Hi all,
I’m looking to do some fairly involved photogrammetry with my P4P and looking through the technical specifications I have quite a few unanswered questions.
The one big one that might answer a lot of the others is - what format is the (meta)data stream in the video output? There are two muxed streams and one seems to contain data that might be recording stuff such as gimbal angle, GPS coordinates etc. but I found no way of parsing it. I came across a whitepaper (here https://developer.dji.com/onboard-sdk/documentation/introduction/announcement.html) describing some kind of protobuf which seemed promising but whatever I do I can’t seem to be able to parse it into the structure described. (I found DvtmPayload but the whitepaper mentions DvtmProductName (assuming something like DvtmPhatnom4Pro) which I can’t locate?)
Giving up on the data stream I found metadata using exiftool which seems to specify a lot of the parameters I would need, but still leaves questions open:
what height is provided in the “GPS Coordinates” field? I got “39.9m above sea level” but that’s impossible as the ground is about 210m ASL where my video was taken. Either there is a huge error margin rendering the data unusable, or this is somehow above the takeoff point and not sea level? There is a thread here about this ( https://forum.dji.com/thread-148221-1-1.html) but I couldn’t find any definitive conclusions. with all the pitch, yaw and roll angles provided, how do they relate? Is drone yaw relative to north, and if so - magnetic north or geographic north? And do I have to add drone yaw to gimbal yaw? Or is gimbal yaw also relative to (geo/mag) north? Is pitch relative to the horizontal or vertical axis of the drone? How about sensor use? Say I’m running at a lower resolution than maximum, does it use the entire sensor and only subsample some of the pixels? Or does it use every pixel but only from a part of the sensor, effectively decreasing sensor size? What kind of error should I expect from these numbers? What are the “(err)” values representing in the exif? They are exactly the same as the values for pitch/yaw/roll without the “(err)” suffix, so they don’t seem very useful. Are there any estimates on this?
|