NEW ENGINES * GPT Image 2 for when you want to edit an existing image without repainting everything, or when you need a complex image with readable text in it, like a sign, magazine, or poster. * Seedance 2 Fast is a new video engine you can use with just text, with a starting frame, or with a handful of reference images you weave into your prompt. * LTX Fast make video clips from 6 to 20 seconds long, and you can pin both the opening and closing frames so the motion lands where you want it. LTX Fast runs at about two-thirds the price, so it's a nice way to mess around without spending much. DIFFIE CLOUD Introducing Diffie Cloud: a workshop for your creative materials, alongside the Gallery you already use. Your current Gallery is more like a portfolio, showing the finished work you've generated. Diffie Cloud is where the materials you build with live: the source photos you upload as references, and the half-finished edits you've saved for later. That means when you're starting something new, the photo you used last week is one tap away, without re-uploading every single time. VIDEO EDITING When you generate a video clip, you can now trim it to exactly the moment you want before doing anything else with it. Pick your start and end points, then send the result wherever you'd send a finished piece: save it to your photo gallery, save it to Diffie Cloud to come back to later, publish it to your feed, or share with a friend. Scrubbing also tracks accurately at any speed now, so finding the exact frame to cut on doesn't feel like guesswork anymore. SOCIAL When a fan likes a bunch of stuff in a row, the Social feed bundles those into one item instead of burying everyone else. MISC Tons and tons (600 commits for you nerds) of bug fixes, UI and UX improvements, consolidations, alignments, tweaks, and nudges.

Socials, rebuilt. Notifications now bundle by day, expand on tap, and surface fresh activity in a "New" section up top. No more scrolling through the same three likes ten times. Artists get notified (and credited with Diffies) when their work lands in the public feed. A header that earns its space. Your Diffie balance lives right up there. No more losing your wallet. Sign up. Start making. New accounts get an anonymous handle and pfp on the spot, so you can create before you've decided who you are. Claim a real username when you're ready, or don't - you can be anon if you really want. Cursor, restored. The text overlay tool finally has a blinking cursor and better controls. Editing captions feels like editing text, not wrestling with a label maker. Studio polish. More intuitive UI for choosing a sampler. It's a pretty technical factor that can really improve your image quality. Quieter wins. Smoother animations. More consistent colors, typography, alignment. Smarter and faster feed filtering that respects your content prefs.

Bug fix: Prompt fields not accepting input after first focus.

A smoother sign-up - SignUp is now the first screen you see, with a friendlier heads-up about what to expect - Quicker verification flow for Google and Apple sign-ins - A new "Verify Email" shortcut in Settings - Celebration when your verification bonus lands Meet your new Inbox - Alerts and Activity now live together in one Social tab - See likes, follows, remixes, and refines grouped by person and time - Tap any notification to jump straight to what it's about A refreshed Profile - Redesigned profile editor with an easier way to edit your avatar - A checklist to guide you through completing your profile - A celebration overlay when you earn your profile bonus - Cleaner profile card with bonus info front and center More Diffies, more clarity - Starter Diffies are more secure and reliable for new accounts - A gentle nudge when your balance is running low - Better tracking so bonuses and grants land correctly Under the hood improvements - WAN is now the default video engine for higher-quality results - Faster, more reliable notifications and background updates - iPhone install support for sharing with friends and testers - Lots of polish: smoother empty states, fewer image flashes, better keyboard behavior, and more

New Features - Before/After Compare — Slide to compare your original and generated images, works while zoomed in - Multi-Select Sharing — Share multiple images at once, works correctly on both platforms - Better Error Messages — Failure notifications now tell you the specific reason instead of a generic error Bug Fixes - Video jobs no longer hang — Proper timeouts and automatic cleanup for stuck video generation jobs - Correct engine labels — zImage Turbo, Stable Diffusion, and others now display their proper names everywhere - Refine/Upscale shows original workflow — Viewing a refined or upscaled image now displays the source workflow's details - No more annoying gray flash on launch — Smooth transition from splash screen to feed - Refining images now respects source image dimensions better. - Autosave works reliably — Images now autosave when completing in the foreground, not just from background notifications

Minor bug fix

New Features - Image Upscaling — Upscale your images to 4x resolution directly from the image viewer - Refine Mode — Tweak and refine existing images with a simplified editing interface, plus a before/after comparison slider to see what changed - Video Generation — Added Wan v2.5 video engine alongside v2.2, with improved controls and native video playback - Cost Approval — See the estimated Diffie cost before generating, with the option to approve or cancel - Before & After Comparison — Drag a slider to compare refined images against the original; tap to toggle, hold to flash - Engine Selection — Choose which generation engines appear in the Studio from Settings - Share to Public Feed — Share your creations to the public feed for others to discover - New token (Diffie) based economy. Generous initial grant plus monthly subscriptions now available Improvements - Landscape Mode — Image viewer now supports landscape orientation - iPad Polish — Better spacing, popover menus, and responsive scaling across the app on tablets - Task Manager Upgrades — Canceled jobs stay visible, retry failed jobs, and see the cost of each generation - Haptic Feedback — Tactile feedback on interactive elements (can be toggled off in Settings)

Release Notes v6.2.0 Studio Improvements - Slider controls for higher resolution settings values - Clearer settings descriptions - Refreshed engine selector - Better engine descriptions Gallery Improvements - Smoother image loading Bug Fixes - Fixed issue where workflows weren't displaying correctly in workflow details - Improved handling of older saved images in the gallery - Extensive code refactoring to improve reliability and performance

iPad Improvements - New sidebar navigation - Gallery, Images, and Videos tabs are now accessible from the iPad sidebar for easier navigation - Active Jobs button  - Added to the iPad sidebar so you can quickly check on your generation progress; shows a spinner when jobs are running - Better text and button sizing  - Text and buttons now scale properly on iPad for improved readability - Improved touch responsiveness - Fixed an issue where touches weren't registering properly on iPad Gallery & Filtering - Filter button added to Gallery - Gallery images now load smaller preview thumbnails, reducing memory usage and improving scroll performance Generation & Notifications - "Don't show again" option added to informational dialogs - Better notification handling - Improved UX when you tap on generation completion notifications Bug Fixes - NSFW Settings and other scrollable panels now scroll all the way to the bottom - Fixed workflow details panel - Stability improvements - Fixed crashes that could occur when quickly scrolling through images

Generate AI videos from text or images with the new Kling engine User diffusions now have their very own screen Create images with Gemini's semantic composition from multiple source photos Videos now appear in your feed and gallery with inline playback Crop source images to perfectly match your output aspect ratio Full iPad and tablet support with optimized layouts Blur or reveal mature content with customizable NSFW presets Search the public feed by keywords Remix any video to tweak and regenerate with new settings See active generation progress and completion badges in the tab bar Push notifications now jump directly to your finished diffusion Paste images from clipboard as source material Delete images and videos directly from the viewer Control whether videos autoplay in your feed Fixed Autosave when app in background

What's New Content Safety Controls - New NSFW content filtering — Choose how mature content is displayed with customizable blur settings. - Tap the eye icon to reveal blurred images when you want to see them. - Content filtering now works consistently across your feed, image viewer, and saved images. Improved Model & LoRA Selection - Search and filter your models and LoRAs to quickly find what you're looking for. - Tap any model or LoRA to see detailed information including descriptions and sample images. - Checkpoint and LORA search! It's been a long time coming and it's here! Feed Improvements - New update indicator — A badge now shows when new images are available in your feed. - Faster image loading with smarter caching. - Long model names now automatically resize to fit on screen. iPad Enhancements - Improved navigation and layout for iPad users. - Better responsive sizing throughout the app. General Improvements - Redesigned sign-in and sign-up screens. - New unified notification system for a cleaner, more consistent experience. - Double-tap to zoom works again in the image viewer. - Various layout and spacing fixes for a more polished look.

Native iPad support Better support for LORAs Improved user experience and fixed many bugs

COMPLETELY re-written app with improved stability and reliability. This new version will quickly achieve parity with the iOS-native version and allow for a massive increase in functionality.

This version incorporates subscription convergence with Diffie Studio and enhancements for professional token replenishment logic.

‣ Pro tokens are now visible in admin view ‣ Bug fixes to replenishment logic

FLUX has been upgrade to 1.1 - 6x faster image generation compared to Flux 1 Pro - Improved image quality with better prompt adherence and diversity - Support for ultra high-resolution images up to 2K - Enhanced realism, though potential issues with faces in multi-person scenes - Initially available on Freepik's AI image service, with API access for other providers - Outperforms previous models in user ranking tests and Artificial Analysis image arena - Fixed Seed value not retaining in Advanced Settings - Fixed Step slider jump issue in Advanced Settings

The new Pro-Level Flux feature is here! ‣ Async Mode always enabled ‣ Feed Filter button clear enhancement ‣ App Renamed to DiffieArt to align with other Diffusitron Apps such as DiffieAI and DiffieStudio ‣ New awesome launch screen image Enjoy enhancing your creations!

Our new Pro-Level Image Enhance feature supercharges your images using the original diffusion model. Here's how it works: • Select any image you've created • Hit "Image Enhance" • Watch your image transform with improved details and clarity Other Improvements: • Faster image generation • Bug fixes • UI improvements

you can receive a notification when new models and LoRAs are published. Ding! Images now load and scroll faster In addition to blur, you can now completely hide all nsfw content.

By request, users now have the choice to use the new multi-Diffusion mode or go back to single-Diffusion mode Better feed filtering and faster image rendering across the entire app Bug fixes and many performance improvements Added yearly subscription option

Better process control of in-flight diffusions Fixed up some UX and UI elements Bug fixes and performance improvements

You can now run multiple diffusions at once - no more staring at spinning circles! ControlNet! Now you can diffuse and maintain poses, edges, etc. Better filtering to search the feed and find what you're looking for. Fix up janky faces, weird eyes, etc, with our face fixer. Upscale images to bigger and better dimensions Smoother UX and better performance.

New realtime feed showcasing all the best diffusions from our users. More models and more LoRAs!

Base- and LoRA-model pickers now more compact. When repeatedly diffusing, will now default to using the current diffused image as the next initial image. This pattern is now also reflected accurately on screen - previously would continue to show wrong initial image. Prompt boxes default to using height set by user. The previous "auto-hug" functionality can be enabled in Settings. Many small user experience (UX) and performance improvements.

New models! The introduction of LoRAs! (they're cool) Inpainting-specific models (for better results) Resizable prompt boxes Profile picture will now stick around across reinstalls Long-press on Info to copy said info to your device's clipboard More robust networking. Won't swoon and give up as easily now Many small UI/UX enhancements to make things just a little better here and there A few too many bug fixes