🎉 ArtTic-LAB 4.1.0

📝 What's Changed

• Refactor XPU validation, update install scripts, add tests (f9dca84)

---

View the previous release for detailed info
✨ Full Changelog: 4.0.0...4.1.0

🎉 ArtTic-LAB 4.0.0

📝 What's Changed

• fixed the README.md and write a github workflow using AI for publishing release (2cec924)
• fixed the version mismatch across files (0e890d6)
• Updated the script to support all major linux distro (020c801)
• New Logo (e63ad36)
• Add Share button to the main UI template (6063e49)
• Implement frontend logic for Ngrok sharing and SDNQ toggling in the node graph (325bbdd)
• Update CSS for new UI elements (SDNQ toggle, Share button) and responsive layouts (89d14c3)
• Update WebSocket server to handle sharing actions, quantization args, and error handling (a0ed2d1)
• Improve Linux startup script logging, styles, and environment variable handling (efd6ef0)
• Refine Windows startup script loop handling and environment checks (51d2192)
• Update dependencies to include SDNQ, Triton, and system monitoring utilities (e2bfebe)
• Add optional SDNQ quantization support to SDXL pipeline (7acc9fe)
• Clean up SD3 pipeline loading logic and remove legacy text encoder wrappers (290408d)
• Add optional SDNQ quantization support to SD2 pipeline (37053a4)
• Add optional SDNQ quantization support to SD1.5 pipeline (05fcd73)
• Update FLUX pipeline with optional SDNQ quantization and remove legacy XPU hacks (38b6a06)
• Standardize base pipeline device placement and remove legacy CPU-forcing hacks (39e2530)
• Refactor model architecture detection logic to better distinguish FLUX/SD3/SDXL (aa24f76)
• Rewrite Linux installer with auto-hardware detection and distro-agnostic setup (d8021b7)
• Rewrite Windows installer to support hardware selection (CUDA/XPU/CPU) and new deps (e09ec29)
• Enhance system info logging with detailed hardware detection for Intel/NVIDIA/CPU (4e9c251)
• Remove legacy XPU patch; native PyTorch XPU support is now sufficient (2dc1a0c)
• Implement quantization (SDNQ) loading state, Ngrok toggling, and generic VRAM calculation (8adf6a9)
• Add Ngrok sharing logic, warning suppression, and clean up application entry point (d645ee6)
• Update docs to reflect Universal Hardware support (Intel/NVIDIA/AMD) and SDNQ features (9324282)
• Refactor ARC utility to focus strictly on driver verification and native runtime checks (58b4095)
• Trying out new things (471b576)
• Create .prettierrc (747219c)
• using prettier formatting. no code change (c0c8781)
• Add Intel ARC GPU Linux performance optimizer script (f19138a)
• Refactor file handling and VAE tiling logic (8d04ebf)
• Delete error.png (1b96f13)
• remove useless files (10c9965)
• Merge pull request #4 from rootLocalGhost/linux-prep-optimize-12246209573885703571 (530fcf3)
• Prepare for Linux run, add device fallback, and mark img2img as unstable (ee3475b)
• Prepare for Linux run, remove img2img, and optimize for cross-platform (55d87f4)
• Refactor: Update username from Md-Siam-Mia-Man to rootLocalGhost (f428c66)
• Merge pull request #2 from Md-Siam-Mia-Man/perf-improve-arc-ipex-optimization-1459662916832583792 (6dd9a01)
• Optimize pipeline components with IPEX and memory format (2bf43e8)
• v3.2.0 (accf340)
• remove the img2img pipeline completely (f5a3ff1)
• Add input image UI element to index template (b38be6d)
• Add img2img functionality to UI (619dccb)
• Add styling for img2img UI elements (13cd1a6)
• Add websocket support for init image and strength parameters (d9ca653)
• Add img2img support to SDXL pipeline (7b98fd3)
• Add img2img support to SD 3 pipeline (08f4da7)
• Add img2img support to SD 2 pipeline (71dd803)
• Add img2img support to SD 1.5 pipeline (8693ab3)
• Add img2img support to FLUX pipeline (ea02df8)
• Add pipeline mode switching functionality (e9aeed2)
• Add image processing and img2img support to core logic (8c6e2f9)
• v3.1.0 (61f6126)
• icon assets (6a42e55)
• Update server.py (4bac20d)
• Delete package.json (c0c1e5c)
• Update index.html (659bf60)
• Create install.ps1 (eec3c39)
• Update install.bat (5c750ea)
• Update .gitignore (dd9601d)

---

✨ Full Changelog: 3.0.0...4.0.0

ArtTic-LAB v3.0.0 Release Notes: The Phoenix Update

✨ The Star of the Show: The All-New Node-Based UI

We have officially retired the Gradio interface in favor of a fully custom, fluid, node-based user interface. This is the culmination of our vision for a truly modern and efficient creative workflow.

• An Infinite Canvas for Your Ideas: Break free from static forms. Your entire workflow, from model selection to final image, is now laid out on a spacious, pannable, and zoomable canvas.
• Complete Visual Overhaul: ArtTic-LAB now features a stunning new design language with a cohesive green-toned color palette for both Light and Dark modes. Every element has been redesigned with a rounded, floating, "glassy" aesthetic.
• Non-Blocking Notifications: Gone are the intrusive pop-ups. All progress indicators (model loading, image generation) and confirmations (VRAM cleared, etc.) now appear as elegant, temporary "toast" notifications, keeping you informed without interrupting your flow.
• Intuitive & Redesigned Controls:
  • Aspect Ratio Buttons: Now a sleek, icon-based control group that is both beautiful and functional.
  • Stateful Buttons: The "Load Model" button is now intelligent. It automatically transforms into an "Unload Model" button when a model is active, and this state persists even if you reload the page.
  • Application Feel: Text selection has been disabled across the UI for a more focused, application-like experience.

🚀 Major New Features & Quality of Life

• Share Your Canvas with the World: The most requested feature is here! You can now generate a public, shareable link for your ArtTic-LAB instance by launching with the --share argument. Built on ngrok, this allows anyone, anywhere, to access your UI.
• A Reborn Prompt Book: The Prompt Book has been completely reworked with a new card-based design that is both beautiful and functional. Adding, editing, deleting, and using prompts is now a seamless experience.
• Rock-Solid Backend Restart: The "Restart Backend" feature is now completely reliable. We've moved the logic to the launcher scripts, eliminating the "port in use" errors and ensuring a smooth restart every time.
• Polished CLI Experience: The Command-Line Interface has received a major visual and functional upgrade.
  • It now shares the same beautiful green color palette as the GUI.
  • Web server logs are now intercepted and reformatted into clean, concise, single-line messages (e.g., [Web] GET / -> 200 OK), making the console a pleasure to read.
  • When log filtering is active (the default), all verbose clutter from underlying libraries is completely silenced.

🛠️ Under the Hood: A Leaner, Meaner Engine

• Farewell, Gradio: By removing the Gradio dependency entirely, ArtTic-LAB is now leaner and more focused.
• Fully Offline Capable: We no longer rely on Google's CDN for fonts or icons. The installer now automatically includes all necessary assets, making the web UI fully functional without an internet connection.
• Dependency Management: The install scripts have been updated to automatically handle all new dependencies (pyngrok) and web assets (npm install), ensuring a smooth one-click setup.

⬆️ How to Upgrade

Due to the significant changes in dependencies, a fresh installation is highly recommended.

1. Download the latest v3.0.0 release.
2. Run the installer script for your OS: install.bat (Windows) or install.sh (Linux). This will create a clean environment with all the correct packages.
3. Launch the application using start.bat or start.sh.

---

Thank you for being part of the ArtTic-LAB journey. This update is for you—the artists, the creators, the tinkerers. We can't wait to see what you create.

Happy generating

Full Changelog: 2.6.0...3.0.0

ArtTic-LAB v2.5.0 - The Intelligence & Interface Update

✨ New Features & Major Enhancements

🧠 VRAM-Aware Resolution Guidance

• Problem: Users were encountering "Out of Memory" (OOM) errors when trying to generate images at resolutions too high for their GPU's available VRAM.
• Solution: ArtTic-LAB is now VRAM-aware! When a model is loaded, the backend analyzes your GPU's available memory and the model's architecture. It then calculates an estimated maximum safe resolution you can generate at.
• In the UI: This guidance is now displayed directly in the "Dimensions" card (e.g., "Est. max resolution: 1536px (VRAM), 2048px (Offload)"), empowering you to choose settings that will succeed and preventing frustrating OOM crashes.

🖼️ Unified & Modern Image Viewer (Lightbox)

• The old lightbox has been completely replaced with a powerful, unified viewer that works seamlessly from both the Generate and Gallery tabs.
• Full Keyboard Control:
  • ← / →: Navigate between previous/next images in the gallery.
  • Esc: Close the viewer.
  • + / -: Zoom in and out.
  • F: Fit the image to the screen, resetting zoom and pan.
  • Delete: Permanently delete the current image (with confirmation).
• Intuitive Pan & Zoom: Zoom into an image and then simply click and drag to pan around and inspect details.
• Integrated Actions: The viewer includes all essential actions: zoom, fit, delete, and close.

💬 Custom Dialog System

• All native browser pop-ups (alert, confirm) have been replaced with a custom, non-blocking dialog system. This provides a polished, professional, and consistent user experience for confirmations (like deleting an image) and error messages.

🗑️ Full Image Deletion Workflow

• You can now permanently delete images directly from the Image Viewer.
• A custom confirmation dialog ensures you don't delete images by accident.
• Deleting an image automatically updates the UI for all connected clients, removing it from the gallery and the generation preview if it was the current image.

🎨 UI & UX Improvements

• Non-Blocking Interface: The backend now runs all heavy tasks (model loading, image generation) in a separate thread. This means the UI remains 100% responsive at all times. You can browse the gallery, change settings, and navigate the app while a model is loading or an image is generating.
• Collapsible LoRA Section: The LoRA options are now housed in a collapsible section with a clean toggle switch, saving significant vertical space in the UI for a more compact and organized look.
• Dynamic Image Preview: The image preview area on the Generate tab no longer has a fixed aspect ratio. It now dynamically resizes to perfectly match the aspect ratio of the generated image, eliminating letterboxing and providing a true-to-form preview.
• Intelligent "Load Model" Button: The "Load Model" button is now state-aware. It automatically disables itself if the currently selected model, LoRA, and other settings are already loaded, preventing redundant work. It re-enables as soon as you change any relevant setting.
• Stateful Control Toggles: The "VAE Tiling" and "CPU Offload" checkboxes are now correctly wired into the state logic. Changing them will re-enable the "Load Model" button to apply the new configuration. Their states are also correctly reset when a model is unloaded.
• Auto-Resizing Prompts: The prompt and negative prompt text areas now automatically grow as you type, eliminating scrollbars for a much cleaner and more pleasant writing experience.
• New Defaults: The default generation parameters have been updated to 50 steps and a Guidance (CFG) of 3.0, based on modern generation practices. A new, more creative default prompt has also been added.
• Sequential Image Naming: Images are now saved with a simple, sequential naming scheme (ArtTic-LAB_1.png, ArtTic-LAB_2.png, etc.) for easier organization.
• Cleaner Status Messages: Refined UI messages for clarity, such as the "No model loaded" status.

⚙️ Backend & Stability Fixes

• Strict Log Filtering: The terminal output is now pristine. All non-essential warnings from underlying libraries (like safety_checker and autocast deprecation) and all progress bars have been completely suppressed, showing only clean, relevant ArtTic-LAB logs.
• Critical AttributeError Fixed: Squashed a major bug that caused the application to crash when loading non-FLUX models due to faulty logic in determining the model type. The model loading process is now robust for all architectures.
• Asynchronous Callback Correction: Fixed a critical bug where progress updates were not being sent to the UI from the backend due to an issue with the asynchronous event loop, ensuring smooth progress bar updates every time.
• Robust OOM Error Handling: In addition to the proactive guidance, if a generation still runs out of memory, the error is now gracefully caught, the VRAM cache is cleared, and a user-friendly error message is displayed in the UI instead of crashing the backend connection.

---

Thank you for using ArtTic-LAB! We're committed to making it the best-in-class tool for AI artistry on Intel hardware.

Full Changelog: 2.4.4...2.5.0

🎉ArtTic-LAB v2.0.0: The Experience Update🎉

🎉 Highlights of v2.0.0

• ✨ Brand New Custom UI: A bespoke, professional-grade interface is now the default experience! Built with a modern tech stack (FastAPI + WebSocket + Vanilla JS), it's faster, more responsive, and packed with custom components and animations that make creating a joy.
• 📉 CPU Offloading for Low VRAM: Run massive models like SDXL and SD3 even on GPUs with limited VRAM! Our new CPU Offloading mode keeps the bulk of the model in system RAM, making high-end AI art accessible to everyone.
• 🏛️ Dual-UI Architecture: We haven't forgotten our roots! The powerful Gradio interface is still available as an alternative. Launch with the --ui gradio flag to use the classic UI, now powered by a more robust, decoupled backend.
• 🎨 Complete UI/UX Polish: Every component has been re-imagined. Enjoy custom-styled sliders, dropdowns, and checkboxes, a sleek top navbar, a beautiful image lightbox viewer, and smooth animations across the entire application.
• ⚙️ Decoupled Core Logic: The application's "brain" has been separated from its "face." All core functionality now resides in a UI-agnostic core module, making the entire system more stable, maintainable, and ready for future expansion.

---

✅ What's New in v2.0.0: The Full Changelog

🎨 User Experience & Interface

• New Default UI: The Gradio interface has been replaced as the default by a custom-built frontend.
  • Technology: Powered by a FastAPI backend, WebSocket for real-time communication, and a lightweight, modern JavaScript frontend.
  • Custom Components: Features fully custom-built sliders, dropdown menus, checkboxes with helper text, and a gallery lightbox for an integrated and polished feel.
  • New Layout: The old sidebar has been replaced with a sleek top navbar and a professional three-column layout designed to be fully usable without scrolling on most displays.
  • Animations & Theming: The entire UI is animated with smooth transitions and adheres to a new, cohesive color theme.
• Gradio as an Alternative: The classic Gradio UI is still fully supported and can be launched using the --ui gradio command-line argument.
• Refined CLI Theme: The command-line interface has been updated with a new color theme and a gradient-style banner to match the new branding.

🚀 Features & Performance

• CPU Offloading (Low VRAM Mode): Implemented enable_model_cpu_offload() as a user-selectable option. This allows users to run large models that would otherwise exceed their VRAM capacity, at the cost of generation speed.
• Improved User Feedback: The new UI provides real-time progress bars for both model loading and image generation, sent directly from the backend over a WebSocket connection.

⚙️ Backend & Architecture

• Core Logic Decoupling: All application logic (model loading, generation, etc.) has been moved from app.py into a new, UI-agnostic core/logic.py module. This critical change allows different UIs (FastAPI, Gradio) to share the exact same stable backend.
• FastAPI & WebSocket Backend: The app.py launcher now starts a Uvicorn server for the new default UI. A new web/server.py file handles all API requests and manages the persistent WebSocket connection for a responsive experience.
• Robust Installation Scripts: Both install.bat and install.sh have been improved with more reliable Conda detection and clearer error handling to ensure a smoother setup process.

---

A Note on Our Journey

If v1.5.0 was about building a high-performance engine, v2.0.0 is about designing the perfect cockpit. A powerful tool is only as good as its ability to be wielded by the artist. This update was driven by the desire to eliminate friction, to make the interface disappear so that nothing stands between your idea and the final creation. We believe this new experience achieves that goal.

What's Next? 🔮

The new architecture opens up a world of possibilities. Here's what we're dreaming up for the future:

• 🧩 LoRA & LyCORIS Support: A full-featured system for loading and managing LoRA/LyCORIS files to customize your creations with incredible flexibility.
• 🖼️ Image-to-Image (Img2Img): The highly-requested feature to guide generations using an input image.
• 🛂 ControlNet Integration: The ultimate tool for fine-grained control over composition and poses.
• 🛠️ Model Management: Tools to download and manage your models directly from within the ArtTic-LAB interface.

---

Thank you for being on this journey with us. The canvas is ready.

Now, go create something extraordinary. ✨

Full Changelog: 1.5.0...2.0.0

ArtTic-LAB v1.5.0: Next-Gen Models, Seamless Performance

We are thrilled to announce the release of ArtTic-LAB v1.5.0: The Power & Efficiency Update!

This version represents a major leap forward, transforming ArtTic-LAB from a powerful tool into a truly professional-grade creative suite. We've gone under the hood to completely re-engineer the pipeline system, enabling support for the latest next-generation models while simultaneously introducing critical features for memory stability and workflow efficiency. This update is all about pushing the boundaries of what's possible on Intel ARC hardware.

🎉 Highlights of v1.5.0

• 🚀 Next-Gen SD3 Support: Be among the first to experiment with Stable Diffusion 3! ArtTic-LAB now intelligently handles single-file SD3 models by automatically fetching the required base components, giving you cutting-edge power with zero hassle.
• ✨ Full SD2.x & SDXL Compatibility: The model detection engine is now a master of its craft, flawlessly identifying and configuring pipelines for SD1.5, SD2.0, SD2.1, SDXL, and SD3 models.
• 🧠 Memory-Safe VAE Tiling: Say goodbye to "Out of Memory" errors on high-resolution generations! VAE Tiling is now integrated and enabled by default, ensuring smooth and stable performance even when pushing the limits.
• ⚡ Instant Model Unloading & Hot-Swapping: A new "Unload Model" button gives you direct control over your VRAM. Plus, the app is now smart enough to automatically unload the old model before loading a new one, making switching between models seamless and crash-free.
• ⚙️ Smarter, Faster Pipeline System: A complete architectural overhaul of the backend makes the entire application more robust, faster to load, and easier to extend for the future.

---

✅ What's New in v1.5.0: The Full Changelog

✨ Features

• Stable Diffusion 3 Support: The application can now load single-file SD3 models. It automatically downloads the official base components (like the text encoders and VAE) from Hugging Face on the first run and injects the local model's weights, giving you a seamless experience.
• Stable Diffusion 2.0/2.1 Support: Full support has been added for SD2.x models. The backend automatically detects them and uses the appropriate pipeline and 768x768 native resolution.
• Dynamic Aspect Ratio Presets: The one-click aspect ratio buttons are now even smarter! They check the loaded model type (SD1.5, SD2, SDXL, or SD3) and apply the correct resolution presets, removing all guesswork.

🚀 Performance & Stability

• VAE Slicing & Tiling: This critical feature is now integrated to dramatically reduce peak VRAM usage during the final image decoding stage. It is enabled by default to prevent out-of-memory errors when generating large images.
• Explicit Model Unloading: A new "Unload Model" button has been added to the UI, allowing you to instantly clear the active model from your GPU's VRAM.
• Automatic Model Hot-Swapping: The model loading process has been fortified. When you load a new model, ArtTic-LAB now automatically unloads the previously loaded model first, ensuring a stable transition without memory conflicts.

⚙️ Backend & Architecture

• Unified Pipeline Architecture: The pipeline system has been completely refactored around a new base_pipeline.py. This centralizes all core logic like IPEX optimization and device placement, making the code cleaner, more efficient, and follows the DRY (Don't Repeat Yourself) principle.
• Advanced Model "Sniffing": The model detection logic in pipelines/__init__.py has been significantly upgraded. It can now reliably differentiate between all major Stable Diffusion versions by inspecting their unique tensor key structures.
• Expanded IPEX Optimization: The IPEX optimization step is now more comprehensive. In addition to optimizing the main U-Net or Transformer, it now also optimizes the VAE, providing a broader performance uplift across the entire generation process.

---

A Note on Our Journey

The evolution to v1.5.0 was driven by a single question: "How can we make this not just work, but work flawlessly?" This update is a testament to the rigorous testing and forward-thinking vision of our lead tester, who pushed for the stability and next-gen features that define this release. We've moved from building a functional car to engineering a high-performance vehicle.

What's Next? 🔮

With a rock-solid and future-proof foundation, we're positioned to bring even more incredible features to ArtTic-LAB. Here's a look at what's on the horizon:

• Low VRAM Mode: Implementing "CPU Offloading" to allow users with less VRAM to run massive models like SDXL and SD3.
• LoRA Support: A full-featured system for loading and managing LoRA files to customize your creations with incredible flexibility.
• Image-to-Image (Img2Img): The highly requested feature to guide generations using an input image.

---

Thank you for your continued support. The creative ceiling has been raised.

Now, go make something amazing. ✨

Full Changelog: 1.0.0...1.5.0

🚀 ArtTic-LAB v1.0.0 - Official Release Notes

It is with immense excitement that we announce the official release of ArtTic-LAB v1.0.0!

Born from a need for a simple, powerful, and clean AI image generation tool on Intel ARC hardware, ArtTic-LAB has evolved from a single script into a full-featured creation studio. This release marks the culmination of a rapid development and testing cycle, focused on stability, user experience, and raw performance. Our mission was to build the tool we always wanted, and we believe we've achieved it.

🎉 Highlights of v1.0.0

• 🚀 Forged for Intel ARC: A native, optimized experience for Intel ARC GPUs using the latest PyTorch and IPEX extensions.
• ✨ Sleek & Modern Web UI: Ditch the command line! A beautiful Gradio interface provides all the controls you need in an intuitive layout.
• 🧠 Intelligent SDXL & SD1.5 Support: The app automatically detects the model type you load and uses the correct pipeline, every time.
• 🎛️ Full Creative Control: Fine-tune your creations with advanced options for Samplers, Image Resolution, Steps, Guidance, and more.
• 📦 One-Click Installation: New installer scripts for Windows and Linux make getting started easier than ever with a dedicated, isolated Conda environment.

---

✅ What's New in v1.0.0: The Full Changelog

✨ Features

• Advanced Generation Options: You can now control Width, Height, and select from a list of popular Samplers (Schedulers) like Euler A, DPM++ 2M, and more.
• Dynamic Model Resolution: The UI now intelligently suggests a default resolution (1024x1024 for SDXL, 512x512 for SD1.5) when you load a model.
• Image Gallery: A new "Gallery" tab has been added to browse all your past generations, sorted with the newest first.
• Refresh Models Button: Added a 🔄 Refresh Models button to the UI, allowing you to scan the models folder for new additions without restarting the application.

🎨 UI & UX (User Experience)

• Professional CLI: The command-line interface has been completely overhauled. It now features a beautiful startup banner, structured system information, and is completely silent by default, hiding all library warnings and noise.
• Themed Aesthetics: All CLI colors have been themed to a blueviolet, matching the Gradio UI and the new banner for a consistent, polished look.
• Graceful Shutdown: You can now press Ctrl+C in the terminal to shut down the application cleanly and gracefully.
• Robust Error Handling: Added fallback error messages for models that are corrupted or of an unsupported type, guiding the user instead of crashing.

⚙️ Backend & Performance

• Professional Architecture: The entire application was refactored from a single script into a professional, multi-file structure separating UI (ui.py), logic (app.py), and pipelines (pipelines/). This makes the code cleaner, more stable, and easier to maintain.
• Rock-Solid Model Detection: The logic to differentiate between SD1.5 and SDXL models has been rewritten to be 100% reliable by directly inspecting the model's tensor keys.
• Universal Callbacks: The backend now uses a modern, universal progress callback system that works flawlessly with both SD1.5 and SDXL pipelines.
• Targeted Suppression: Implemented surgical-precision code to suppress all C++ and Python warnings during startup, ensuring a pristine console without causing race conditions or crashes.

📦 Installation & Packaging

• Conda-Based Environment: The setup process has been standardized to use a dedicated Conda environment named arttic-lab, ensuring no conflicts with other Python projects on your system.
• Installer & Launcher Scripts: Added install.bat/.sh and start.bat/.sh scripts for a simple, one-click installation and easy launching experience.

---

A Note on Our Journey

ArtTic-LAB wasn't built in a vacuum. It was forged through a rapid and intensive collaboration between a developer and a visionary lead tester. A massive thank you to our Lead Tester for their invaluable feedback, rigorous testing, and clear vision that pushed this project from a proof-of-concept to a truly polished application. Every bug found and every feature requested has made this tool better for everyone.

What's Next? 🔮

The journey doesn't end here! The v1.0.0 release provides a rock-solid foundation. We're already looking at what's next, including:

• Image-to-Image (Img2Img) functionality
• Inpainting and Outpainting tools
• One-click upscaling for generated images

---

Thank you for believing in this project. We can't wait to see what you create with ArtTic-LAB!

Now, go make something amazing. ✨