For most cases the Auto option is a great solution. Higher upscale like 16х can give less accurate results since the network is tasked to generate more details from the same input. 

Try to hit the needed output size for your use case. If you know the numbers you can use Custom size input.

Please check out our tips to choose the right algorithm or a preset that fits your use case the best.

Now users can choose the AUTO option or select one of these algorithms based on the picture you have:

Photo (Legacy) algorithm works for images of architecture, nature, people etc. Most photos captured by a mobile device or digital camera qualify for the Photo algorithm.
Illustration is recommended for digital art (drawings, illustrations, paintings, cartoons, anime, logos, etc)
Photo 2.0 (beta) is a redesigned algorithm of Photo (Legacy). Better image compression removal, higher upscaling accuracy and detail preservation
Photo 2.0 Faces (beta) is recommended for compressed or pixelated images with faces. It is used to rebuild face details.

If you have a particular use case in mind check out our Presets tab to find the right operation:

Real Estate to automatically adjust property images
E-commerce to automatically adjust product images
One of the Printing presets to optimise photos for printing in top quality
Was this article helpful?
Thank you!