Release notes
2.1.0
New Features & Improvements
Updated Torch-CUDA Version: Upgraded
torch-cuda
to version 2.5.1+cu124 to ensure compatibility with the latest CUDA updates and improve performance.
2.0.9
Permission Updates: The following processes are now configured to use UID 1200 and GID 1200:
Blur Worker
Directory Worker
Manager API
Technical Details
These updates improve permission management and enhance security for service execution.
Why UID/GID 1200? This choice ensures compatibility across different systems and avoids conflicts with system-reserved or commonly used IDs. It provides a dedicated user/group for these services, improving security and isolation.
2.0.8
New Features & Improvements
Added asynchronous CPU and GPU processing to improve performance.
Introduced a new feature to clear the GPU cache for better resource management.
2.0.7
New Feature: Intuitive Dashboard for Task & Consumption Tracking
We’ve added a Dashboard page to give you a clear overview of service usage and resource consumption.
🔹 Key Features:
📊 Service Usage – Total API calls, daily & monthly breakdown 📉 Consumption Tracking – Data usage (MB) per day & month 📆 Dynamic Charts – Yearly & monthly trends for better insights
2.0.6
Starting from this version, custom port configuration has been removed. Applications are now accessible exclusively through the standard ports 80 (HTTP) and 443 (HTTPS). Use the dedicated paths /api
and /flower
to access the API and the Flower
2.0.2
New Features:
SSL configuration for cloud version
Hotfixes:
Web App: Improved input validation to allow spaces in the input path.
BlurIt Model: Resolved an issue with handling videos with low FPS.
2.0.1
New Features:
Added HTTP Basic Authentication for the web application, enhancing security.
2.0
New Features:
High-Performance Algorithm : This new algorithm delivers enhanced computing power and speed, resulting in better overall performance for your tasks.
Lightweight Model : In addition to the high-performance algorithm, we’ve introduced a smaller, more efficient model that uses less GPU RAM, making it ideal for environments with limited resources. Despite its compact size, it still provides robust results, ensuring you don't have to compromise on accuracy or efficiency.
1.1.0:
CLI configuration all in one
1.0.0:
installation script
NEW CLI
fix: not processing image or video with uppercase extension (e.g. image.PNG)
Last updated