2 added 221 characters in body
source | link

I hate to tell ya, but the 'hand' of a set of coordinates is actually subjective -- it depends on which way you imagine you are looking at things. An example, from the OpenGL cube map spec: if you imagine being inside the cube looking out, the coordinate system is left-handed, if you think of looking in from the outside it is right handed. This simple fact causes endless grief, because to get a numerically correct answer all of your assumptions have to be consistent, even though locally it doesn't actually matter which assumption you make.

The OpenGL spec is officially "handedness neutral" and goes to great lengths to avoid this issue -- that is, to put it off onto the programmer. However there is a "hand change" between eye coordinates and clip coordinates: in model and eye space we are looking in, in clip and device space we are looking out. I'm always struggling with the consequences of this no doubt accidental complication.

The cross-product is our friend, because it lets you generate a definitely right handed or left handed 3D coordinate system from a pair of vectors. But which hand it is depends on which of those vectors you call "X".

I hate to tell ya, but the 'hand' of a set of coordinates is actually subjective -- it depends on which way you imagine you are looking at things. An example, from the OpenGL cube map spec: if you imagine being inside the cube looking out, the coordinate system is left-handed, if you think of looking in from the outside it is right handed. This simple fact causes endless grief, because to get a numerically correct answer all of your assumptions have to be consistent, even though locally it doesn't actually matter which assumption you make.

The OpenGL spec is officially "handedness neutral" and goes to great lengths to avoid this issue -- that is, to put it off onto the programmer. However there is a "hand change" between eye coordinates and clip coordinates: in model and eye space we are looking in, in clip and device space we are looking out. I'm always struggling with the consequences of this no doubt accidental complication.

I hate to tell ya, but the 'hand' of a set of coordinates is actually subjective -- it depends on which way you imagine you are looking at things. An example, from the OpenGL cube map spec: if you imagine being inside the cube looking out, the coordinate system is left-handed, if you think of looking in from the outside it is right handed. This simple fact causes endless grief, because to get a numerically correct answer all of your assumptions have to be consistent, even though locally it doesn't actually matter which assumption you make.

The OpenGL spec is officially "handedness neutral" and goes to great lengths to avoid this issue -- that is, to put it off onto the programmer. However there is a "hand change" between eye coordinates and clip coordinates: in model and eye space we are looking in, in clip and device space we are looking out. I'm always struggling with the consequences of this no doubt accidental complication.

The cross-product is our friend, because it lets you generate a definitely right handed or left handed 3D coordinate system from a pair of vectors. But which hand it is depends on which of those vectors you call "X".

1
source | link

I hate to tell ya, but the 'hand' of a set of coordinates is actually subjective -- it depends on which way you imagine you are looking at things. An example, from the OpenGL cube map spec: if you imagine being inside the cube looking out, the coordinate system is left-handed, if you think of looking in from the outside it is right handed. This simple fact causes endless grief, because to get a numerically correct answer all of your assumptions have to be consistent, even though locally it doesn't actually matter which assumption you make.

The OpenGL spec is officially "handedness neutral" and goes to great lengths to avoid this issue -- that is, to put it off onto the programmer. However there is a "hand change" between eye coordinates and clip coordinates: in model and eye space we are looking in, in clip and device space we are looking out. I'm always struggling with the consequences of this no doubt accidental complication.