I am having trouble with the accuracy of my models, is there any way to increase this? i set GCPs all around each site and input the coordinates into Pix4D, yet when i Tagg the locations the Blue marker indicates it is not where i have physiccally surveyed, most of my sites are getting between 50mm - 100mm error is there any way i can improve this? please see photo my marker is clear but yet the location (blue marker) says otherwise.
Thank you
Hi Max,
I’ve had the same issue a couple of months ago and left a remark on the forum but there hasn’t been any response.
(see one of the previous posts).
We did a test project with RTK positioned camera in Belgium and in the Netherlands.
Turns out that the projection parameters for those countries used in Pix4d are only approximate (not cm correct for the entire country).
So we had to shift our data 15cm to 20cm manually after processing
My advise would be to check if the logged Latitude and Longitude tagged to the pictures is correctly projected into your local grid (Initial position) by Pix4d.
G,
Thomas
Thomas,
Thanks for the reply, the cameras are in the correct location, the only way around it I can see is to change the GCP coordinates to the ones on pix4d (the blue circle), but will that make them correct even though it shows minimum RMS errors?
Max
Hi Max,
I saw on your screenshot that you have only marked the GCP in 2 images.
Normally if you mark the GCP in more images, the blue dot will be more accurately calculated.
Select the GCP, mark in more images, click apply and then reoptimise.
G,
Thomas
There are more GCPs with 19 images with the same result, I have adjusted all to computed position and the accuracy is 20mm and lower! Seems to have done the trick,
Pix4d does not take into account global positioning,(movement of the satellites) I’m hoping this will work, I will check against and RTK survey and see the results.
Thanks for your input Tom (£250 a month must not be suffice to get a response from pix4d!!!)
Dear all,
The blue circle with the dot in the middle indicates the estimated GCPs’ position in one particular image. Depending on the quality of the initial reconstruction this blue dot can be further away from the real location of the GCP. However, when the GCP is manually marked on the images, the marks are used to estimate a new 3D position, by reoptimizing the project. Thus, the marked positions are the ones that are in the end taken into consideration.
Please note that indeed it is important to mark the GCPs on as many images as possible and to make always sure that the same point is selected on the images.
Regards,