Image recognition is very generous :-)

The Search for image action (above) also finds these two rather different images in addition to the selected image. Which is good in this specific case. I'm just wondering why the color of the background doesn't matter (blue/gray), nor does the direction of the arrow. Is that how it should be?

image
image

It mostly depends on the selected recognition treshold. However the next version of BTT will also allow you to select from different matching algorithms that require different levels of accuracy

I played around a bit with the treshold. Something is changing, but it's difficult to say what. I think with the value 0.6 the color orange plays the decisive role. If I change it, recognition becomes much more difficult.

However, the fact that everything is recognized is exactly what I want :slightly_smiling_face:

the latest alphas make it a bit easier to work with this.

There is now an option to take color into consideration when matching and there is a "test search" button, that will show you the treshold value that was required:

1 Like

I experimented a bit with the new options. To be honest, I'm still not quite sure when which option is best. A longer field test will show that.

But I'm pretty sure that a “search only in focused window” option would make the search even easier for BTT. (Instead of: Search on screen with focused window)

yes that I‘ll add before the next stable release. I think I now have a pretty good idea how to set the values for perfect matching, I‘ll add some docs for that

Mm, an addition, I don't know if this is possible, but it would be extremely good if you could even limit the search area of the front window.

So, search only in the upper right quadrant of the focused window. Or only in the lower half.

The more information BTT has, the faster and more reliable the search should be, I think.

v4.681 alpha adds "focused window", "around mouse cursor" and allows to further limit the search area by quadrants or halves. Documentation will follow soon :slight_smile:

Very, very nice. Thank you very much! :smiley: