed for the full result
Standout feature. A passport photo checked against the actual official rules before you pay for it — not just cropped and hoped for.
Limassol, Cyprus
Free trial download; full editing unlocked via a one-time Lifetime License or a Cross-device Lifetime License, price shown at checkout and varies by region
Free trial download
Best for: Photographers who want generative editing inside a full RAW workflow
Luminar Neo puts generative AI editing inside a conventional RAW photo editor rather than a standalone web tool. GenErase, GenExpand and GenSwap remove objects, extend a frame or replace elements with AI-generated content that matches the surrounding image — the same category of tool as Photoshop's generative fill, built into an editor that also does layers, masking, RAW processing from over a thousand camera models and full portrait and landscape retouching.
Sky AI, Atmosphere AI and the portrait tools (Skin AI, Face AI, Body AI) automate the retouching a photographer would otherwise do manually, and batch tools — panorama stitching, focus stacking, HDR merge across up to a thousand images — cover volume work the web-based tools in this category are not built for. It is a desktop application for Windows and macOS, downloaded and licensed rather than used in a browser.
"Skylum" is a trading name of Shinyfields Limited, a Cyprus company, and its own terms of use name US affiliates — Skylum Software USA, Inc. and Photolemur, Inc. — and state that those terms are governed by New York law, not Cyprus or EU law. That is a real jurisdictional wrinkle worth knowing before it comes up in procurement, alongside Clipdrop's French-team-owned-by-a-US-company position. The app is free to try, with full editing unlocked through a one-time Lifetime License.
What Luminar Neo does well
- Generative erase, expand and swap inside a full RAW editor
- Automated sky replacement and portrait retouching
- Batch HDR, panorama and focus-stacking tools
- RAW support for 1,000+ camera models
- One-time Lifetime License rather than a forced subscription
Where Luminar Neo falls short
- Cyprus entity, but terms of use run under New York, US law
- Named US affiliates alongside the Cyprus company
- Desktop app, not a lightweight web tool
- Exact pricing not published on the website itself
Standout feature. Generative erase and expand inside a real RAW workflow — Photoshop's generative fill logic, but built for photographers rather than designers.
Estonia
Free tier with 20 credits, no card required; Creator $19/month, Premium $79/month, Restless $199/month, credits carry over while the plan is active
20 free credits, no credit card required
Best for: Fast, credit-priced AI edits on product photos and portraits in the browser
Pincel packages a wide set of narrow AI editing tools — object remover, generative fill, background remover, AI upscaler, AI portraits, virtual try-on, clothes swap — behind one browser interface with no installation, aimed at commercial content: product photos, marketing visuals and portraits that need to look brand-safe rather than experimental.
The pricing is transparent and usage-based rather than a flat subscription: twenty free credits with no card required, then Creator at $19 a month, Premium at $79 and Restless at $199, each with a monthly credit allowance that carries over while the plan stays active.
For a small team running frequent but unpredictable volumes of product or marketing images, paying by the edit rather than committing to a seat licence is a genuinely different shape of pricing than most of this category.
MULTIMEDIA DESIGN OÜ operates Pincel as a company registered in Estonia, with its terms of use and privacy policy both governed by Estonian law. The trade-off for the breadth of tools is depth: this is fifty-plus focused operations rather than one deep generative model, so for pure image generation from a prompt the model labs in this category go further.
What Pincel does well
- 50+ browser-based AI editing tools, no installs
- Object remover, generative fill and background remover
- Credit-based pricing that carries over month to month
- 20 free credits, no credit card required
- Estonian company, EU jurisdiction
Where Pincel falls short
- Breadth over depth compared with a dedicated generation model
- Credit consumption varies by tool and complexity
- Not aimed at from-scratch artistic generation
- Exact plan limits need checking against actual usage
Standout feature. Twenty free credits and fifty-plus AI tools with no install — a browser toolbox priced by the edit rather than the seat.
Limassol, Cyprus
Founded 2004
From about €40.95 for a licence (EUR pricing shown to EU customers, price varies by plan and region); free trial download available
Free trial download
Best for: A familiar desktop photo editor with AI retouching bolted on
Movavi Picverse Photo Editor is a conventional desktop photo editor — colour correction, presets, effects — with AI features layered on top rather than built around them: AI-assisted object removal that generates a natural-looking replacement background, AI colour correction, and colourisation for old photos. It is closer in spirit to a traditional editor like PaintShop Pro than to a generation-first tool, which makes it an easier switch for someone who already knows that kind of software.
That familiarity is the whole pitch. Thirty-four presets and a large effects library cover the fast, non-technical edits most people actually make, and the AI object removal handles the one job a manual eraser tool does badly — filling in what was behind the thing you deleted. It is a downloaded Windows and Mac application rather than a browser tool, sold as a licence rather than a subscription.
Movavi Software Limited is registered in Limassol, Cyprus, and the company itself dates to 2004, by its own account, making it one of the longer-running vendors in this category even though its AI features are recent. Pricing shown to EU customers runs in EUR, from about €40.95 for a licence, with a free trial download to try it first. Where it does not compete is generative depth: GenExpand-style frame extension and prompt-driven generation are not part of what it does.
What Movavi Picverse Photo Editor does well
- AI object removal with generated background fill
- AI colour correction and old-photo colourisation
- 34+ presets and effects for fast edits
- Company operating since 2004, EUR pricing for EU customers
- One-time licence rather than a forced subscription
Where Movavi Picverse Photo Editor falls short
- Less generative depth than Luminar Neo or the model labs
- Desktop application, not a browser or mobile tool
- Interface aimed at traditional editing more than AI-first workflows
- Cyprus entity — check the terms that apply to your purchase
Standout feature. AI object removal bolted onto a familiar desktop editor — the easiest switch here for someone who already knows this kind of software.
Karlsruhe, Germany
Free tier with limited free images; additional credit packs and subscription plans for higher usage
Free tier with limited free images
Best for: Quick AI upscaling and restoration from a long-established German software company
Nero AI is a browser-based upscaling and restoration tool from a company most people know for burning CDs two decades ago. Image upscaling multiplies resolution using AI rather than simple interpolation, and face enhancement and photo restoration apply the same kind of reconstruction Remini offers, packaged instead as one tool inside a broader Nero AI suite that also covers video.
Its position in this category is a utility rather than a destination: a free tier handles occasional upscaling or restoration jobs, with credit packs and subscription plans for heavier use. It is not aimed at generation from a prompt or at product-photo batch work, and it does not try to be — it does one well-understood operation and lets you leave.
Nero AG is a German stock company registered in Karlsruhe, Baden (Mannheim commercial register, HRB 362519), long predating the current wave of AI image startups. That corporate longevity is a different kind of reassurance from a young lab's funding round: Nero has been operating software businesses in Germany for decades, under GDPR, and its AI tools are an extension of that rather than a new venture built from nothing.
What Nero AI does well
- AI upscaling to several times the original resolution
- Face enhancement and photo restoration in the same suite
- Free tier for occasional use
- Long-established German company, GDPR jurisdiction
- Video tools alongside the image tools
Where Nero AI falls short
- Utility tool rather than a generation or editing destination
- Credit-based paid tiers for heavier use
- Less specialised than Remini for face restoration specifically
- Exact pricing not published in plain text on the site
Standout feature. A two-decade-old German software company doing AI upscaling — corporate longevity where most of this category is a recent startup.
