Camera Location and Centre Vector

[photo_man]

Hello,

I have a chunk in a local coordinate system with scale bars.
For all enabled cameras I want to extract their X,Y,Z coordinate in the local system and a vector that goes through the centre point of the image.

If it's also possible, I would also like to get the X,Y,Z coordinate of the intersection point of the vector with the mesh.

Is this possible through python scripting?

Thanks!

[saduka]

Hello,

I have a chunk in a local coordinate system with scale bars.
For all enabled cameras I want to extract their X,Y,Z coordinate in the local system and a vector that goes through the centre point of the image.

If it's also possible, I would also like to get the X,Y,Z coordinate of the intersection point of the vector with the mesh.

Is this possible through python scripting?

Thanks!

Hey Photo_man, you can use the following code to retrieve points and rotation in local 3D coordinates like the following if the chunk has not been assigned with a geographic coordinates. The vector through image center is just [0,0-1]*rot(yaw)*rot(pitch)*rot(roll)
how metashape defines the rotation order and rotation matrix have to be checked on somewhere else

Code: [Select]

label_l,  yaw_l, pitch_l, roll_l=[],[],[],[]
    cameras=chunk.cameras
    for camera in cameras:
        T = chunk.transform.matrix
        m = chunk.crs.localframe(T.mulp(camera.center))
        R = m * T * camera.transform * Metashape.Matrix().Diag([1, -1, -1, 1])
        row = list()       
    for j in range (0, 3): #creating normalized rotation matrix 3x3
        row.append(R.row(j))
        row[j].size = 3
        row[j].normalize()
    R = Metashape.Matrix([row[0], row[1], row[2]])
    yaw,pitch,roll = Metashape.utils.mat2ypr(R) 
    yaw_l.append(yaw)
    pitch_l.append(pitch)
    roll_l.append(roll)
    label_l.append(camera.label)   
    table= pd.DataFrame(data={'Label':label_l,'Yaw':yaw_l,'Pitch':pitch_l,'Roll':roll_l}, 
                        columns=['Label','Yaw','Pitch','Roll'])   
    table.to_csv(pose_file)     


[photo_man]

Thanks Saduka!

That got me set in the right direction, but I found a more direct way to get the camera vector in my chunk coordinates. This is the code I'm using, if it's useful to anyone.

Code: [Select]

import PhotoScan #Photoscan 1.4.1

file1 = open(path, "wt")
chunk = PhotoScan.app.document.chunk #active chunk
cameras=chunk.cameras

for camera in cameras: # Loop through cameras
    if camera.enabled is True: # Only apply to enabled cameras
        estimated_coord = chunk.crs.project(chunk.transform.matrix.mulp(camera.center)) #estimated XYZ in coordinate system units
        cam_vec_int = camera.transform.mulv(PhotoScan.Vector([0,0,1]))
        cam_vec_ext = chunk.transform.matrix.mulv(cam_vec_int)
        cam_vec_ext.normalize()
        print(camera.label, estimated_coord, cam_vec_ext)
file1.write(camera.label + "," + "{:.8f}".format(estimated_coord) + "," + "{:.8f}".format(cam_vec_ext) + "\n")  #writing cameras
file1.write("#\n")
file1.close()