Hi, just wondering if anyone has any experience or understanding of any likely benefits of using the DJI Phantom 4 RTK version versus the standard Phantom 4 Pro V2 when taking photos and then using Reality Capture?
There is a big price difference it seems between the two, but does it really make that big a difference with the accuracy of the end model made - say for modelling something at the scale of a building?
Interested to know your thoughts thanks!
Hi Anthony,
I guess you are referring to different sensor resolution? Or are there other differences?
Could you specify a bit more?
In any case it will be difficult to predict possible benefits of one camera over another, since there are so many variables involved. In my experience the camera plays only a part in the quality of the end-result. Less resolution can be made up for by more images, if needed…
Hi Götz,
Actually both drones use the exact same camera hardware (20MP). What makes the RTK version about 4X more expensive is that it can tag each photo taken with GPS with an accuracy down to within a cm or so in terms of vertical and horizontal location in the sky …apparently. Normally GPS with a device like a phone is accurate within a few metres, so this big leap in accuracy is what you are really paying for (survey grade accuracy). The question then is if Reality capture can make use of this exact location of each photo taken and how much better if at all the end result models are.
The hard way to find this out is to purchase the Phantom RTK drone and try it with Reality capture, but its an expensive experiment! 
Does anyone at Reality Capture know if such accurate location metadata associated with each photo is taken into account by the software? This could be a first question to ask. Thanks.
Hi Anthony,
ah, that’s much clearer now.
I have no experience whatsoever using GPS but I know that there are sometimes problems caused by imprecise GPS data, so I’d say it will definitely benefit. As to how much this will pay off for your personal workflow is quite another question. I think that ultimately the GPS data is only a first step for RC and might just speed up the processing time a little bit. The final result is probably largely independent from GPS data and the last bit of precision for aerial photography is ultimately derived from ground control points, the coordinates of which you have to aquire separately.
Hi Anthony,
I am working on a project using RTK/PPK GNSS solution of UAV camera shots for georeferencing the 3D model of a large landscape using ONLY the precise flightlog data. My Camera positions are accurate 15-30 cm. I have GCPs to countercheck the accuracy, at the moment I am at 40-50cm precision, but hoping to improve this result. High precision camera positions are much better for georeferencing via flightlog than normal GPS, but so far RC has not proven to me that camera and lens parameter calibration can make sure to get the full potential out of this. At the moment flightlogs dont seem to replace GCPs!
However I am working on it and hopefully we can improve accuracy together with RC.
All the Best
All the Best
Hi ,
Appreciate the information. May I ask what hardware setup are you using? Thanks.
Hi Anthony,
I use:
Emlid reach M+ (it has a ublox MT6 inside) as PPK receiver (I dont do real-time) on a Sony Rx1R2 with hotshoe adapter flying with DeltaQuad Pro (VerticalTechnologies)
Your Welcome
Hi ,
Wow the vertical take off plane looks awesome, all the best and good luck with the mapping!
Gents, I just upgraded my Inspire 2 + X4S with a Topodrone PPK system. I was wondering if between you you have figured out the best processing settings. The Topodrone software writes accurate coordinates to the image EXIF. So in the first instance I would like to process to the camera prior positioning and compare with the GCP’s.
Thanks