[Bf-committers] bone matrix docs clarifications?

[Campbell Barton]

On Sat, May 22, 2010 at 5:58 AM, Bassam Kurdali
<bkurdali at freefactory.org> wrote:
> Hi, I've been trying to document the various bone matrices for my own
> usage in python, and ran into some inconsitancies/missing docs, possible
> bugs or clarifications. I'd like to update the docs with the results of
> my findings.
>
>
> editbone matrices:
> ******************
>
> *matrix: DOC: "Read-only matrix calculated from the roll (armature
> space)"
> ------------------------------------------------------------------
> notes: it's a 4x4 matrix, seems to store bone location and rotation
> (including roll).. saying it is calculated from roll alone seems not
> fully true. It is in Armature space, as the docs say.
>
>
> bone matrices:
> **************
>
> *matrix: DOC: "4x4 bone matrix relative to armature"
> ----------------------------------------------------
> notes: 4x4 transformation matrix in armature space.
>
> *matrix_local: DOC: "3x3 bone matrix"
> -------------------------------------
> notes: this seems to be a bone rotation matrix in parent bone space.
>
>
> pose matrices:
> **************
>
> *matrix: DOC: "Final 4x4 matrix for this channel"
> -------------------------------------------------
> The matrix is in armature space, represents loc/rot/scale, and does seem
> to show the final position; fits the docs
>
>
> *matrix_channel: DOC: "4x4 matrix, before constraints"
> ------------------------------------------------------
> I don't understand this one: but it seems constraints do change it,
> negating the idea that it represents the unconstrained transformation.
> Is that a bug? or is the doc wrong? or am I crazy?
>
> *matrix_local: DOC: "Matrix representing the parent relative location,
> scale and rotation. Provides an alternative access to these properties."
> ------------------------------------------------------------------------
> notes: the doc here does not jibe with experimentation (unless I
> misunderstood the doc or the experimentation): this matrix gives the
> bones' transformation from it's *rest* position, nothing todo with
> parent.
>
> cheers
> Bassam
> --

Hi Bassam, I'll try reply to your mail within a week or 2, need to
look into it and see the differences myself, possibly modify/add to
existing properties.
In 2.4 we really did a bad job in this area, so I want to be very
clear on the difference between all matrix types too!

-- 
- Campbell