Current tool version is using a Ken Burns implementation, but it doesn't always produce the best results (i.e., sometimes large images are very pixelated on zoom-in, irrelevant parts of the image are zoomed into, head cutoff problem on people/animals, etc.)

1. Investigate options for different ways to visualize a somewhat unpredictable image stream (some ideas from Design are here: https://www.figma.com/proto/o7idnBrc4TOqWGab88zEh4/Sprinthackular-Group-D?node-id=273-12719&node-type=FRAME&t=zyneyW5Q5lnqjiXD-1&scaling=scale-down&content-scaling=fixed&page-id=270%3A7987&starting-point-node-id=270%3A12709&show-proto-sidebar=1)
2. Generate video previews for the highlighted enwiki DYKs here (https://docs.google.com/spreadsheets/d/1_6R0z4MQqWTWZL3jZYGV_SEUuwzukbewlhuTf39Khss/edit?gid=616783496#gid=616783496) with the 1-2 most promising alternatives to Ken Burns and take a first pass at assessment to see whether this is better or worse than Ken Burns.
3. Pass off to @Maryana for final assessment

Goal:
Aiming for ~75% of topics to produce acceptable video