Translation vector of camera pose

I have calculated the camera pose using the Aruco marker. The translation vector of the marker from the camera is [-114.91, -45.44, 1913.51 ], while that of the camera from the marker is [-188.38, 74.10, 1906.78]. The question is can I check if the translation vectors are correct by looking at the location of the marker and the image center as per the image below.


The blue cross is the image center. Would it be correct to say for marker position from camera x should be negative, y negative, and z positive? while for camera position from marker, x should be positive, y negative, and z positive?


hard to tell. the marker is too far away to see a clear perspective. you’d have to look at the camera from the position of the marker… and then you’d have to know the orientation of the marker. the z-distance makes sense though.

The marker in camera coordinates looks right, but the camera in marker coordinates doesnt look right because x should be positive and y negative. The marker is oriented just as shown in the image.

Rotation by -180 degrees seems to solve the problem but I dont know why that should be the case.

marker frame:

  • x right
  • y up
  • z near

marker faces slightly to the right of the camera, i.e. the camera is to the marker’s right, so camera’s pose in marker frame should have negative x.

this is hard to tell because it faces very nearly to the camera.

if you had a different picture, where the marker doesn’t face the camera so directly, this would be a lot easier to tell. your picture is rubbish for telling precisely on which side of the zero in x and y the camera’s pose should be. the marker IS facing the camera so both are near zero.

neither the marker nor the algorithms are at fault. it is your perception.

Well as an observer of the setup I can tell you that the orientation is just as I have said (see image below), secondly the camera to marker pose is another clue. This seems to be a general problem from what i gather from looking at what others are saying on the forum.

that picture (the direction of the blue Z-axis) says: the camera, viewed from the marker, is more -x, and not as much +y.

do you agree? your numbers agree:

The ‘y’ I can agree but what of the ‘x’? shouldn’t it be positive?

why? the red axis is +X
“we”, the camera, are on the other, the negative (-X), side of the YZ plane

what others are saying

this is getting “philosophical”.

we can argue about what we see, and what we think we see, and how it agrees or disagrees with the math we think we understand. that’s an issue on the level of individuals.

in general, it’s a typical case of confirmation bias. when others agree with you, that doesn’t mean they’re right or you’re right. it merely means you agree. you can all be wrong because you all have the same misconceptions. that is a general point. “consensus” is worthless when it comes to humans. it used to be “consensus” that babies don’t feel pain, so you can operate on them without anaesthesia. with all due respect, FUCK consensus. this isn’t meant to be personal.

the camera center is to the left of the marker, or postive x direction. Will sleep over it and see if i can see it your way. cheers

the marker faces towards the camera. its red axis, which is the positive X, points to “its” left (it can be said). since we are to its right, we are on the negative X side.

imagine you are the marker. you are facing the camera. your left arm is the red axis. the blue axis is your nose, it points away from you towards the camera.

are you trying to find an explanation for your conclusion, or are you trying to reach a conclusion from observations? you seem to be tricking yourself.

Edited my comment,. I should have said the camera is to the left of the marker.