Palette
Palette turns black-and-white photographs into realistic color images in seconds, using 21+ named color filters plus keyword and reference-image tuning. A permanent free plan lets you test it without signing up, watermarked at 500x500 pixels.
What is Palette?
Palette is a single-purpose tool: it puts realistic color onto black-and-white photographs. The homepage promise is to let you see your photos in realistic colors, and the workflow it displays has three steps, upload or drag in one or several images, pick a filter, then download the result or adjust the colors with keywords.
The filter library carries most of the product: 21+ named looks including Classic Palette, Bright Essence, Colorful Past, Dusk Dream, Nature's Palette, Human Hue, Grainy Cinema, Nostalgia Pop, Ocean Mood and Analog Wilderness, plus a separate Modern Filters group. Colorization can also be steered by automatic prompts, custom keywords or a reference image, then finished with automatic color correction, white balance, color temperature and saturation. A prompt generator can produce up to 120 variations to test on the same picture. No account is needed to try any of it, the site advertising a free online tool with no sign-up needed.
The engine is proprietary, developed by a former Google resident and presented as a second generation of AI colorization, returning an image in 100-300 milliseconds with dynamic skin tones, vivid colors and careful handling of small details. Palette claims 6.6M+ users, 54 million photos already colorized, a 4.9-star rating and 150+ testimonials, alongside named endorsements from Kevin Kelly, founding editor of Wired, from the Photoshop channel PiXimperfect and from the AI channel Bycloud, and adoption by professional colorists at large production companies. A homepage comparison sets it against unnamed known alternatives on color accuracy, interface simplicity, detail capture, batch upload and one-off pricing.
Developers get a documented photo and video API on docs.palette.fm, distributed through RapidAPI, with bulk colorization above one million images per day by script and a beta video API covering SD, Full-HD and 4K frames. Partners extend the reach: simplified video colorization through Neural Love, and a Windows desktop application from Covenant built on the Palette engine.
Technical limits are stated plainly: 10 MB per image, up to 4K through the API and 5000x5000 pixels in the app, with very broad format support including jpg, png, tiff, webp, heic, psd, gif, avif, bmp and ico. Video has no native temporal consistency, the site suggesting you reuse the previous frame as a reference with the same prompt on short scenes.
What it does
- Colorize a black-and-white photograph in seconds, the engine returning an image in 100-300 ms
- Apply one of 21+ named color filters, from Classic Palette and Nostalgia Pop to Ocean Mood and Analog Wilderness
- Refine the colorization with keywords or a written description
- Colorize from a reference image so a series keeps the same look
- Process several images at once, from multi-file drag and drop up to a million images a day by script
- Colorize video frames in SD, Full-HD or 4K through the video API, still in beta
- Retouch the result with automatic color correction, white balance, color temperature and saturation
When to use Palette / When not to
A quick filter to help you decide if Palette is the right fit.
When to use Palette
- Professional colorists, whom the site claims as users "at many of the largest production companies"
- Production companies and TV channels handling archive footage and stills at scale
- Individuals restoring family photographs, for what the site calls "mass colorization of your family memories"
- Archivists and picture librarians with large holdings, since a script can push over one million images per day
- Developers embedding colorization in their own product, either building an app from scratch or adding Palette to an existing one
When not to use Palette
- Anyone looking to generate an image from nothing: Palette recolors a photograph you already have, it never creates one
- Anyone who needs enlargement or higher resolution: there is no upscaling, the free plan actually shrinks output to 500x500 px, and hard limits are 10 MB per image, 5000x5000 px in the app and 4K through the API
- Video post-production teams: the video API is still in beta, there is no editor, and the engine "isn't specifically designed for video frame consistency"
- Organizations under a compliance obligation: no data processing agreement is published, no Article 27 EU representative is named, no SOC 2 or ISO certification is mentioned, and data is processed in the United States
- Under-18s, whom the terms bar outright since the service is "intended only for access and use by individuals at least eighteen (18) years old", and anyone needing a non-English interface, as the site and documentation exist in English only
How to use Palette
A typical end-to-end flow, from setup to results.
- Open the colorization tool on the filters page, no registration required to try it
- Click Upload image, drag and drop one or several files, or click one of the sample photographs
- Browse the 21+ filters and pick one
- Click the filter adjustment icon to refine it, the site advising "clear descriptions and key words for optimal colorization"
- Hit Colorize again after changing the keywords to regenerate the image
- Download the result, watermarked and capped at 500x500 px on the free plan
- Create an account to claim one free trial credit and a logo-free, full-resolution render
- Buy credits as a one-off purchase, or subscribe for volume, watermark-free previews and the pro features (pinned, organized and duplicated filters)
- Developers: register for the photo and video API on RapidAPI from the documentation site
- Call the endpoints (/colorize_image_with_auto_prompt, /colorize_image_with_reference_image, /generate_image_prompt and others) or use the supplied Google Colab notebooks for bulk processing
Pros & Cons
Pros
- A genuine trial with no account and no card, since the tool opens straight onto the filters page
- A permanent free plan with unlimited previews rather than a trial that expires after a few days
- One-off purchase available at 49 USD for 75 credits valid two years, so no subscription is forced on occasional users
- Fine control over the render through filters, keywords and reference images, where many colorizers offer a single button
- Fast processing, claimed at 100-300 ms per image
- Documented bulk pipeline up to a million images a day through Google Colab notebooks, Google Drive and Dropbox, plus a public API with both photo and video endpoints
- Output ownership left with the user and commercial use explicitly allowed without restriction, with subscription credits rolling over up to five times the monthly allowance and a 14-day money-back guarantee under conditions
Cons
- The free plan is heavily constrained: Palette logo in the bottom-right corner, images cut down to 500x500 px, advertising and promotions
- API documentation is showing its age, with pages last updated between one and three years ago and an empty "Other resources" page; the move to RapidAPI is flagged only as a note, with no up-to-date API pricing on the site itself
- Legal identity is inconsistent: the terms and privacy policy name Palette SAS under French law, while the footer names PALETTE FM, UNIPESSOAL LDA in Porto
- No data processing agreement, no Article 27 EU representative and no security certification such as SOC 2 or ISO is mentioned, while data is transferred to and processed in the United States whatever the user's country
- No About, Contact or pricing page with its own URL: everything lives in unanchored homepage sections, and a single address, support@palette.fm, serves support, legal, GDPR and DMCA matters
- The video API is announced as beta and temporal consistency is not designed for video, so frame-to-frame stability has to be improvised
- Interface and documentation are English only, with no list of supported languages, and access can be blocked if more than 15% of your requests fail
Pricing & Plans
Palette runs a freemium model. A permanent free plan is available, offering unlimited colorization previews with all filters and customization, but returning watermarked images resized to a maximum of 500x500 pixels and served alongside advertising; registration adds one free trial credit good for a single logo-free, full-resolution colorization. The lowest paid entry point actually displayed is a one-off purchase of 49.00 USD for 75 credits, that is 0.65 USD per credit, valid for two years. Smaller one-off tiers exist for 1, 10, 200 and 500 photos, but their prices are rendered client-side and could not be recorded. The annual subscription is displayed at 72.00 USD per year for 480 credits, or 0.15 USD per credit. On the developer side, a free Basic API tier provides 200 credits, followed by Pro at 39 USD, Ultra at 79 USD and Mega at 169 USD. All prices are shown in USD.
- Free — 0 USD — unlimited colorization previews
- 21+ filters and customization
- Palette watermark
- output capped at 500x500 px
- advertising
- plus one free trial credit on registration
- Subscribe — 72 USD per year — 480 credits a year at 0.15 USD per credit
- with tiers of 480
- 2
- 400 and 6
- 000 photos
- up to 10x faster processing
- no watermark
- pro features and workflows
- Pay once — 49 USD — 75 credits at 0.65 USD per credit
- with tiers of 1
- 10
- 75
- 200 and 500 photos
- valid for two years
- single payment
- limited advertising
- API Basic — free — 200 image credits
- 0.01 USD per extra credit
- 1 request per second
- API Pro — 39 USD — 8
- 000 credits
- 0.005 USD per extra credit
- 2.5 requests per second
- API Ultra — 79 USD — 20
- 000 credits
- 0.004 USD per extra credit
- 3.3 requests per second
- API Mega — 169 USD — 48
- 000 credits
- 0.0035 USD per extra credit
- 5 requests per second
- API credit scale — 1 credit buys one watermarked SD image or one prompt call
- 10 credits one SD image
- 15 credits Full-HD and 20 credits 4K
Data, GDPR & hosting
A consolidated view of how Palette handles your data.
GDPR overview
GDPR handling is documented rather than certified. The privacy policy, effective 24 September 2024, carries a dedicated section titled "Your Data Protection Rights Under General Data Protection Regulation (GDPR)" linking to EU Regulation 2016/679. Palette declares itself the data controller. The rights listed are access, rectification, erasure, objection, restriction, portability and withdrawal of consent, plus the right to complain to a data protection authority in the EEA; requests go to support@palette.fm. CalOPPA and CCPA sections follow, and Do Not Track signals are honored. The policy also states that data is transferred to the United States for processing. What is absent matters: no Article 27 EU representative, no data protection officer, no published DPA, and the phrase "GDPR compliant" appears nowhere. Compliance is inferred from the rights section, never claimed outright.
Who owns the data?
The terms are unusually direct: you keep ownership of every image or video you produce with the service, and Palette states that it and its licensors claim no ownership rights in content generated through the API or the interface. Content belongs to the user who submitted it for processing. In return, posting content grants Palette a license to use, modify, display, reproduce and distribute it on and through the service; that license extends to making your content visible to other users only where you have given explicit permission. Palette has the right, but not the obligation, to monitor and edit posted content. The service itself, outside your content, remains the exclusive property of Palette SAS and its licensors.
Reuse rights
Reuse needs no permission request. The terms state that you are free to use these images or videos for any purpose, including commercial purposes, without any restrictions, and no rights are assigned back to the publisher in exchange for that commercial use. The limits are commercial rather than legal: one credit buys one colorized photo with no logo at full resolution, up to 5000x5000 pixels, while the free plan returns a file carrying the Palette logo in the bottom-right corner and resized to a maximum of 500x500 pixels. You also warrant that you hold the rights to every image you upload.
Data retention & training
Hosting summary
No hosting country and no data centre is named anywhere in the published pages. The only jurisdiction statement concerns processing: If you are located outside United States and choose to provide information to us, please note that we transfer the data, including Personal Data, to United States and process it there. The site itself is served behind Cloudflare, on an anycast address geolocated in the United States, which describes delivery infrastructure rather than storage. The privacy policy does name its service providers: Google Analytics, Firebase, Cloudflare Analytics, Mixpanel, GitHub, Google AdSense, Google Ads, Twitter and Facebook, plus the payment providers PayPal and Braintree, Stripe, Apple and Google Play in-app purchases, Alipay, Klarna, Square, WePay and WeChat. The terms add error-tracking tools, namely Bugsnag, Sentry and Firebase Crashlytics. What is missing is just as telling: no data processing agreement, no standard contractual clauses and no Article 27 EU representative. The terms are governed by French law while processing happens in the United States.
Things to keep in mind
Risks and trade-offs to weigh before adopting Palette.
- Colorization is interpretation, not restoration: the colors produced are plausible, never historically attested, which is a real problem on archive, press or family photographs that are then read as documents
- Memory displacement is a known side effect: a recolored family photograph tends to replace the original in the mind of whoever looks at it
- Journalistic, editorial or museum use without an explicit mention of the AI treatment turns a pleasant image into visual misinformation
- Uploaded files often contain the faces of third parties who never consented, while the terms put the burden of holding the rights entirely on you
- Data is transferred to and processed in the United States whatever your country of residence, and the legal entity behind that processing is uncertain, since the terms name Palette SAS under French law while the footer names PALETTE FM, UNIPESSOAL LDA in Porto
- A single generic address, support@palette.fm, handles support, legal matters, GDPR requests and DMCA claims, which leaves poor traceability for a sensitive request, and no DPA or Article 27 representative covers professional use on personal data
- Advertising and promotions are built into the free plan, with advertising cookies from Google AdSense and DoubleClick, Google Ads remarketing, Twitter and Facebook; credits perish (monthly ones at the end of the cycle, one-off ones after 24 months) and access can be blocked if more than 15% of your requests fail
Setup & Integrations
Technical difficulty
Very low for everyday use: nothing to install and no account needed to try it. Open the page, drop in an image, pick a filter. An account is only required to obtain a full-resolution, watermark-free result. Developers face a moderate step, with registration on RapidAPI, an API key, HTTP POST calls and Python snippets provided. Bulk colorization runs from ready-made Google Colab notebooks connected to Google Drive or Dropbox, which is advanced-user territory rather than engineering. No client software is needed; a Windows desktop application exists but is published by a third party, Covenant.
Deployment
Integrations
Behind Palette
Social
Resources
All the official URLs gathered for verification and reference.
Frequently asked questions
Do I need an account to try Palette?
Does the free plan watermark my images?
What exactly is a credit?
Do credits expire?
Can I use the colorized images commercially?
Are my photos used to train the AI?
Is there an API?
Can I colorize a large batch at once?
What formats and file sizes are accepted?
Is there a refund policy?
Should you pick Palette?
Palette is a specialist rather than a suite: it does one job, colorizing black-and-white images, and it goes deep on it. That focus shows in the controls. Where most colorizers offer a single button, Palette layers 21+ named filters, keyword prompts and reference-image guidance over a fast engine, then backs the workflow with a documented photo API, batch notebooks and a beta video API.
Credibility here is claimed rather than audited, but the claims are specific: 54 million photos colorized, adoption by professional colorists at large production companies, and named endorsements from Kevin Kelly of Wired, PiXimperfect and Bycloud. Pricing is unusually reader-friendly for the category, since a one-off purchase of 49 USD for 75 credits valid two years lets you use the tool seriously without ever subscribing.
The reservations are real. The free plan is generous in volume and tight in output: every preview is watermarked, capped at 500x500 pixels and served with advertising, so any serious use converts into credits. Legally, two identities coexist, Palette SAS under French law in the terms and privacy policy and PALETTE FM, UNIPESSOAL LDA in Porto in the footer; data is transferred to the United States for processing; and there is no published DPA, no Article 27 EU representative and no security certification. Technically, the API documentation shows its age, with pages last updated between one and three years ago, a migration to RapidAPI mentioned only in a note, and a video API still in beta.
Two audiences fit naturally. Individuals with a box of family photographs will get a convincing result in minutes and can stop at a handful of credits. Image professionals and developers get real depth, batch pipelines, prompts and an API, provided their compliance requirements are light.
- Choosing a selection results in a full page refresh.
- Opens in a new window.