Upscaling

Traditional upscaling adds pixels by estimating values between existing pixels. Bilinear, bicubic, and Lanczos interpolation can make an image larger, but they cannot recover detail that the camera never captured. AI upscaling, often called super-resolution, predicts plausible textures and edges from patterns learned during training, so the result may look sharper than ordinary interpolation.

Choose the scale from the intended output rather than enlarging as much as possible. A 2× enlargement doubles width and height and creates four times as many pixels; 4× creates sixteen times as many. Start with the sharpest, least-compressed source, then inspect faces, text, logos, fine lines, and repeating patterns at the final display or print size. AI-generated detail is an estimate, not recovered evidence, and stronger enlargement can introduce altered letters, unnatural facial features, halos, or invented texture. Magic Eraser’s AI Enhance can upscale and sharpen images; keep the original and verify the result before archival, documentary, or other accuracy-sensitive use.