2 Camera Stereo Rig

PIX4Dcapture PIX4Dcapture Questions/Troubleshooting
1

I am interested in a 2 matching camera rig in stereo. 

We are modeling tall thin, very complex tower structures. 

What impact would 2 matching cameras in stereo, fixed at an oblique angle, have on Pix4d processing? 

Would the cameras be better in parallel or angled in toward each other? And what distance do they need to be apart from each other to see the most benefit?

The current plan is to use 2 Sony a5100’s in parallel stereo set exactly 1 meter apart. We fly about 5-7 meters from the tower.

Thanks

2

If there is any test imagery available, happy to run it through Pix4D for you!

1 Like
3

Hi Daniel,

Could you please give me more details on your project?

Why are you looking for a rig set-up? Is it a flight time constraint?

To set up the rig, you will have to know the relatives and translation between the sensors which could be a tricky process. 

Looking forward to hearing from you,

Ina 

4

Hello Ina,

The project is self supporting cellular and radio towers. We find it exceptionally difficult to capture the intricacies of the lattice work of the tower, as well as around the antennas. 

We get

  1. a lot of noise from the surrounding area. For example, if the tower is in a green field, the tower and interior is cased in green ‘fuzz’ that takes forever to mask and remove. 

  2. poor resolution of the tower.

We have tried a Phantom 4 Pro taking over 1,000 pictures at 20mp, and a Sony UMC R10C at 20mp. We are flying about 20 feet away from the tower and imaging at an oblique angle. We are mainly using another photogrammetry suite and getting good results, but it is geared more toward virtual reality than BIM, and would like to switch to Pix4D but not if we can’t get it to work.

So as cameras in stereo are used for depth mapping, would it help the Pix4D algorithms process a complex structure like this, if we used 2 cameras in stereo? Would a low resolution dense cloud from a LiDAR unit help Pix4D process? Would a much higher resolution camera like a 40-50mp camera help? Would adding precision GPS help? 

Thanks for your help.

5

Hi, please find my answers below:

  1. I theory it would help having two cameras mounted together, but they have to be triggered at the same time so that the rig can be configured in PIX4D. If they are not, then you should process them individually and then merge the projects. For merging please see: https://support.pix4d.com/hc/en-us/articles/202558529-Merging-Projects

  2. Adding an external point cloud would not be helping for the noise since it can be used only for creating generate the DSM, orthomosaic, and other outputs after processing step 3 and will not be combined with the one generated from the images.

  3. Having a higher resolution sensor could help but what is more important is to have proper camera parameters so that you could perform a calibration. See https://support.pix4d.com/hc/en-us/articles/206065716-How-to-calibrate-a-Perspective-Lens-Camera

4.The GPS does help the reconstruction but not the noise and artefacts in particular; the most import for the latest one is right camera parameters, enough overlap, quality of images (not blurry).

I hope this helps,

Ina