10. MediaShow 5: Criticisms, Cont'd

At the top of the Pending work area are two buttons: Apply Default and Skip All. Say the four groups shown on-screen contain 100% accurate matches of Bob, and you’ve done enough tagging already that MediaShow puts a blue outline around the Bob name button above each group. Click Apply Default and MediaShow will apply the Bob tag to the faces in all four groups at once. In theory, this is a great shortcut that also highlights the need for MediaShow to have a more flexible UI for making more information visible at once. In reality, I never had an instance where I had enough accurate hits in enough visible groups to make this feature practical. The Skip All button works similarly, allowing you to defer all visible groups.

The trouble with these two buttons, apart from accuracy issues, is that you’re stuck with however CyberLink chooses to organize its returns. Check out this image:

You can see that MediaShow displays four association groups. In the Pending bar running down the right side of the UI, we see the first face shown in each group. The sixth face shown in this bar is the same person as shown in the third and fourth image. I should be able to lump that sixth group in with the third and fourth, but I can’t. Similarly, I should be able to batch select the first and second groups and click something either to Skip or Remove remove these two groups together, but there’s no way to do this. I have to Skip or Remove each group individually. That doesn’t sound like much of a difference, but there’s a two to three second lag between clicking one of these buttons and being able to proceed. When you deal with many dozens of groups in succession, this lag between each action becomes really annoying. A more sensible approach might have been being able to expand that Pending bar and then having the ability to drag group thumbnails into super-groups of your own arranging. With this, suddenly those Apply Default and Skip All buttons would have a lot more power.

My last major complaint centers on the Regroup button at the top-right of the UI. The idea of this button is that as you go along and occasionally uncheck faces from groups, you can end up with a very large collection of unassigned faces. Presumably, MediaShow keeps learning your faces more accurately as you continue tagging. Pressing Regroup performs a second analysis that should refresh your groups and provide for more accurate associations. This was not the case for me. Rarely did I get groups of any larger size than I had before, and, as you can see in the following image, accuracy seemed little helped. Two out of 16 is a real head-scratcher.