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
If we could record user position, and camera position and orientation this would be very good to have for automating experiments, for example the messy latency experiments.
We could then do one session live where we also record the raw camera streams plus the camera trajectories for each user.
With these datasets we could repeat the exact same sequence automatically, but then with different protocols, codecs, settings, etc.
The most important right now is the MMSYS use case, but let's not forget that this functionality is more widely usable. For example to create trajectories for an automatic camera to record a session.
If we could record user position, and camera position and orientation this would be very good to have for automating experiments, for example the messy latency experiments.
We could then do one session live where we also record the raw camera streams plus the camera trajectories for each user.
With these datasets we could repeat the exact same sequence automatically, but then with different protocols, codecs, settings, etc.
Related to #181
The text was updated successfully, but these errors were encountered: