Hello,
I am running a Klau PPK X4S camera with a Dji M200. I was curious why Pix4D recognizes multiple cameras when setting up project and importing images. This happens even though I use only one X4S camera for the missions. I also see these cameras listed in the Pix4D quality report. Why would this happen? Pix4D seems to number them, but they have the same serial # and parameters as they should…just multiple cameras.
Thanks,
Hello Michael,
This is a known issue which has been reported form other clients already.
There is a community topic that you can look at: Multiple camera models
As my colleague Rhea explains, there are many reasons why that happens, "To name a few possibilities
Sometimes, the camera serial number written in the image EXIF tags changes (it happens randomly and we did not find the reason yet) and the software will create several camera models according to that. Or, as when exterior conditions (weather, exposure) vary greatly during the flight. Also, working with a merged project will, of course, cause the software to detect multiple camera models. "
Anyway, if you can share a couple of images which are assigned to different cameras, we can analyze better and tell you what is exactly going on.
Thank you very much.
Thanks Daniel.
I have a link below to a few of the images from the related mission mentioned above in my first post.
https://1drv.ms/f/s!AshJwvSEGv7qiv8c9z3-uGPn-9V5Sw
Please note that Klau recommends using the following camera settings for an Dji X4S:
However, I actually used “auto” mode and “auto focus” for this mission, so maybe that is the problem? The above recommended-Klau settings maintain consistency by setting “Shutter Priority” mode, which could prevent this from happening?
Thanks for your help.
Mike
Hello Mike,
I have downloaded the images and create a project with them but there is only one camera created.
Would you send me photos which have a different camera assigned?
Thank you very much.
Hi Daniel,
I uploaded all the images now to same link above.
I forgot to mention that the X4S camera I used had a laboratory calibration on it by Klau Geomatics. They provided me with the parameters to enter into Pix4D. I edited the X4S camera model that was default to Pix4D for this. Below is the camera parameters I used in case you need it.
Thanks for your help.
Mike
Hello again Mike,
I downloaded all of your images and indeed there are four cameras detected when importing the images.
I then looked at the EXIF/XMP tags embedded in the JPG files and what happens is that the EXIF tag “Exif.Photo.BodySerialNumber” has four different values. Most of the images have “e962248556df3bd0f1e6a8b3c98b995c” and they have been detected as camera (1) but there are other images which have “11334300000000000000000000000000”, “2016041101” and “2016041101df3bd0f1e6a8b3c98b995c”. That is the reason why four cameras are detected.
The shutter speed value is also different so that is probably the reason why there are 4 different serial numbers.
I hope this answer helps you.
Thank you very much.
Thanks for your investigating this Daniel!
Other than having these extra cameras show in the Quality Report, is this something that could degrade how Pix4D processes the images thereby affecting the accuracy of the project?
Thanks again!
Mike
Hello Michael,
What happens when there are multiple cameras created is that all of them can be optimized differently.
For example, if there were different flights within the same project, meaning that the drone landed several times, I would say that there should be different cameras for each flight even although the same drone and camera are used. That is because the internal camera parameters can be different for each flight and therefore, they should be optimized separately.
When you have more than one camera within the same flight, the camera should be the same as it does not change. If you have two or more, it means that they will be optimized independently and that should not be the case.
Thank you very much.
Thanks Daniel,
This makes sense. In this mission I had 3 discrete flights, so 3 of the 4 cameras makes sense.
In Pix4D Processing Step 1, Klau Geomatics recommends setting the camera calibration method \ Internal Optimization to All Prior because I have a laboratory-calibrated X4S camera with defined internal camera parameters (please see 5th post above).
As I understand it, this means that minor internal camera optimization is occurring during step 1 (i.e., Pix4D is keeping internal parameters optimization values as close as possible to what I manually defined for my lab-calibrated camera). Can you please confirm this is the case? If so, then having multiple cameras show up in my case, should not be a concern.
Thanks!
Mike
Hi again,
You are right about “All prior”, it forces the software to not change much the internal parameters during Step1 and in case you have a lab calibration, I would recommend it.
Thank you very much.
Today I did a job using our new system and the software identified three different cameras even if the souce of images was always the same.
What could be the reason? Should I repeat the flight?
Images was taken using DJI GSPro and Matrice 210/X4S
Below the link to the quality report
https://www.dropbox.com/s/jf0xhrlly3m7z6d/1126_001_20171018_Ticino_report.pdf?dl=0
Couple things, Pix4D responded to another post saying they had a bug making multiple cameras but shouldn’t affect the results. Is there something wrong with your results? The camera calibration is way off so that is a concern.
In the GSPro, rotate your flight to be parallel with the long side.
I would never run Pix4D on a Xeon computer or use anything but a GeForce video card…at least running the Desktop flavor. Different results come from different computer architecture…it is a pretty simple project if you want me to test it on my Core i9 machine. Adam.Jordan@nhiae.com
Hi Massimiliano,
It’s difficult to tell, could you share the images of your dataset? I’d like to have a closer look at the project.
Have you made a single flight or were there more than one?
Hi pierangelo,
thank you for the support.
Could you please send an email to massimiliano.talento@comal.ch so I can send you a link to download the project. Of course, results of your analysis could be shared in the community.
Hi Massimiliano,
I’ve sent you an email 
Ok, I’ll post my findings here.