image-quality-enhancement

To enhance blurry images and improve sharpness, this skill uses a Laplacian variance metric to automatically assess image quality and applies AI upscaling with Real-ESRGAN via Replicate to restore detail.

Laplacian variance works for image quality assessment by calculating the variance of the image's second derivative, providing a quantifiable sharpness metric to decide if AI upscaling is necessary for low-quality inputs.

Yes, you can use Real-ESRGAN with Replicate to upscale low-resolution images by providing a local file or URL, which the workflow processes to output an enhanced image ready for design workflows.

The best way to automate image enhancement for design workflows is using this script-based approach, which evaluates incoming image URLs or files and conditionally applies AI upscaling to output optimized assets automatically.

Yes, you need OpenCV to assess image sharpness before upscaling, as the workflow relies on opencv-python to compute the Laplacian variance metric that determines whether Real-ESRGAN enhancement is required.