Talk:Getting Started: Difference between revisions
No edit summary |
No edit summary |
||
Line 55: | Line 55: | ||
I like the new ideas. Thanks for your contributions! | I like the new ideas. Thanks for your contributions! | ||
PS: to do the IMU perspective thing, I consider using photos of an actual camera and to blend it into the axis thing in the IMU configure tab... |
Revision as of 13:32, 8 May 2014
Okay, getting closer. But your IMU example does not match mine. I as well have the dot left bottom, I as well look at the camera from the bottom (so that "right" is on the wrong side, but my index no is 15, in your example it is the zero. So either we change the index numbers or we assume the IMU was mounted on top of the camera and the right arrow is pointing to the right.
Agreed?
->
I'm not sure I get exactly what you say, but the "old" picture describes a different situation than seen in your setup photo.... in your photo one is looking at the camera FROM BELOW while in the "old" picture one is looking at the camera FROM ABOVE (I guess I said this before ;))... the "right" is correct and no index number needs to be changed, it's just this different perspectives... maybe have a look at Greg's addition to this figure in the rcg thread... I'll provide you with some additional figures
Understood but I find it confusing. I would still not be able to relate this picture to my concrete example. What should we do?
->
Not sure what to do. I'll send a new picture with an explicit photo of a camera instead of the blueish scheme, I would hope that this avoids at least missreading it. I have my doubts though that this by itself would simplify relating things or remove any confusion. Ideally you would figure out what it exactly is what makes you confused and makes you not to be able to relate the pictures.
I think the basic problem is that it's not everyone's cup of tea to relate coordinate axes seen from different perspectives. I.e. one photo shows a camera from below, the next from above, the next from the front etc. etc.
I could imagine if one would stay within one perspective things might get simpler. In fact, I tried to aid the user into this with the grey axes labeled front, right, up shown in the GUI window. Following up on this I could suggest the following general recipe. Look at your camera from the front, i.e. directly into its lens. Then the meaning of the grey axes and their relation to the camera should be obvious. Now, still looking at your camera from the front, identify in which direction the IMU's z axis points to. It can only be one of six cases: it can point towards you, away from you, to the left, to the right, to the top, or to the bottom. Sounds fairly easy to me. Next, still looking at your camera from the front, do the very same for the x axis. What do you say?
PS: maybe the figure with teh many configurations should be totally removed
->
Frankly I find the current method to most easy one. The z-axis does grow out of the IMU chip, so where does it point to? It points down in my example. The x-axis can be seen on the board but is the one pointing in reading direction (from the white dot left to right in the direction of the printed text on the chip), it points forward. Done.
Your image I find confusing. How about that change?
The last thing I find confusing is where the camera right side is. I would have thought that if I hold the camera in my hand for shooting, the right hand side would be to the right. But be it as it is, at least it is not confusing anymore with all the examples, especially the coordinates (front/right) you have added to my picture.
->
just have seen your new IMUorientation figure... yes, that's what I had in mind too... I only would also remove the camera figure, so that this figure is generally valid
your comments however gave me another idea. One equally well could turn my above suggestion around, namely instead of looking at the camera from a given perspective and determine the orientation of the IMU one also could also look at the IMU module with a given perspective and determine the orientation of the camera!! so, a recipe could go as follows: Look straight at the top of your IMU module, such that the x axis arrow points to the right. The dot on the IMU chip should be in the left top corner. Now, still looking at your IMU from atop, identify which face of your camera you see. It can only be one of six cases: you can see the front, the back, the right side, the left side, the top or the bottom. How the story goes on for the next setting I have to think briefly about... but you may see what the point is
left and right depend on the perspective, if you look at your camera the right is the opposite to when you hold your camera in shooting position
maybe all this would be easier if one just would refer with the axes to the situation where on holds the camera in shooting position?
->
I believe we are getting closer. I like the idea of holding the IMU in one position and then describe the camera. If you would change left/right in the doc because of the more natural "shooting position" point of view, we would need to change the GUI as well. Something I'd shy away. In that case I would align the documentation to the current reality only.
Do you want to make the changes in the docs?
->
Both the IMU perspective and the right/left thing will need changes in the GUI, and accordingly in the docs... before we jump to the docs let me first change the GUI and let us see in what this results... then everything else will unfold by itself
I like the new ideas. Thanks for your contributions!
PS: to do the IMU perspective thing, I consider using photos of an actual camera and to blend it into the axis thing in the IMU configure tab...