Suggestion Box Color Meaning and Tolerance

We just upgraded our software to a new version. There were no changes made to this version that deals with the look and feel (GUI) of our software. We have been hit by eggplant with failed image searches more than usual though with this update.

When eggplant fails to find the image, it will pause the script and give a dialog box. Then it will search the window for images that look similar and make recommendations. Sometimes these recommendations are WAY OFF, highlighted in red boxes, while other times it correctly finds the image and highlights them with orange or green boxes.

Is there a way to change the acceptance level? For example, if eggplant finds an image that would be considered a “green box” suggestion, go ahead and click on it and continue.

Additionally, what do the different box colors mean?

There should be a tab in the dialogue box that pops up when you can’t find an image called Diagnostics. It should list the search types that can be done by eggPlant and any locations that meet their criteria. Each entry is colored the same as the rectangle it will draw on the connection window.

The diagnostic searches conducted are:

Standard Search: Searches for the image again using the original specifications. A result for this search means that the image has appeared since the first failure to find the image. Most likely this means that the timing of the script at the step needs to be adjusted to allow more time for the desired image to appear.
Dynamic Tolerance: Searches for the image while increasing the RGB color tolerance value up by a particular increment (default = 5) until a match is found. A result for this diagnostic means that the RGB values of the image have shifted.
Alternate Types: Searches for the image while using the different SearchType filters, which are Text, Pulsing, and Text and Pulsing. If the image is text-based, a result indicating Text suggests that the SUT is using anti-aliasing in its text. A result for Pulsing or Text and Pulsing suggests that the background portion of the image either has a pulsing effect or perhaps another effect such as a variable gradient.
Scaling Search: Searches for the image at a different scale than the original image.
Discrepancy Search: Performs the search while allowing a certain percentage of the pixels in the image to differ.
OCR Search: Scans the original captured image for text, and then conducts a search for that text. For example, if the original image is of a button that says “Search” on it, this search will use OCR to look for the word “Search” on the screen.
Prior Location: References the results for the current script in which the current line of code was successfully run. Highlights the location on the screen where the image was found in that successful run of this test. The results that this search uses can be specified manually using the “Mark As Prior Run” button in the results portion of the suite for any script. Note: This search is not used when the Image Doctor is set to “Auto”.
Original Location: Lists the coordinate pair of the location at which the image was first capture.

There may be a way to programatically to specify the search type in eggPlant. I did a quick search of the documentation and I didn’t find anything.

Thanks, sphennings, for a great description of the different diagnostic searches that are done!

Is there a way to change the acceptance level? For example, if eggplant finds an image that would be considered a “green box” suggestion, go ahead and click on it and continue.

This is what the Auto Doctor does. If you select Run > Update Image > Auto Doctor from the menu then instead of showing the panel, eggPlant will try certain diagnostics. If one of the diagnostics allows eggPlant to find the image exactly once on the screen then the script will use that location and go on.

A warning is entered in the log when that happens, allowing you to review it later. There is even an action available on the Results tab that will let you apply the needed changes to the image so it will be found in the future.

This is a great synopsis of what I was looking for. I will play around with these, especially Auto Doctor.