We are trying to get the limelight 2 working and haven’t had much luck after getting a clean threshed image. Here is the image we get: https://imgur.com/a/eDo6Pbq
When trying to calculate the angle the robot needs to finish at in relation to the target the values for the raw contour are inverted from what we need. For example, if we set the sort mode to left most the raw contours get returned from right to left. On the other hand, if we set the sort mode to right most the raw contours get returned from left to right. This is also the same issue with using highest and lowest sort modes. This was a screenshot when the limelight was in sort mode: leftmost - https://imgur.com/a/2vV8ZuQ is it possible to enable more than 3 values? Currently we only get tx0, tx1, tx2 from network tables and if we were able to get a tx3 it would solve most of the issues we are having.
Another issue that we have ran into was that some of the new features for the limelight are not in the limelight docs. The main items that were not in the docs that were hard to interpret the meaning was: Contour simplification % and W/H Ratio. From our testing the W/H Ratio seems to be backward thinking than explained in the docs.
Is anyone else having these issues?
Devs at Limelight: Is it possible to get the pipelines that you guys used on this year’s field elements?
the contour sorting modes are for telling the limelight which target to lock onto. I tested all of them again this afternoon and they all seem to be working well. If I select leftmost, then the limelight locks onto the contour which is furthest to the left in the image, etc. We usually use “closest”.
For this year’s field elements the easiest way we’ve found is to use target grouping - dual target and intersection filter - above and sorting mode - closest.
the w/h ratio seems to be working as well, if you set w/h to be 0.0 - 1.0, it only accepts targets which are taller than they are wide and vice versa for 1.0 - 20.0. Does this make sense?
teamVision, you should be able to prevent the behavior you are seeing with the right combination of filtering and target grouping. Can you try the ones I mention above.
Ok I think I see what you guys are struggling with. It doesn’t look like we’re sorting the tx0,tx1,tx2 values correctly. We can look into that for the next update. However I think you will have good luck if you use the sorting and filtering methods I mention above.
The w/h ratio is applied to individual contours. You are using a target grouping of two (Dual Target) and it looks like both contours are within the w/h ratio.
Thanks for the explanation @Hjelstrom. Do you know if there will be an update to have the W/H ratio apply to the dual target? Or better yet, have a W/H discriminator for the individual targets, and a second W/H filter for the dual target? Those ultra long targets are giving us a lot of grief.
