# Calculating "Camera to facade distance"

I have taken a series of overlapping oblique images of a building facade with an iPhone camera (with geotags) and generated a point cloud and orthomosaic in Pix4D. When I import the point cloud and orthomosaic into eCognition, it shows that the point at the ground adjacent part of the facade has a lower Z value and the point of the upper part has a higher Z value but I need “camera to facade distance” as Z value to help extract cracks on the facade using OBIA in eCognition.

Later I generated point cloud and mosaic by deleting geolocation information and selected “Arbitrary”
(output) coordinates in Pix4D. Still, the Z value is the height above the ellipsoid (according to WGS 1984), not the distance from camera position to facade. What can I do?!

Thanks a lot for your feedback. Your feature request has been forwarded to the product team.

Best,

Hi Nikoleta,

Thank you for your reply and the action taken. I am eagerly waiting for the ultimate response from the product team.

Best regards,

Pix4Dmapper will not export “camera to facade distance” as Z value by default.

One thing you could do is delete the image geolocation and add an orientation constraint:

Hi Christina,

Thank you for clarifying your position with regard to the problem and also for the idea to find the solution.

I am going to experiment with your idea.

Best regards,
Selim

1 Like

Hi Christina,

I am going to experiment with your idea.

By changing orientation while taking orientation constraints, I got the “Camera to facade distance” which varies object to object (object depths) on facade. But it shows in negative numbers.

Is there anything to do to get the distance in positive numbers?

I look forward to hearing from you.

Best regards,
Selim

Hello Selim,

One thing you could try is to add some arbitrary GCPs that will give an arbitrary coordinate system to your facade. For example, you could add a GCP at the lower left part of your facade with coordinates 0,0,0.

Similarly, you should add more GCPs so that the numbers come out positive.
I believe the numbers came out negative because you removed the image geolocation and this lead to an arbitrary coordinate system.

I hope it helps.
Christina

Hi Christina,

Thanks for the nice idea!

Yes, you are right. I have removed the image geolocation and taken orientation and scale constraints.

I understand that I can add the coordinates 0,0,0 at the lower left part of my facade as a GCP.

If I add another GCP at the lower right corner, the coordinates should not be 0,0,0 I guess. So, how to find out the coordinates of that point?

One thing I can do: after inserting 0,0,0 coordinates at the lower left corner, I can check on the facade image what coordinates at the lower right corner come up.
Then, I can add that as its GCP. Does that make sense?

I look forward to hearing from you.

Best regards,
Selim

Hi Selim,

I was thinking something like this:

GCP1 (0,0,0)
GCP2 (X,0,0)
Where X is the real distance between point 1 and 2 in one direction
GCP3 (X,Y,Z): a point in the direction of the camera and the facade in order to define the Z direction

I hope it makes sense
Given that this is an application that is not officially supported by Pix4Dmapper, unfortunately, I cannot give further assistance.

Currently, the software does not export “camera to facade distance” as Z value by default.
But you can give it a try with arbitrary GCPs. I’ve seen similar projects in the past from other customers.

Christina

Hi Christina,

Really interesting! Thanks a lot. I am going to experiment it today. I will let you know the results.

Best regards,
Selim

1 Like

Hi Christina,

Thanks for your lovely idea. It really worked in my case.

I am facing a new problem. This is the point cloud layer with scale and orientation constraints added.

I see some problems with it. They are-

• point density is not the same throughout the whole facade
• some noise

First, I cleaned the noise and clipped the point layer as below-

Then, I run the third process (ortho mosaic & DSM).

Results
Got the following ortho mosaic and DSM

It is notable that the ortho mosaic and DSM are not generated according to my point cloud clip.

Best regards,
Selim

@m.s.hossain-2, thank you for the screenshots.

Instead of using the Clipping tool (1), you should use the Processing Area (2) to limit the point cloud.

I hope this helps,

Thanks a lot. I think this time it will work. Regards, Selim

Hello Christina,

This email finds you well. Thanks for your continued cooperation.

I have some problems generating good reconstruction in Pix4D. I need your help soon.

You know that I am working with building facades (oblique photogrammetry). I capture images with an iPhone camera.

I see it is very difficult to generate a good reconstruction. What is the problem I don’t know? I take images very carefully with much overlap. After the acquisition, I check the images on the laptop one by one and delete the blurry images. Despite that, I do not get a good reconstruction of the facade. It is giving me much pain killing my valuable time. I am attaching some of my experiment results for your reference.

I look forward to hearing from you soon.

Best regards,
Selim

Hi Christina,

Another problem I am facing is: to change the orientation of a project, when I remove geolocation, take orientation constraints and then, reoptimize the project, orientation constraints do not work, sometimes I see them displaced (you will see in the attachment).

What are the possible reasons for this? How to solve this problem?

I look forward to hearing from you.

With kind regards,

Selim

Orientation change problem.pdf (195 KB)

Hi,

May I ask if you have tried to remove image geolocation before processing Step 1.?

Best,

Hi,

No, I did not. I followed the steps written in Pix4D online materials.

First, I ran processing step 1, removed geolocations, took orientation constraints & then reoptomized the project.

Best Regards,
Selim

Hi,

In the link Christina suggested on how to orient the project, it is written that if the project has images with geolocation and the Orientation Constraint is too different from the orientation given by the image geolocation, the Orientation Constraint will not be taken into account. When deleting the geolocation of the images without reoptimizing or running Step 1. again, this action is not taken into account.

I would suggest removing image geolocation , running Step 1., adding orientation constraints and then reoptimize.

I hope it helps.

Best,

Hi,

The orientation given by the image geolocations is very different from the one I need. So, I am going to follow your suggestion “removing image geolocation, running Step 1., adding orientation constraints and then reoptimize”.

Hope this time it will work.

Best regards,
Selim

Hi,

I prepared an orthomosaic with RGB images. When I loaded the mosaic in eCognition, I found 4 layers. What is the 4th layer and why?