Having issues while using visual commands on Kantu

I have tried Visual Commands in Kantu, they’re the best thing. But I found 2 problems that i don’t know how to fix it.

Problem #1

If i give my macro to my friends, it’s not going to work to them so how can i make it work for them

Problem #2

I tried visualAssert and to see it it needs to be shown fully in your screen now like you need to see it not like click. I don’t know how to make the page scrolldown or up and that’s the issue im having

any help about this

Problem 1: Our goal is that the machine vision works so well, that you can give the macro to your friends (= run on different machines) without any tweaking. That said, we are not there yet 100%. What you can do then is to tweak the so-called “confidence level”. It tells Kantu how similar the input image and the image found in the website need to be.

What to do depends on the type of issue you have:

If the wrong image is found => increase the confidence level eg. 0.9

If no match is found => lower the the confidence level, e. g. 0.5 is often a good value.

There are two ways to do this:

In the replay settings you find the “default vision search confidence” box. By default it is set to 0.8 (1 = exact match needed… 0.1 = almost anything matches).

You can do the same on a per command level by adding @0.xx to the image, example

VisualAssert | testimage.png@0.50

Problem #2:

By default, the vision commands search on the currently visible part of the web page (called “viewport”).

But you can use the visionLimitSearchArea | full command to tell Kantu to search the complete website for an image. The only drawback here is that this takes longer, as full page screenshots need to be stitched together from single viewport screenshots (this is a browser limitation). For more info see https://a9t9.com/kantu/docs/visual-ui-testing#visionlimitsearcharea

1 Like

Isn’t there anyway to make the page scroll down a bit and then checks if there is the image and then presses it because problem 2 didnt work.