better support for rigid camera rig

[nickfan]

Hello,

Thanks a lot for creating such beautiful software. I have read a lot forum posts here and are amazed by the rapid development of photoscan. Just a suggestion that may make photoscan even better for photo-scanning with synchronized multiple rigid cameras. I see camera import and export can be used for a fixed camera rig. However, after the rig is moved, a new chunk is needed for reusing the calibrated data. Am I right here? I hope there can be a calibration that can save the relative positions of the cameras and optimize photo aligning as a whole. one only have to open a calibration file that contain the data about number of cameras and relative positions, then the program will assume the relative positions of a sequence of photos in different folders (for CAM1, CAM2...).
This way it should increase accuracy of photo aligning process and solve issue of no enough texture for aligning. This can be especially helpful for aerial scanning for eg area with lots of vegetation. Another scenario that will greatly benefit from this feature is object 3D modeling using turntable.

What do you think?

Nick

[Alexey Pasumansky]

Hello Nick,

If the cameras remain still and their relative orientations and positions do not change it is possible to use the same camera data (Export cameras feature on the first camera position). Please note that camera labels should be the same.
But it will work only if you have absolutely the same camera positions and orientations for all cameras in chunk.

However, I'm not sure how it could be used in aerial surveys.
Maybe you can explain the idea if I have understood it not correctly?

[RalfH]

This sounds like a potentially useful feature. Alexey, the way understand it it is not about a fixed setup of all positions from which images are taken for a complete set of photos. Rather, I think this is about using a fixed set of cameras (e.g., a stereo pair or, in aerial imagery, a forward-nadir-aft setup) and take multiple time-synchonized images with all cameras in this setup. As the relative positions and orientations of the cameras are know, I think this could help to improve geometric quality or at least reduce processing time in photo alignment. I see multiple potential applications, from makro to studio to PAP to aerial.

[nickfan]

This sounds like a potentially useful feature. Alexey, the way understand it it is not about a fixed setup of all positions from which images are taken for a complete set of photos. Rather, I think this is about using a fixed set of cameras (e.g., a stereo pair or, in aerial imagery, a forward-nadir-aft setup) and take multiple time-synchonized images with all cameras in this setup. As the relative positions and orientations of the cameras are know, I think this could help to improve geometric quality or at least reduce processing time in photo alignment. I see multiple potential applications, from makro to studio to PAP to aerial.

That is exactly what I mean. By constraining the relative positions of camera rig with accurate calibration, it should reduce the possible output and hence increase accuracy and reduce alignment time.

Take object 3D modeling as example. Assuming I make a calibrated stereo camera rig to improved the quality and I take 36 shots around at 10 deg interval for each camera, using the current implementation, I need to make 36 chunks in order to use the calibrated camera info. That is not too productive. Now what if you can constraint the relative positions of the camera and optimize the rig as a whole, then I can import all 2x36 shots at once and still benefit from the calibrated stereo pair.
ideally, you allow to save a calibration for any number of cameras whose images are saved at separate folder.

Nick 

[FoodMan]

+100

[mrrafs]

+1

[totoromo]

Hi,

I'm exhuming this old topic as i haven't found anything new about the subject.
Has there been any advance a better support for camera rigs ?

[patorf]

Is there anything new about this topic. Would love to use a calibrated multi-camera rig.

[dskarlat]

So, i suspect it is valid to be done assuming a rigid rig airborne or mobile (the later being my interest). We are doing the same on board a car. We suppose that the two wire-synchronized cameras are rigid, but we would like to test that as well. So i should prepare (i.e) 1000 directories with cam1.jpg and cam2.jpg inside (two camera rig) and import folders as groups. OK so far
When processing alignments etc, will i be able to calculate the relative positions among the two cameras and their uncertainty across the 1000 positions?

[dskarlat]

It is mentioned in the manual 'The default assumption is that synchronized cameras have the same position in space. In case distancebetween the cameras cannot be neglected, it is possible for Metashape to calculate relative camera offset.If offset parameters are known you can set them manually.'. As far as i understand Metashape does not support moving rigid rigs (such as those in UAVs with multi camera configurations, or multicamera rigs on top of cars for mobile mapping. Is this right?

[Alexey Pasumansky]

Hello dskarlat,

If you have several cameras rigidly mounted relatively to each other, you can use multi-camera system approach.

After the images are loaded via Add Folder command according to the folder layout, you can go to the Tools Menu -> Camera Calibration dialog where in the Slave Offset tab (for each slave camera) you can define the measured offsets between the sensors (if known) or allow Metashape to estimate them by enabling Adjust Location option.

[jenkinsm]

Hello dskarlat,

If you have several cameras rigidly mounted relatively to each other, you can use multi-camera system approach.

After the images are loaded via Add Folder command according to the folder layout, you can go to the Tools Menu -> Camera Calibration dialog where in the Slave Offset tab (for each slave camera) you can define the measured offsets between the sensors (if known) or allow Metashape to estimate them by enabling Adjust Location option.

Hi Alexey,

I am wondering about this in regards to cameras mounted in/on a vehicle (being used to capture a road while driving along it). In my case, the camera mounting positions are fixed relative to each other, but the mounts are not 100% stable (they vibrate a little) so the relative positions change slightly from one shot to the next. Also, the cameras are not synchronized and the FOV is different for each camera.

In this case, is it recommended to use the "Adjust Location" option or not?

Thanks!

[James]

+100

Another +100 please

Would be great to be able to export the calculated slave offset and import as reference in future projects, rather than have to type all those number in each time!

I'm using a stereo rig for close up handheld scanning and being able to import the camera offsets would allow me to get my scans scaled properly without worrying about markers and scalebars.

Alternatively, being able to select a master camera and create a scalebar between it and the slave would achieve the same thing for me, but this doesn't seem to be supported.

If there is an existing python solution I'd be happy with that, and otherwise I I'll try to figure one out myself!