Hi all,

I am brainstorming ways of displaying to the user the orientation of
an object over time in as compact and meaningful way as possible. I
thought I would see what others have tried and used.

Hi all,

I am brainstorming ways of displaying to the user the orientation of
an object over time in as compact and meaningful way as possible. I
thought I would see what others have tried and used.

trace of the normalized orientation-vector onto the xy-plane.

On 22 Okt., 02:53, Dave <thedaverud... at (no spam) gmail.com> wrote:

Hi all,

I am brainstorming ways of displaying to the user the orientation of
an object over time in as compact and meaningful way as possible. I
thought I would see what others have tried and used.

If you'r going through the trouble of putting numbers into some visual
display, why not 'paint' the vector itself?

I don't know weather accessing pixels on a surface has become easier
lately, or weather I just have been hm .. slow .. But, it's
available.

An ordinary surface (D2) is not the complicated 3D you would like, but
when you've only got one item to draw, it's a pretty moderat problem
to make it look '3D' .. draw the 'head' and the 'rod' in the order
called for (pointing in or out), and let them be drawn in different
colors/tones.

On 22 Okt., 02:53, Dave <thedaverud... at (no spam) gmail.com> wrote:

Hi all,

I am brainstorming ways of displaying to the user the orientation of
an object over time in as compact and meaningful way as possible. I
thought I would see what others have tried and used.

Or, if you stick to a graph-type of display .. calculate the projected
trace of the normalized orientation-vector onto the xy-plane.
Eventually two graphs in different colors indicating z positive or
negative. The output would be a curly line around (0,0)

Carsten

trace of the normalized orientation-vector onto the xy-plane.

'normalized orientation-vector' ...  is not necessarily obvious.
Transform (0,1,0) for a series of angle-values and catch the
transformed x and y for your graph.

Carsten

On Oct 22, 1:58 am, Carsten <cTroelsga... at (no spam) email.dk> wrote:

trace of the normalized orientation-vector onto the xy-plane.

'normalized orientation-vector' ...  is not necessarily obvious.
Transform (0,1,0) for a series of angle-values and catch the
transformed x and y for your graph.

Carsten

Again, this won't display the roll of things.

Hi all,

I am brainstorming ways of displaying to the user the orientation of
an object over time in as compact and meaningful way as possible. I
thought I would see what others have tried and used.

I was thinking of having a 2D graph where the x axis is time, but am
open to other ideas. For the graph, I can think of several ways:

Euler Angles - i.e., have three lines along the time graph, one for
each of the individual euler angle values. This is most compact, but
strange things happen toward the poles.

Quaternions - Not as compact, and don't freak out as much. However, I
would probably have to put special code in to handle dual values
(i.e., Q represents the same rotation as -Q)

Axis Angles - As compact as Quaternions, and also have the same
duality problem as quaternions, but also have the problem of angle x
representing the same orientation as x + 2 pi.

3x3 Rotation Matrix - Not compact in the least. The upside is that
there is no duality problem, no singularities, none of that special
case handling that other rotation representations have. Or am I
missing something?

Thanks.

Dave

I am brainstorming ways of displaying to the user the orientation of
an object over time in as compact and meaningful way as possible.