You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
A couple years ago, we ran an updated set of regressions to approximate the distance of a label from the camera, taking into account factors such as zoom level and the height of the panorama. This hasn't been incorporated into predict_crop_size() yet! I imagine that it would marginally improve the sizes of the crops, and it would also create more consistency across Project Sidewalk tools.
Even if we don't do that, we should at least update it to account for different panorama sizes. I updated this for a partner we're working with yesterday, so I'm pasting it below. Just did some simple algebra to figure out what the function would look like when we aren't assuming that panos have a height of 6656 pixels!
// Old version
distance = max(0, 19.80546390 + 0.01523952 * old_pano_y)
// New version
distance = max(70.52258646 - 101.43424512 * (pano_y / pano_height), 0)
// New version using pitch instead of pano_y if we want
distance = max(70.52258646 - 101.43424512 * ((90 - pitch) / 180), 0)
A couple years ago, we ran an updated set of regressions to approximate the distance of a label from the camera, taking into account factors such as zoom level and the height of the panorama. This hasn't been incorporated into
predict_crop_size()
yet! I imagine that it would marginally improve the sizes of the crops, and it would also create more consistency across Project Sidewalk tools.The code is in the following file in Project Sidewalk, with other helpful links at the line directed to from the link:
https://github.com/ProjectSidewalk/SidewalkWebpage/blob/develop/public/javascripts/SVLabel/src/SVLabel/label/Label.js#L13
The text was updated successfully, but these errors were encountered: