Multiple Camera Models

 Why does pix4d display multiple camera models from one set of images from one camera?

1 Like

Hi Brian,

The software pulls the camera model information from the image exif tags. In most of the cases, it then indicates with an icon if it could match it to a known camera model in the database of Pix4D (which is what happened in your case) or of the user. For more information about the camera model and associated icon, please click here.

In your case, the image exif tags indicate you used the FC6510_8.8_5472x3078 (RGB) camera model to capture you dataset. This camera model exists in the database of our software, and this is indicated by the   icon.

For one dataset, there are many reasons why the software would detect several camera models, but it is difficult to say without having your dataset. Could you please send us two images of each group so that we can investigate?

In any case, we do not expect this to negatively influence the reconstruction of your project.

Looking forward to hearing from you!

Thanks for the response.

This forum will not allow me to upload images larger than 2mb.

Could you list some of the reasons it would detect multiple camera models? There doesn’t seem to be a pattern to the software detecting multiple cameras. Sometimes it happens, sometimes it doesn’t. sometimes it detects 1 camera, sometimes 2 cameras, sometimes 3 cameras. 

 

It is usually exposure…bright sun to clouds will do it and a big reason I shoot in RAW only now.

If there is an issue with your point cloud then you will likely need to use multiple sub-projects and merge…otherwise if it produces “1 block” in the point cloud then don’t worry about it as everything worked out fine internally to Pix4Dmapper.

Adam,

That’s an interesting take. We have been having issues with exposure. What’s a typical size project that you work on? We have some projects that have over 2,000 photos. Reading up on the RAW format, not sure how well that would work for me…

My current project has over 11,000 RAW 42MP pictures that is over 450GB…nothing small here :slight_smile:

I wouldn’t bother with DJI DNG as they don’t have the image sensors to make edits worthwhile in RAW.

What camera are you using?

Hey Brian,

You need to use a file sharing service to share your images (Dropbox, Google Drive, WeTransfer, …) and post the link. As mentioned, please include 2 images of each group for starters.

To name a few possibilities for the reason why the software would detect multiple camera models:
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 @Adam mentioned, 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. However, I would prefer not give a definitive answer until I can have a look at your pictures :slight_smile:

Thanks for the reply. Here is a link to the photos.

https://www.dropbox.com/sh/w2annytl17km76c/AAA3U-dNfbhFlFYTpJ1IBOw4a?dl=0

 

Hi Brian,

Thanks for providing the link.

After reviewing your pictures, I can confirm that the reason why our software is detecting several camera models is because there are different camera serial number tags in the image Exif data. Other users are in the same situation and we are not sure as to why this is happening. The camera manufacturer, being DJI in your case, is responsible for the way these tags are written, so there is nothing we can do about it.

As shown in your screenshot, the software detected three different camera models. You sent me two images from each group and indeed, your dataset was split into three groups because there are three different serial numbers in the Exif data of the images. See the screenshot below, with group1 in red, group2 in green and group3 in blue.

As explained in my first reply, we do not expect this phenomenon to negatively influence the reconstruction of your project.

I hope this helps!

I’d like to add that if users deal with mapping cameras and the imagery includes world files or GeoTIFF Tags in the imagery, this also confuses Pix4D and causes it to create multiple cameras.

I just wanted to add this in case it helps others as it caused us a lot of confusion.

 

I still have this problem with an Inspire 2 + X5S. In my case it definitely is the serial number. In any given flight I’ll have maybe 5-10 (out of several hundred) photos recorded with a different serial number.

======== DJI_0635.JPG
Serial Number : 67b7f95cc986543a148cc8e74586fa42
======== DJI_0636.JPG
Serial Number : 67b7f95cc986543a148cc8e74586fa42
======== DJI_0637.JPG
Serial Number : 67b7f95cc986543a148cc8e74586fa42
======== DJI_0638.JPG
Serial Number : 67b7f95cc986543a148cc8e74586fa42
======== DJI_0639.JPG
Serial Number : 67b7f95cc986543a148cc8e74586fa42
======== DJI_0640.JPG
Serial Number : 201604110186543a148cc8e74586fa42

I use exiftool to read the tags from my whole set of images to identify the non-conforming ones:

exiftool -SerialNumber *.jpg > serialnumbers.txt

Then you can manually fix each one by re-writing that tag.

exiftool -SerialNumber=67b7f95cc986543a148cc8e74586fa42 DJI_0640.jpg

exiftool will backup the edited files by default in case you screw it up. If you have a bunch of images you could just move the ones that need to be fixed into their own folder and change all of them at once.

Now at least Pix4D will see all the images as being from the same camera and it will produce a much more sensible quality report.

I reported this issue to DJI and they didn’t seem particularly concerned about resolving it so I wouldn’t hold your breath.

Hi @dboge, thanks for sharing the workaround.

Another option is to assign the same camera model to the images inside the Pix4D’s Image Properties Editor:

  • Order the images by Camera Model.
  • Assign all the images to the same Camera Model.

This way you will be able to process with only one model and also only one model will show up in the Quality Report.

Best,

:joy: Well that’s a lot easier… thanks!