Define the relationship between device orientation/motion events and screen orientation in the spec about deviceorientation HOT 14 CLOSED

Why am I reminded of this cartoon? :)

from deviceorientation.

Further discussion on this topic is at http://lists.w3.org/Archives/Public/public-webapps/2014JulSep/0203.html.

In this thread @sicking also argues that we should incorporate Screen Orientation in to returned Device Orientation data – therefore providing alpha, beta and gamma values that match the current 'up' frame relative to the screen – by adding the following to the spec:

partial interface DeviceOrientationEvent
{
  readonly attribute double? screenAlpha;
  readonly attribute double? screenBeta;
  readonly attribute double? screenGamma;
}

from deviceorientation.

Proposed spec change is now available as Pull Request #10.

from deviceorientation.

Is there some way we can stop with the alpha, beta, gamma nonesense?

from deviceorientation.

What do you want to see instead @marcoscaceres?

from deviceorientation.

Angles and x, y, z.

from deviceorientation.

Generally speaking we can say that alpha = z, beta = x, gamma = y.

But depending on the world coordinate system you use that can change. e.g. three.js employs a different coordinate system where alpha = y, beta = x, gamma = -z (see: https://github.com/mrdoob/three.js/blob/master/examples/js/controls/DeviceOrientationControls.js#L48).

from deviceorientation.

What three.js is doing might be good - Do you know of other libs that are changing the coordinate system?

from deviceorientation.

I believe CSS transforms match the z-x-y convention of device orientation events.

There are lots of different conventions for coordinate systems though. You can have left-handed and right-handed conventions, the Y and Z axis switched around (like in three.js) or the axis that is perpendicular to the ground being either positive or negative toward the center of the earth. You can also have any combination of those factors together. It isn't a one-size fits all thing.

from deviceorientation.

Sure, but no one is doing alpha, beta, gamma and negative angles, right?

from deviceorientation.

The angles are mathematically sound :) That is to say, you can compute quaternions and rotation matrices from them as they are defined today. Euler angles are generally defined using both positive and negative offset from a default position. More info is available at https://en.wikipedia.org/wiki/Euler_angles#Signs_and_ranges.

from deviceorientation.

I'm sorry if that is how this discussion came across.

We do not plan to re-define standard 3D Cartesian coordinate space though. If you have more info on an alternative coordinate system then we can discuss it here.

from deviceorientation.

Wikipedia appears to say that it's not a requirement that one use negative angles. I don't think we need to define some alternative system: just one that is more developer friendly. If we are seeing evidence that developers are changing coordinate systems, they might have a good reason.

from deviceorientation.

Closing as a long overdue task per #10 (comment) decision to not add new features that are not implemented by any vendor.

from deviceorientation.