The latest features, improvements, and fixes shipped to the GrainLux platform. We ship updates weekly.
You can now serve images in JPEG XL format using f_jxl. Delivers up to 86% compression with superior quality retention compared to legacy JPEG. Automatic fallback to WebP for unsupported browsers when using f_auto.
Completely rebuilt ML model for background removal. 3x faster processing (under 200ms), significantly improved edge detection for hair, fur, and semi-transparent objects. Now supports batch processing of up to 100 images per request.
Generate HLS and DASH manifests automatically from uploaded videos. Supports up to 4K resolution with configurable bitrate ladders, automatic thumbnail sprite generation, and preview clip extraction.
Enhanced face detection accuracy by 40% across diverse demographics. New object-aware gravity mode detects products, text regions, and logos for intelligent cropping in e-commerce and editorial scenarios.
Added CDN points of presence across South America (São Paulo, Buenos Aires, Bogotá), Africa (Lagos, Nairobi, Cape Town), and Southeast Asia (Jakarta, Bangkok, Ho Chi Minh City). Total network now spans 300+ locations.
Receive real-time notifications when async transformations complete, usage quotas are reached, or processing errors occur. Supports custom endpoints with configurable retry logic and HMAC signature verification for security.
Doubled AVIF encoding speed through hardware acceleration and optimized encoder settings. AVIF transformations now complete in under 100ms for typical web images, making it viable for real-time delivery.