I am evaluating Keyboard Maestro (trial version) and trying to build a macro to automate my photo editing workflow.
The Goal: I got the idea for this setup from this tutorial: How to Automate Your Imagen AI Workflow. Once Imagen AI finishes processing an edit catalog, I want Keyboard Maestro to automatically click the "Review Edits" button to download and import the edits into Adobe Lightroom Classic, and then trigger 1:1 previews.
The Problem: The Click at Found Image action fails to detect or click the target image on screen during normal macro execution. To troubleshoot this, I downloaded the exact macro provided in the article, but the image recognition feature still fails to execute on my setup.
Troubleshooting completed so far:
Downloaded macro test: Tested the premade macro from the tutorial; the "Click at Found Image" action fails during execution.
Isolated simple test: Created a minimal test macro containing only the "Click at Found Image" action targeting a simple UI element (the "GO" button in the Speedtest app). It still fails during macro playback.
Fuzziness adjustments: Adjusted the fuzziness slider across all levels without success.
The "Try" button behavior: When I open the target image in Finder and click the Try button inside Keyboard Maestro while the app is visible, it works instantly.
Because the "Try" button works when the image is open in finder, screen capture permissions seem to be fine, but execution fails every time the action runs as part of the actual macro flow.
(Screenshot of the macro attached below)
Has anyone experienced this behavior or found a solution for image detection issues in similar app automation flows?
Are you using Control + Shift + Command + 4 to capture the area of screen containing the image you want to use and pasting Command + V this into the image well of the Click at Found Image Action?
This is the best way to get the image as it is pixel-perfect and that it is what Keyboard Maestro will be matching.
Not getting the image into the Action this way is the most common reason for this Action not working reliably.
Which suggests you screenshot to a file, then copied from that file to put into KM. As @Zabobon says, that often fails -- the saved image has no resolution set and will default to 72dpi, which probably doesn't match your screen's resolution.
⌃⇧⌘4, drag over what you want to capture, Paste into the Action's image well.
You should also limit your search area, especially when using the "Unique" constraint, to limit the chance of false positives. Since your first "Pause" only checks the front window my bet is you can limit the other Actions to the front window too. And try using the "Best" constraint so the Actions don't fail when there are other, sort-of-but-not-really, matches in the search area.
