RTK Processing

I was just doing more testing with the Matrice 300 and processing data in DD and Pix4d and found this at one site. Using no GCP, and Emlid RS2 as my base casting to the M300 for corrections. Processed the data in DD check my again my point from the RS 2 and elevation is off by about 4 feet. Processed the same images in Pix4d and elevation is still off but only by .8 less than a foot. I didn’t need accurate elevation for this just using for documentation as is untouched.
I’m curious to know if this is because Pix4d actually support the P1 camera and DD doesn’t or if it’s something in the way they process the data.
Anyone got any insight on this?

1 Like

What’s up Joe?! You definitely need to work with them directly and have them do some reprocessing. We had similar problems with the Yuneec RTK at first. Once they got the system to recognize the hardware correctly everything started dialing in. I bet the camera profiles have something to do with it. Especially if it is a consistent offset.

Hey Michael. I figured as much. Looking at the processing report they actually show the camera as the ZenmuseP1. I’ll open a ticket with them.
I also flew a smaller site same setup, didn’t define my GCP in Pix and compared elevation to a benchmark our survey guys put down…was within a 1/10. I was pretty impressed with that. I still put GCP’s down for everything I fly requiring close elevation…that’s only cause I don’t trust my RS2 setup entirely yet but I’m getting there, someday you’ll see :slight_smile:


The biggest drawback of DD for me is the $49/map for GCPs. So, do you think that using P4PRTK or M300RTK, with PPK’d images could help side-step that hurdle?

1 Like

@txdroneco! nice to see you around. Cold much today? Lol. Yes, RTK/PPK will alleviate the need for GCP’s but understand that you will still need to be able to collect checkpoints and most likely a way to align data after the fact. Especially in construction as there are all pretty much local coordinate systems. To just make a map that’s not related to any other data you probably won’t have to do anything.

Hi, older topic but I have had some issues using the M300 RTK tagged images using the DJI base.

Todays data was distorted (stretched about 2m over 100m) compared to the previous map (checked by design overlays) flown with DJI pilot indicating a good connection to the base.

I have not checked what the model looks like using pix4d.

I corrected the data by picking control points from the previous map. This is of course very bad practice but it was the only solution this time.

After 4 years of using PPK and more recently RTK tagged images, I found that the best way of having reliable data over time is using the same 4 - 5 GCP’s and couple of check points.

PPK only especially in larger sites datasets seem to be still subject to variance so I definitely can’t rely on that for important survey type work (checking work against design)

Equipment M200 & x4s with geocue Loki PPK gps. RTK m300 & p1 and dji base or network RTK when available.

Also DDs automatic alignment feature on previously GCPd maps can be a bit unreliable. Although It’s handy for quick rough jobs.

1 Like