Image Misalignment Issue in Pix4Dmatic

PIX4Dmatic PIX4Dmatic Questions/Troubleshooting
1

Hi, dear community members :wave:

I hope you can help a newcomer.

I am writing to seek assistance regarding an issue I am encountering with the alignment of images in Pix4Dmatic. Our photogrammetry project involves the use of a viDoc RTK rover and an RTK drone (Mavic 3 Enterprise RTK), and we recently captured a site using these devices.

The workflow includes the use of ground control points (GCPs) for accurate georeferencing. The coordinates of the GCPs were obtained using the viDoc rover and are in the NAD83 (2011) coordinate system (EPSG: 6318) for horizontal positioning, with vertical positioning based on ellipsoidal height and no geoid transformation applied.

However, when I upload the images captured with the viDoc rover into Pix4Dmatic, they are positioned incorrectly compared to the GCPs and the images from the drone. The coordinates in the metadata of the misaligned images show significantly different values compared to the correctly positioned images.

For example, the latitude and longitude values of one of the misaligned images are 9.910808333333334e-05 (latitude) and -2.817591666666667e-05 (longitude), whereas those of one of the properly positioned images are 40.739088583333334 (latitude), -73.84138852777777 (longitude).

Unfortunately, I do not have information on the coordinate reference system (CRS) of the misaligned images, which complicates the troubleshooting process.

I would greatly appreciate any guidance or assistance you can provide to resolve this issue. Specifically, I am seeking insights on how to ensure proper alignment of images captured with the viDoc rover in Pix4Dmatic, considering the known CRS of the GCPs.

Any input is highly appreciated.

Best regards,
Alek Alergant

2

Hi @alek,

Thank you for your message and the description of the issue.

If I understand correctly, if you process in PIX4Dmatic only the DJI Mavic 3 Enterprise RTK images, the result is located at the correct location and the coordinate system is set correctly. Do you confirm?

For the images taken with PIX4Dcatch RTK, the geotags are in the Coordinate Reference System used by the NTRIP service you selected to broadcast corrections.
Can you retrieve this information from PIX4Dcatch?

Best regards,
Daniele

3

Hi @Daniele_Lecci,

Thank you for your prompt reply.

Yes, as you wrote, only the imagery taken with the drone seems georeferenced correctly.
The rover take-off seems to be misaligned.

Please see the attached images, log files and a video of screencast from pix4DCatch.
There are two take-offs done on the same day with a 30-minute difference. They are similar and similarly behave in pix4Dmatic.

Appreciate your help and am looking forward to hearing back from you.

Best regards,
Alek Alergant

4

Hi @alek

Thank you for your reply and the files.
I couldn’t find any image to test, but from the screenshots, I can see that the two projects taken on 2024-01-26 are missing the RTK tag (image below), indicating that the geotags contain only standard GNSS accuracy. This will explain the vertical inaccuracy.

image
image723×764 47.6 KB

What if you try to use one of the older projects (RTK), will it be located correctly?

Kind regards,
Daniele

5

Hi @Daniele_Lecci

I appreciate your reply.

Yes, I’ve noticed the lack of the RTK tag too.
Most likely, it’s the key. Can you tell me why that could’ve happened and how I can avoid this situation?

The previous take-offs worked fine for us.

Many thanks for your help, Daniele.

Best regards,
Alek Alergant

6

Hey Alek,
I had a look into your logs and both of them indicate the loss of RTK signal. When scanning a site, please make sure you avoid obstacles such as trees or buildings. I also recommend looking at the RTK signal while capturing, since we do not provide warning sounds when the signal drops.
The GNSS accuracy of your 2 projects is very low, which is expected under these conditions. If really needed, you can process the dataset separately and anchor some high-accuracy GCPs to correct the georeferencing. This will help position the model where it should be.

Hopefully this helps.
Cheers,
Teo

7

Hey Teodora :wave:

Thanks a lot for your expert assessment.
I appreciate that.

Will be watching the signal and obstacles more carefully next time.

Best,
Alek

8

From your experience what would you recommend to do in the situation when the signal disappears in the middle of the take-off?

Many thanks :pray:

Best,
Alek

9

This is tricky, because your RTK signal might be affected by numerous factors, such as atmoshperic conditions, radio or multipath interference. Did you notice a drop in the RTK signal in the exact same place? Can you identify a pattern in the signal strength? Could you compare the viDoc RTK signal with a second GNSS receiver?

Best,
Teo

10

I understand. Thank you for your help Teodora.
We’ll try to watch the signal more carefully and find the pattern.

Best,
Alek