Explore viability of multi-density images as UI element identifiers - use this line

[kensoh]

For eg, capturing screenshots on a high-density computer, and use in lower-density PC through a config option.

Ie downsize on the fly before execution. Otherwise computer vision matching will break on different screen densities.

Source - Telegram Group Chat

[kensoh]

Adding on inputs from @1RedTea

Hi Kaushik, i just tried out a flow with different resolution and it works. Ken is spot on saving the higher resolution screen shot instead of a lower resolution Screenshot. I use PNG for all my Screenshot instead of Jpg. ☺️

[kensoh]

Solution can be as simple as running below. Where 0.5 is the resizing that is done to provided input images.

vision Settings.AlwaysResize = 0.5

[kensoh]

Solution is take image snapshots at the computer with the 'largest' images. Note that big resolution doesn't mean large image. Because icon size will still be the same although there is a larger screen viewing and search area.

Probably this means taking image snapshots at the computer using the largest % zoom. For eg 125% zoom. Then to run on a computer with 100% zoom, set vision Settings.AlwaysResize = 0.8

If this works, to explore having an option in tagui_config.txt for easier usage instead of running that line for every robot.

[kensoh]

Above solution found from a similar problem at RPA for Python - tebelorg/RPA-Python#170 (comment)