跳至内容Low-resolution screenshots, old scans and compressed JPEGs look soft when enlarged with normal resize. Tentaco's AI upscaler uses ESRGAN super-resolution to reconstruct detail at 2× or 4× — ideal before printing, cropping zoomed regions or fixing tiny e-commerce assets.
Everything runs in your browser after the model loads once. Your image is not sent to a cloud GPU farm. That makes Tentaco a practical alternative to Let's Enhance, Bigjpg or Photoshop's super resolution for quick one-off upscales.
Choose 2× for social posts and web hero images; use 4× for small logos, avatars or archival scans that need maximum pixel count. Output stays PNG for lossless follow-up editing.
Your image stays on-device during upscaling. Only the ML weights download from the network.
Choose 2× (double width and height) or 4× (four times each dimension). Output size depends on your source image.
Yes for photos and textures — super-resolution predicts missing detail. Pure vector art may not improve as much.
No. Upscaling runs in your browser with WebGL/CPU after the model caches locally.
Very large images are slower and memory-heavy. For best speed, upscale images under ~2000 px on the long edge first.
2× or 4× helps small sources reach acceptable print DPI. Always check final dimensions in mm/inches before ordering prints.
