Processing DJI Phantom 4 RTK to Include GCPs

Hi All, I have been using a P4RTK for a while now with mixed accuracies compared to GCPs. I always add a number of GCPs to check against through the processing stage. I have watched a very good video Pix4Dmapper masterclass with James Pick - YouTube and it mentions that you can incorporate your GCPs into the project with the RTK images to better the accuracy.

My question is… does Pix4D then only use the those GCPs for the processing? or if it can use the RTK and the GCPs to strengthen the residuals at what stage do you change the GCP’s from “Check Points” to “3D GCP”?, so both are used?

Thanks, Mark

Hi Mark,

I have seen mixed results from the RTK drones as well, and agree with backing it up with GCPs. With PIX4Dmapper you have the flexibility to select options based on this data to adjust accordingly ensuring the best results even if one method fails.

I strongly recommend using GCPs whenever possible, with good RTK accuracies, they can be used as checkpoints to confirm/measure this operation. Should the checkpoints indicate that there are errors with the RTK accuracies, you have the ability to assign less confidence in the image geolocation, thus giving the software room to move the camera accordingly. In this case, you would then change the checkpoints to GCPs for accurate geolocation.

I suppose you could use both, in this case, PIX4Dmapper would put higher weight on the GCPs. In case there are conflicts, you would need to isolate one method over the other using the above workflow.

Hope this info helps.


Hi Jon,

Thanks for your reply. So, If I find the accuracy from the RTK survey is not so good and I change the GCPs to 3D GCPs I assume these are used over the RTK readings? There is “chatter” saying you only need 1-3 GCP with RTK, depending on flight size… hmmm not sure about that, I still put down a minimum 6 over my flights areas as the errors seem to vary across the know GCPs, and generally use these during the processing stage.

Just wanted to confirm that when the GCPs are changed to 3D these are used/weighted over the RTK coordinates.



Hi Mark,

Yes, start with the GCPs as Checkpoints. In the face of errors outside the acceptable range, then change them to 3D GCPs. The 3D GCPs will be given priority over the RTK geotags.

I have seen situations where the error is so large, that it becomes necessary to loosen the tolerances and select Standard or Low Geolocation Accuracy in the Image Properties Editor.

This gives PIX4Dmapper the room to shift the camera position to where it calculates it should be. Standard gives 5 - 10m of wiggle room, whereas Low allows for 50 - 100m of movement.

There is “chatter” saying you only need 1-3 GCP with RTK

I don’t have enough confidence in most drone RTK to support this statement either. If taking the time to measure 3, seems irresponsible not to add a few more. What happens when RTK fails, and you are left with insufficient GCPs to correct?

I guess it ultimately depends on the importance of the job, and the amount of time given to complete it.

Hi Jon,

Appreciate your comments and assistance