1.0.2

Smart Segments v1.0.2 - CUDA Compatibility Enhancement 🚀

🎯 Major Fix: CUDA Compatibility Issues Resolved

This release addresses critical CUDA compatibility problems that were causing crashes for users with GPU driver/PyTorch version mismatches. The plugin now works reliably on all systems with automatic CPU fallback when needed.

🔧 What's Fixed

Critical: "CUDA kernel image not available" Error

• Problem: Plugin crashed with CUDA error: no kernel image is available for execution on the device
• Solution: Added intelligent CUDA compatibility testing and automatic CPU fallback
• Result: Plugin continues working seamlessly in CPU mode when GPU issues occur

✨ New Features

Smart GPU Detection

• 🔍 Real CUDA Testing: Actually tests GPU operations instead of just checking availability
• 🔄 Automatic Fallback: Seamlessly switches to CPU mode when CUDA compatibility fails
• 📊 Detailed GPU Info: Reports GPU memory, CUDA versions, and compatibility status

Enhanced User Experience

• 💬 Friendly Error Dialogs: Clear explanations instead of cryptic technical errors
• 🛠️ Troubleshooting Guidance: Actionable steps to resolve GPU compatibility issues
• 🛡️ Graceful Degradation: Full functionality maintained in CPU mode (just slower)

🎯 For Users Experiencing CUDA Issues

If you've been getting CUDA errors, this update will:

1. Automatically detect the compatibility problem
2. Switch to CPU mode without crashing
3. Show a helpful dialog explaining the situation
4. Continue working with full functionality

🔧 Technical Improvements

• New CUDACompatibilityChecker class for comprehensive GPU analysis
• Enhanced model loader with device fallback mechanisms
• Improved error categorization and logging for CUDA-related issues
• Specialized error dialogs with platform-specific troubleshooting

📦 Installation

Download the plugin package and install via Krita:

1. Download SmartSegments_v1.0.2.zip
2. Open Krita → Tools → Scripts → Import Python Plugin from File
3. Select the downloaded ZIP file
4. Restart Krita

🐛 Reporting Issues

If you encounter any problems, please open an issue with:

• Your GPU model and drivers
• PyTorch version (if known)
• Error messages from the plugin
• Log from plugin folder ./log/smart_segments.log

💾 System Requirements

• Krita: 5.0+ with Python support
• RAM: 4GB+ recommended
• Storage: 5-6 GB free space (1 GB models + 4-5 GB tooling)
• GPU: Optional (automatic CPU fallback if incompatible)

---

Note: This release ensures the plugin works on all systems regardless of CUDA compatibility. GPU acceleration is nice-to-have, but the plugin is fully functional in CPU mode.

1.0.1

🎨 Smart Segments for Krita - v1.0.1

Windows Compatibility Hotfix

This patch release addresses critical Windows compatibility issues that were preventing proper installation and operation on Windows systems.

🐛 Bug Fixes

Windows Platform Improvements

Fixed: Virtual Environment Creation on Windows
• Issue: Virtual environment creation was failing on Windows due to Krita's embedded Python interpreter conflicts
• Solution:
• Added intelligent Python interpreter detection that prefers Windows Python Launcher (py -3) when available
• Implemented fallback mechanisms to avoid using Krita.exe as the Python interpreter
• Added proper detection and handling for Windows-specific Python installations

Enhanced: Bootstrap Process for Windows
• Improved path handling for Windows file systems
• Added robust fallback methods for virtual environment creation
• Better error handling and logging for Windows-specific issues
• Fixed Script directory path resolution on Windows

Improved: Model Loading and Inference
• Resolved path separator issues affecting model loading on Windows
• Fixed compatibility issues with Windows file system operations
• Enhanced error recovery for Windows-specific edge cases

📦 What's Included

• Distribution Package: Pre-packaged SmartSegments_v1.0.1.zip for easy installation
• Updated Bootstrap: Completely rewritten bootstrap process with Windows-first approach
• Enhanced Logging: Better debugging information for troubleshooting Windows installations

💾 Installation (Windows)

1. Download SmartSegments_v1.0.1.zip from the releases
2. Open Krita, Menu Tools -> Scripts -> Import Python Script from File... Select File Above.
3. Restart Krita
4. Enable the plugin via Settings → Configure Krita → Python Plugin Manager
5. Run the setup wizard from Tools → Scripts → Smart Segments Setup

🔧 Technical Details

Files Modified
• smart_segments/bootstrap.py - Windows-compatible virtual environment creation
• smart_segments/minimal_bootstrap.py - Enhanced Windows Python detection
• smart_segments/core/inference_engine.py - Windows path handling fixes
• smart_segments/core/model_loader.py - Cross-platform model loading
• smart_segments/smart_segments.py - Main plugin Windows compatibility updates

✅ Tested On
• Windows 10 (21H2, 22H2)
• Windows 11 (22H2, 23H2)
• Krita 5.0.0 - 5.2.3

🚀 Quick Start

After installation, if you encounter any issues:

1. Check the log file at %APPDATA%\krita\pykrita\smart_segments\logs\smart_segments.log
2. Ensure Python 3.7+ is installed on your system
3. Try running the minimal bootstrap from Krita's Scripter

📝 Notes

This is a compatibility patch release focused exclusively on Windows platform issues. Linux and macOS users running v1.0.0 without issues do not need to update, though the update is backwards compatible.

🙏 Acknowledgments

Special thanks to our Windows users who reported these issues and helped with testing the fixes.

@CainTagwell
@yoshyteru
@SimonH (Krita-Users)
@yosh1 (Krita-Users)
@sanjeevan_madhu (Krita-Users)

Full Changelog: 1.0.0...1.0.1

1.0.0

🧠 Initial Release – Smart Segments v1.0.0

Released by Hawkins.Tech Inc

We’re excited to launch Smart Segments, a Krita plugin that brings AI-powered image segmentation directly into your creative workflow using Meta’s Segment Anything Model (SAM v2).

---

✨ What It Does

• Automatically segments the current layer using SAM v2
• Lets you click or Shift+Click segments to select one or more areas
• Converts selected segments into a Krita selection mask
• Supports both GPU (CUDA) and CPU-only systems
• Cross-platform: Windows, macOS, and Linux supported

---

🚀 Features

• Fully self-contained: no external installs or CLI steps required
• Automatically creates and manages its own Python virtual environment
• Downloads SAM v2 checkpoint and dependencies on first run
• Fast local inference using PyTorch
• Qt-based overlay to select and combine segments before applying them

---

🛠 How to Install

1. Download this ZIP and extract it into your Krita plugin directory:
   • Linux: ~/.local/share/krita/pykrita/
   • Windows: %APPDATA%\krita\pykrita\
   • macOS: ~/Library/Application Support/krita/pykrita/
2. Restart Krita
3. Go to Tools → Scripts → Manage Python Plugins and enable Smart Segments
4. Launch via Tools → Scripts → SAM Segment Picker

---

🧪 First Run Notes

• Downloads ~375MB SAM model and sets up a Python venv
• Detects and uses GPU automatically if available
• Starts a local inference server in the background during use

---

📦 Requirements

• Krita 5.2 or later
• Python 3.8+
• Internet connection on first launch

---

Let us know if you run into issues, have feature requests, or want to contribute.

🖼️ Smart Segments — segment smarter, not harder.
👨‍💻 Released by Hawkins.Tech Inc