Product updates

Refini changelog

A public record of meaningful Refini improvements, written for users who want to understand what changed across image tools, billing, credits, downloads, safety, and account management.

Latest update

Aug 18, 2026

Covered areas

19 release groups

Focus

Trust, billing, tools

· v2.6.7

Homepage and SEO

A clearer AI photo workflow across the site

We refreshed the homepage and core tool pages to make portrait editing easier to understand, improve discovery, and present more trustworthy product information.

New trust strip and tool hub

The homepage now surfaces practical privacy and format information, a fuller AI photo toolkit, and clearer links between related tools.

Portrait-first messaging

Enhancement and upscaling pages now explain the difference between natural facial detail and generic, waxy sharpening with more focused examples and copy.

More accurate search metadata

Structured data, FAQ text, language alternates, and page headings were aligned so search engines receive cleaner, more complete product information.

Anonymous trial setting clarified

When anonymous trials are disabled, uploaders now guide visitors to sign in instead of incorrectly suggesting that a trial quota has been exhausted.

· v2.6.6

Multilingual experience

Japanese, German, and French experiences are now available

Refini now offers broader localized experiences in Japanese, German, and French across the main tools, account areas, policies, and product navigation.

More of the product is localized

Core tools, landing pages, pricing, account screens, legal pages, and navigation now include Japanese, German, and French content where supported.

Localized copy was refined

Translations and page copy were reviewed across the newly supported languages to make key actions, tool descriptions, and guidance clearer.

· v2.6.5

Prompt safety

Safer prompt checks with fewer false positives

Prompt safety checks now handle explicit harmful requests more consistently while allowing ordinary editing, educational, medical, and redaction workflows.

Clearer safety classification

The safety model now uses a focused SAFE or UNSAFE decision for explicit harmful image-editing requests and defaults ambiguous requests to SAFE.

Context-aware multilingual checks

Local checks cover more languages and distinguish harmful person-focused edits from benign requests such as removing clothing from objects, blurring sensitive content, or creating medical and news imagery.

· v2.6.4

Image upload reliability

A more consistent image processing pipeline

Uploaded images now pass through one normalization and validation flow before moderation and processing, making results more consistent across supported formats.

Supported formats use one processing standard

JPEG, PNG, WebP, HEIC, HEIF, and AVIF uploads are validated and normalized to a consistent color space and output format before downstream processing.

Orientation and transparency are handled consistently

Image orientation is corrected automatically, while transparent areas are flattened onto a white background to avoid unexpected results in tools that require opaque images.

Invalid files are rejected earlier

File contents are checked against the declared format, and incomplete or damaged images are rejected before moderation or AI processing begins.

· v2.6.3

Account privacy

Permanent account deletion is now available

Users can now permanently delete their Refini account from account settings with email verification and clear information about what deletion includes.

Deletion is protected by email verification

A six-digit code sent to the account email must be confirmed before a permanent deletion request is accepted.

Active subscriptions are cancelled

The deletion workflow cancels supported active subscriptions and immediately blocks further access to the account.

Account content is cleaned up

Images, API keys, credits, and related account data are removed, while limited records required for legal, tax, fraud-prevention, or dispute purposes may be retained as described in our policies.

· v2.6.2

Trust and account management

Clearer policies and smoother subscription controls

We updated customer-facing policy pages and improved the subscription management flow so paid users can review status and manage recurring access with less friction.

Policy pages are easier to review

Acceptable use, abuse reporting, and DMCA pages were refined to make service rules, reporting paths, and AI image safety expectations easier to understand.

Subscription status is more visible

Account and order screens now surface subscription state more clearly, including management actions where supported by the payment provider.

Payment compliance updates

Checkout and billing flows received compliance-focused updates for PayPal and Waffo, including clearer terms presentation around AI image processing purchases.

· v2.6.1

Safety and product focus

Image moderation rollout and a cleaner tool catalog

We connected image moderation to upload and generation flows, added moderation logging for operations review, and removed the standalone AI image generator from public navigation.

Safer image uploads

Uploaded and generated images now pass through a moderation layer that helps block unsupported or unsafe content before processing continues.

Better moderation feedback

User-facing error messages were improved so rejected uploads explain the issue without exposing internal moderation details.

Operational audit trail

Moderation events are logged for admin review, helping the team investigate abuse reports and improve safety handling over time.

Focused public navigation

The standalone AI image generator entry was removed from the public tool catalog and sitemap while editing, enhancement, restoration, and background tools remain available.

· v2.5.7

Library and unlocks

More reliable single-image unlocks

We tightened the handling around single-image unlocks, image retention, and watermark behavior so one-off purchases and saved results behave more predictably.

Single-image unlock flow improved

Checkout and order handling were adjusted so unlocked images map more reliably to the correct user and processing result.

Retention rules clarified in product logic

Image history handling now better reflects retention expectations for unlocked assets and user plan access.

Watermark edge cases fixed

Background removal and unlocked result paths received fixes for cases where watermark behavior did not match the user's access level.

· v2.5.6

Downloads and credits

Download reliability fixes for paid results

We improved image download logic, fixed paid-user download failures, and cleaned up several credit and image viewer edge cases.

Paid downloads fixed

A download issue affecting paid users was corrected so eligible users can retrieve processed images as expected.

Download panel refinements

Image result and info panels were adjusted to make download actions more dependable and easier to understand.

Viewer and credit bugs resolved

Several image viewing and credit-state issues were addressed to reduce confusing account states after processing.

· v2.5.5

Credits and refunds

Credit accounting improvements

We improved subscription credit handling, refund-related credit deduction, and free-user watermark behavior.

Subscription credits fixed

Subscription credit issues were corrected so user balances better match active plan entitlements.

Refund deductions supported

Admin and order logic now supports deducting credits when a refunded order requires balance reconciliation.

Free-user watermark handling improved

Watermark logic was updated for free users so preview and download behavior aligns with access rules.

· v2.5.4

Pricing experience

Regional pricing and clearer upgrade prompts

We added tiered pricing logic, improved insufficient-credit prompts, and adjusted introductory credit bonuses for the lowest package.

Tiered pricing support

Pricing can now load by region, helping the product show the right plan data earlier in the purchase flow.

Clearer insufficient-credit states

Credit shortage dialogs were refined so users see upgrade options at the right moment with fewer confusing interruptions.

Introductory bonus adjusted

The lowest package bonus was updated to an extra 10% credits as part of ongoing pricing calibration.

· v2.5.3

Editing workflow

Image extender experience improvements

We refined the image extender page and related interaction logic so expanding photos feels clearer and more predictable.

Cleaner extension flow

The image extension page received logic and experience updates around choosing output shape and continuing edits.

More helpful expected-error messaging

Known processing failures now present clearer guidance instead of generic failure states.

· v2.5.2

Access and anti-abuse

Payment eligibility and account safety refinements

We improved anti-abuse handling, adjusted free-credit eligibility for unsupported payment regions, and refined admin usability.

Free-credit rules adjusted by payment availability

Countries where payment is unavailable no longer receive free credits that would lead users into an unsupported purchase path.

Admin and error handling polish

User list layout and expected-error messages were improved to make support and operations work easier.

· v2.5.1

Promotions and resource pages

New-user bonus and resource-page SEO updates

We added a first-purchase bonus mode for new users and improved SEO plus imagery on resource pages.

First-purchase bonus support

A new-user first-purchase flow can now offer up to 30% extra credits as part of onboarding promotions.

Resource pages refreshed

Three resource pages received SEO and image updates to better explain use cases and improve search presentation.

· v2.4.5

Performance and pricing

Faster brand loading and clearer pricing credit displays

We improved logo loading, performance details, subscription pricing, and how credits appear in pricing cards and shortage prompts.

Logo and performance cleanup

Brand assets now load from a simpler path instead of translation content, reducing avoidable page work.

Pricing cards made clearer

Credit display logic in pricing cards and insufficient-credit prompts was adjusted so package value is easier to compare.

· v2.4.4

Analytics, privacy, and downloads

Analytics configuration and anonymous-library safeguards

We added custom-domain support for Plausible page views, improved free-download scaling, and tightened anonymous browsing behavior.

Analytics domain support

Plausible page-view tracking can now use a custom domain configuration.

Library access clarified

Anonymous users are guided to sign in before viewing the library, and watermark handling was updated to use a two-image mode.

· v2.4.3

Operations and trial safety

Usage records, verification, and anonymous trial fixes

We improved usage-record visibility, added more anonymous-user context for admins, and fixed verification issues in the anonymous flow.

Better usage records

Admin usage records now include more detail, including anonymous-user information where relevant.

Anonymous trial protection

Human verification and anti-abuse logic were refined to reduce trial misuse and fix verification loading problems.

· v2.4.2

New tools and anonymous access

White-background portrait tool and anonymous trial access

We launched a white-background portrait workflow and introduced anonymous access with revised watermarking.

White-background portrait tool launched

A new tool combines portrait enhancement and background removal for clean white-background results.

Anonymous access added

Users can try selected image workflows before signing in, with watermark behavior adjusted for trial output.

· v2.4.1

Tool catalog foundation

New tool pages, AI Tools hub, and SSG rollout

We added multiple tool pages, improved multilingual page data, introduced an AI Tools hub, and moved landing pages toward static generation for better load performance.

More public tool pages

Six tool pages were added with shared configuration and multilingual content structure.

AI Tools hub and static pages

A central AI Tools page improved internal linking, while SSG updates helped public landing pages load faster.