If you missed our last update post, take a look at 2020.1:
2020.2 pushes Limelight’s hardware even further by adding the ability to zoom and pan simultaneously. It also adds a brand new video mode and important bug fixes. Let us know what features you would like to see in future updates!
3x Hardware Zoom at 60 FPS
Our last update added 2x Hardware Zoom for all Limelight models at no cost whatsoever. We’ve managed to push our hardware zoom even further with 3x Hardware Zoom. All Limelight models are now capable of 3x Hardware Zoom at 60fps. This makes full-court tracking even easier for every team.
The following gif shows 1x, 2x, and 3x Hardware Zoom from the sector line (full court shot line):
Robots capable of shooting from both long and short distances in 2020 would have found it difficult to use Hardware Zoom on their Limelights without the use of an active, adjustable Limelight mount. We have incorporated Hardware Panning to solve this problem.
This gif shows Hardware Panning while using 3x Hardware Zoom. This should simplify mounting for teams who wish to use Hardware Zoom:
Currently, it doesn’t look like the zoom level or zoom region has it’s own NetworkTables entry for on the fly changes. It’s a pipeline setting though, and you can change those on the fly. This would probably be somewhat clunky for really complex zoom handling, but if you’re just doing 2 zoom levels for increasing precision, it’s probably fine.
I know you had planned the zoom/pan features to be configured using the profiles.
However, as the zoom/pan features evolved into the fantastic features we have now, I feel it’d definitely be advantageous to have them as NT variables instead of having to make different profiles (specifically for panning).
How does panning affect tx and ty? I found that zooming doesn’t change tx and ty values, which is great. Panning Y by 1.0 results in a ty change of 12 degrees, but Panning by 0.5 resulted in a ty change of a bit more than 6 degrees. Is the change supposed to be linear? And also is there any documentation on how accurately tx and ty reflect actual target angles? Thanks!
Want to second the networks table capability. Doing some testing yesterday we found it more beneficial to stay in 1x zoom because of the increased FOV and when we tried to add smart zoom logic (zoom once tx is in some threshold) the pipeline switching was too slow to do on the fly