Some useful tools for visual tournaments

Dormant threads from the high school sections are preserved here.
Locked
User avatar
Mike Bentley
Sin
Posts: 6461
Joined: Fri Mar 31, 2006 11:03 pm
Location: Bellevue, WA
Contact:

Some useful tools for visual tournaments

Post by Mike Bentley »

One common problem in writing visual tournaments such as Eyes That Do Not See is that you can only find a low resolution image. Your first step should be checking something like artstor or a Google Image / Books search to find a higher resolution image. But failing that, there are some new AI tools that can potentially help.

Super resolution services include big.jpg which lets you upres 4x for free: https://bigjpg.com/. Another one I haven't tried myself is Let's Enhance: https://letsenhance.io/signup. If you have a spare GPU and some machine learning knowledge you could also run such a model on your own "for free".

Another common problem is wanting to crop out content. You can crudely do this by drawing boxes. A more sophisticated approach is using content-aware fill. There's a new service at https://cleanup.pictures/ that offers this for free in a web app. Quality varies a bit but in general it looks better than most manual removals of figures / signatures / etc.

These AI-powered techniques are going to introduce elements that aren't "actually there" into the image. But, of course, you're also dealing with reproductions in the first place which don't accurately record everything about an artwork. I think there are probably cases where using techniques like this would allow for better visual questions.
Mike Bentley
Treasurer, Partnership for Academic Competition Excellence
Adviser, Quizbowl Team at University of Washington
University of Maryland, Class of 2008
Locked