Current Status of Improved Selection for Cut Outs

[drdancm]

The Affinity Quick Select Tool is quite good and I find that unlike the same tool in Photoshop CS6, Affinity QSelect will clear out fully closed areas of background (as between strands of hair).  However there is room for improvement.

For example Photoshop Elements 2020 (latest version of a competitor for Affinity Photo) now has an AI powered Quick Select Tool feature that is improved over the prior Quick Select for objects such as people. In the Edit mode simply clicking on a single Subject choice, P Elements, will generally select the person, or animal quite accurately in an image with a simple mostly solid background. With a more complex background the reliability is less good. A common error is failure with smaller areas, for example, such as the fully enclosed area between hands on hips.

Topaz Labs has just recently released Mask AI and while the very first commercially available version had lots of problems, subsequent version have shown improvement. Mask AI comes up with the image painted with red (signifiying background) and the user must outline the object to be cut out with a blue transition marker. After filling the blue enclosed area with green marker (object area to be kept) pressing the process button starts the AI processing. Once again, when challenged with a object figure against a fairly uniform background, it tends to work well, however it takes a lot longer to prepare an image for processing than with Photoshop Elements 2020, which works with a single click.

I and many other happy users of Affinity Photo, would very much love to have an improved Quick Select Tool, which probably means, one that is based on AI, since that technology appears to results in some amazingly capable competence.

So this comment is simply to let everyone who is interested keep up to date with the latest technological advances and hope that Affinity Photo will also create improvements for this tool and do it even better than the competition.

Thanks for listening,

Dan