Bookmarking in visual search results

[Carlo Didier]

I found a folder with images where I'm not sure if I have them all processed and classified in iMatch.

So I added that folder to search for visually similar images. Apart from the astonishing (meaning strange) results I got, I am not able to select a master to bookmark it if it is not found, so that later I can process only those who were missing.

What I mean is, in the screenshot, on the top you see one of the masters for the search and the "matching" images iMatch found (only the second match has some visual resemblance with the master, apart from the fifth match which is the correct one ...). I can select one or more of the matches, but not the master to somehow mark it (bookmark for example or assign a pin, whatever).

As is, this search result is useless for me if I can't bookmark the master where a corresponding match has been found.

[Mario]

1. Experiment with different settings. This is not hard science.
2. Make sure you use the Defautl sort order, which sorts by similarity. Else the results appear in random order.
3. Read the Result Window help again: The Result Window which explains how originals in the result are presented that way.
4. You can also use the enable corresponding setting so your originals are displayed as part of the result, which not only allows you to bookmark them but also use other commands. This is explained in the Result Window help.

[Carlo Didier]

I had the default sort order active in the result window. Switching to "Similarity" put the most similar first. Still, 3 out of 5 images it found have practically no similarities at all with the original ... But for my purpose here that's irrelevant as I only want to find identical images.

As I rarely use a result window, and usually never search by similarity, I had forgotten the option to include the original in the results. I can now bookmark them, but it would be clearer ("übersichlicher") to display it apart from the results and still be able to tag the original.

Thanks for the fast help (as always).

[Mario]

The original thumbnail in the group header is just that - a thumbnail. You can right-click it to open a menu with frequent command, like jumping right to the original in the folder. Then you can perform whatever operations you want. To find duplicates, use the Ctrl+M,D search, which does not look for visually similar images but only for true duplicates (by checksum).

I'm not sure what you expect from the search for visually similar images, or which settings you use (by shape, color, textture, ...) This is not an exact science but it usually allows you to bring the files down from 100000 potentially similar images to maybe a few hundred. And that's already quite a lot, I can tell you. And that's all this is designed for.

If you don't use a feature regularly, it is probably impossible to remember how everything works. A quick press of <F1> to open the help topic and to refresh your memory is often helpful. I put a lot of work in the help system and the infrastructure and I can only recommend to use it.

[Carlo Didier]

I didn't use the search for duplicates because images already in the DB may be processed or just have their metadata changed, so a checksum wouldn't be the same and therefore the images wouldn't be found.