Removing Sensors from Studio recordings

Due to resolution limits for certain video codecs, packing recordings from many sensors into one Multiperspective Combined per Pixel video often results in a small area allotted for each sensor. The more sensors you capture with, the more this is apparent.

The most effective way to prevent these dense CPP's is to be selective about the number of sensors you use to record, but if you have already recorded, and want to remove certain sensors from your export to give each sensor more area and resolution in your CPP, you can modify the project JSON to do so.


Make a backup before editing your project file

Hand-editing the project JSON is a precarious task which can corrupt your project, so always make an unedited backup to restore if necessary.

  1. In Windows Explorer, navigate to your Depthkit project bin, and open the original dkproject.json in a text editor.
  2. Search for the "recordings": {...} object
  3. Within the recordings object are objects for each take - e.g. "TAKE_05_17_13_46_00": {...}
  4. Within each take object is a "compositions": {...} object.
  5. Within the compositions object are objects for each sensor - e.g. "Azure Kinect <serial number>": {...}
  6. Delete the entire sensor object, including the contents of its curly braces (and the comma after, if applicable), for each sensor you'd like to remove.
    Though this removes the sensors from the project file, your recordings will remain intact offline from the project. To reconnect them, paste the sensor object back into the appropriate place in the project JSON.
  7. Save the JSON.
  8. Open the project in Depthkit, and confirm that the sensors are removed in the Edit context's Take List.
  9. Export your capture as normal. The exported video or image sequence will only contain the remaining sensors, and the metadata file will not require any modification.