Bigjpg

Bigjpg enlarges images by 2x, 4x, 8x, or 16x using deep convolutional neural networks. Instead of stretching pixels, the model rebuilds lines and fine detail, which is what keeps a small file looking sharp after a large size jump. Free accounts are limited to 4x, while paid plans unlock the full 16x range.

The upscaler ships two models tuned for different content. Art handles anime, manga, and illustration, keeping flat colors and crisp outlines intact. Photo targets real-world shots and works on portraits, products, and scenes. Picking the right profile matters, because the wrong one can smooth away the texture a drawing depends on.

A noise setting lets you dial in how aggressively the model cleans an image, with levels from none up to highest. This helps when a source file is grainy or heavily compressed, since you can remove artifacts before the upscale runs. It is a small control that often decides whether the result looks natural or over-smoothed.

Paid plans run several images at once on a dedicated high-performance server. Batch mode suits designers and studios working through a set of icons, textures, or photos in one sitting, rather than uploading files one at a time. The free plan processes two images at a time on shared servers.

A paid API connects Bigjpg to your own tools. You call an upload endpoint to get a temporary storage URL, push the file, then send an enlarge task with your chosen style, noise, and ratio. Results come back as task lookups, so a script can scale images without anyone touching the website.

Bigjpg runs in any browser and offers Windows, Mac, iOS, and Android apps. Offline enlarging is available once you log in, so you can close the browser while a task finishes. The settings page also lets you switch the download server between China and the United States for faster results.