The documents each of the coordinate types, establishing the nature of their axes including the respective origins and positive directions.
This is based on local testing with a two cameras, one for which I do not have a valid calibration (though I did find one with the same resolution in sr-robot3) and the confirmed using a C270.
There is definitely some oddness here as the results suggest that positive Y axis values are "down" in the camera's field of view.
Validation of this with another camera would be most welcome.
I've built this on top of #304 in part to avoid conflicts, but also because that PR establishes much more clearly that the types have intended semantics rather than allowing them to be essentially shallow wrappers around whatever data happens to fit their shape.
These changes will inevitably conflict with #302, however I suggest we aim to merge this PR first (as it doesn't change behaviour) and then come back to that one. To that end I've ended up copying over some of the relevant changes in where some of the logic lies as well as the tests, but all keeping the behaviour as it is in master.
The documents each of the coordinate types, establishing the nature of their axes including the respective origins and positive directions.
This is based on local testing with a two cameras, one for which I do not have a valid calibration (though I did find one with the same resolution in
sr-robot3
) and the confirmed using a C270.There is definitely some oddness here as the results suggest that positive Y axis values are "down" in the camera's field of view. Validation of this with another camera would be most welcome.
I've built this on top of #304 in part to avoid conflicts, but also because that PR establishes much more clearly that the types have intended semantics rather than allowing them to be essentially shallow wrappers around whatever data happens to fit their shape.
These changes will inevitably conflict with #302, however I suggest we aim to merge this PR first (as it doesn't change behaviour) and then come back to that one. To that end I've ended up copying over some of the relevant changes in where some of the logic lies as well as the tests, but all keeping the behaviour as it is in
master
.