Iván Eduardo Chavez Ayub
14 KiB
14 KiB
Warlock-Studio is an open-source desktop application for Windows, engineered to integrate state-of-the-art AI models for image and video enhancement.
Inspired by Djdefrag tools such as QualityScaler and FluidFrames, Warlock-Studio provides a unified, high-performance platform for upscaling, restoration, denoising, and frame interpolation.
📥 Download Installer (v4.2.1) from:
|
|
|
📚 User Manuals & Documentation
The documentation provides detailed technical explanations, troubleshooting guides, and best practices.
|
🖼️ Interface Previews
Main Window
Console
✨ Key Features
- AI Upscaling & Restoration – Utilize Real-ESRGAN, BSRGAN, and IRCNN models for denoising, super-resolution, and detail recovery.
- Face Restoration (GFPGAN) – Recover facial details from low-resolution or blurry images and video frames.
- Frame Interpolation (RIFE) – Smooth motion or generate slow-motion content with 2×, 4×, or 8× interpolation.
- Advanced Hardware Acceleration – Intelligent provider selection prioritizes CUDA, falls back to DirectML, and finally CPU for maximum compatibility and performance.
- Batch Processing – Process multiple media files simultaneously, saving time and effort.
- Custom Workflows – Fine-grained control over models, resolution, output formats, and quality parameters.
- Open-Source & Extensible – Fully MIT licensed, for contributors and developers.
🆕 What's New in v4.2.1
This maintenance release focuses on critical stability, usability, and a significant visual overhaul.
Stability & Usability Fixes
- ✅ Critical Audio Fix: Resolved a major bug that caused all generated videos to be encoded without an audio track. The missing
ffprobe.exebinary has been restored, ensuring proper audio stream detection and passthrough/re-encoding. - ✅ UI State Resilience: Fixed an issue where conflicting AI model settings (e.g., blending and frame interpolation options) could persist incorrectly when switching between model types, preventing invalid setting combinations.
- ✅ Robust Thumbnails: Reworked the file thumbnail generation process (
FileWidget.extract_file_icon) to be more robust, preventing potential crashes related toPillow/numpycompatibility issues during image conversion. - ✅ Fixed Window Size: The main application window is now non-resizable (
window.resizable(False, False)) to ensure a stable and predictable UI layout, preventing widget placement errors. - ✅ Dialog Behavior: Removed the aggressive "always-on-top" flag (
-topmost) from dialog windows, ensuring they behave as standard application-modal windows.
Visual & Thematic Redesign
- 🔥 "DarkRed" Theme: Introduced a significant visual overhaul with a high-contrast, dark-mode palette. The theme uses a deep red background (
#480B0B) with bright yellow (#FFE32C) and white text for enhanced readability and a distinct, technical aesthetic. - 💻 Monospaced Font: Changed the global font from
"Segoe UI"to"Consola"for a uniform, technical look across all UI elements, enhancing readability for file names and status messages.
🖼️ Quality Comparison
Enhanced image using BSRGANx2:
AI-Driven Visual Transformation: Before & After (Click on the image to open the video on YouTube)
📊 Model Comparison
| Model File | Use Case | Speed | Quality | Notes |
|---|---|---|---|---|
| GFPGANv1.4 | Face restoration | High | High | Optimal for portraits |
| BSRGANx2 | 2× upscale + denoising | Medium | Very High | Suitable for lightly degraded media |
| BSRGANx4 | 4× upscale + denoising | Low | Very High | For heavily degraded content |
| RIFE | Frame interpolation | High | High | Smooth motion, slow-motion support |
| RIFE-Lite | Lightweight interpolation | Very High | Medium | Faster, lower resource usage |
| RealESRGANx4 | General 4× upscaling | Medium | High | Balanced performance |
| RealESRNetx4 | Subtle restoration | Medium | High | Preserves natural image texture |
| RealSRx4_Anime | Anime / line-art enhancement | Medium | High | Sharp edges for 2D art |
| IRCNN_L | Light denoising | High | Medium | Mild artifact removal |
| IRCNN_M | Medium denoising | High | Medium | Stronger artifact cleanup |
🖥️ System Requirements
- OS: Windows 11 or higher (64-bit)
- RAM: 8GB+ recommended
- GPU: NVIDIA (for CUDA), AMD, or Intel GPU with up-to-date drivers recommended
- Storage: Sufficient free space for input and processed media
📌 Development Status (v4.2.1)
| Component | Status | Notes |
|---|---|---|
| ONNX Runtime Engine | 🟢 Enhanced | Prioritizes CUDA > DirectML > CPU with automatic fallback. |
| Installer & Packaging | 🟢 Overhauled | Full offline installer; heavily optimized package size. |
| Video Audio Passthrough | ✅ Fixed | Resolved critical failure to include audio in encoded videos. |
| UI State Persistence | ✅ Fixed | Settings no longer conflict when switching AI models. |
| Upscaling Models | 🟢 Stable | Includes VRAM recovery integration. |
| Face Restoration (GFPGAN) | 🟢 Stable | High-quality face reconstruction. |
| Frame Interpolation (RIFE) | 🟢 Stable | Smooth motion and slow-motion support. |
| User Interface (UI/UX) | 🟢 Redesigned | New high-contrast "Inferno" theme. |
| Code Quality | 🟢 Improved | Refactored, modular, and more maintainable. |
📂 Project Structure
Warlock-Studio/
├── AI-onnx/ # Pre-trained ONNX models for AI processing
│ ├── BSRGANx2_fp16.onnx
│ ├── BSRGANx4_fp16.onnx
│ ├── GFPGANv1.4.fp16.onnx
│ ├── IRCNN_Lx1_fp16.onnx
│ ├── IRCNN_Mx1_fp16.onnx
│ ├── RealESR_Animex4_fp16.onnx
│ ├── RealESR_Gx4_fp16.onnx
│ ├── RealESRGANx4_fp16.onnx
│ ├── RealESRNetx4_fp16.onnx
│ ├── RealSRx4_Anime_fp16.onnx
│ ├── RIFE_fp32.onnx
│ └── RIFE_Lite_fp32.onnx
│
├── Assets/ # Application assets and third-party binaries
│ ├── banner.png
│ ├── clear_icon.png
│ ├── exiftool.exe
│ ├── ffmpeg.exe
│ ├── ffmprobe.exe
│ ├── ffmplay.exe
│ ├── info_icon.png
│ ├── logo.ico
│ ├── logo.png
│ ├── stop_icon.png
│ ├── upscale_icon.png
│ ├── wizard-image.bmp
│ └── wizard-small.bmp
│
├── rsc/ # UI previews and branding resources
│ ├── Capture.png
│ ├── image_comparison.png
│ ├── CaptureCONSOLE.png
│ └── GitHub_Logo_WS.png
│
├── Warlock-Studio.py # Main application script
├── Warlock-Studio.spec # PyInstaller specification file
├── Setup.iss # Inno Setup installer script
├── README.md # Project overview
├── CHANGELOG.md # Version history and updates
├── LICENSE # MIT License information
├── NOTICE.md # Legal notices and attributions
├── CODE_OF_CONDUCT.md # Contributor guidelines
├── CONTRIBUTING.md # Contribution guide
└── SECURITY.md # Security reporting policies
📊 Integrated Technologies & Licenses
| Technology | License | Author / Maintainer | Source |
|---|---|---|---|
| QualityScaler | MIT | Djdefrag | GitHub |
| FluidFrames | MIT | Djdefrag | GitHub |
| Real-ESRGAN | BSD 3-Clause / Apache | Xintao Wang | GitHub |
| GFPGAN | Apache 2.0 | TencentARC / Xintao Wang | GitHub |
| RIFE | Apache 2.0 | hzwer | GitHub |
| BSRGAN | Apache 2.0 | Kai Zhang | GitHub |
| IRCNN | BSD / Mixed | Kai Zhang | GitHub |
| ONNX Runtime | MIT | Microsoft | GitHub |
| FFmpeg | LGPL / GPL | FFmpeg Team | Official Site |
| ExifTool | Artistic License | Phil Harvey | Official Site |
| Python | PSF License | Python Software Foundation | Official Site |
| PyInstaller | GPLv2+ | PyInstaller Team | GitHub |
| Inno Setup | Custom | Jordan Russell | Official Site |
🤝 Contributions
We welcome contributions from the community:
Fork the repository.
Create a branch for your feature or bug fix.
Submit a Pull Request with a detailed description and testing notes.
📧 Contact: negroayub97@gmail.com
📜 License
© 2025 Iván Eduardo Chavez Ayub
Licensed under MIT. Additional terms and attributions are provided in NOTICE.md.