Reverse engineering a pos-rot correctly

[Beowulf Zessinthal]



oh .. and the gerbil asked me to say hello .. k.

[mikka Luik]

Have to reply just because Gerbils are the second centre of the Universe (after cats of course) but as an attachment.. not sure if av as surrogate root prim does have an easy answer.

Position I think would be simply reverse of Gerbil local pos? Rotations I leave to those I hold in awe who eat quats as snacks.

[Qie Niangao]

Yeah, there's nothing tricky here, except to realize that an attachment can only be positioned relative to its attachment point; there's no way to know where the avatar center is relative to that attachment point at any instant (which changes viewer-side as the avatar skeleton animates). But I'm not sure what's actually being asked here.

[Beowulf Zessinthal]

it sure is a puzzle!

heres a diagram that exposes the trickyness of it..

(assume the gerbil is attached to avatar center so the pos rot is a constant knowable in theory)

I know SL has a few very tricky math functions and am posting this in case its somehow 'easy' (!)

..otherwise i will just have to reconfigure my own approach the thing im trying to solve^^

Happy Days!



[Beowulf Zessinthal]

Ok .. if you like to do the problem thats ok .. but i personally have to back out of this rabbit hole!

My brain can no longer follow it well enough to implement anything useful (!) .. so i need to wipe my brain slate, and solve my design in a different way ..  while avoiding this particular route!  .. ok.. thx for indulging tho!

[Qie Niangao]

The graphic confuses me -- I just cannot figure out what you're trying to do, and it now appears you've moved on to a different approach (which may be wise anyway, depending on the actual underlying objective).

But for completeness, there's no mystery in the numbers shown in that graphic. Below is a little script that generates those numbers, as expected. 

vector CHILD_POS = <0.13972, -0.25801, 0.79175>;rotation CHILD_ROT = <-0.26105, -0.46572, -0.22815, 0.81419>;default{    state_entry()    {        /*        // In root of 2-prim linkset, replicate the orientation in the graphic        llSetLinkPrimitiveParams(2,             [ PRIM_POS_LOCAL, CHILD_POS            , PRIM_ROT_LOCAL, CHILD_ROT            ]);        */        rotation inverseChildRot = ZERO_ROTATION / CHILD_ROT;        llOwnerSay("\nInverse Pos = "+(string)(ZERO_VECTOR - (CHILD_POS * inverseChildRot))            +"\nInverse Rot = "+(string)inverseChildRot);    }}

Now I'm the worst possible choice for explaining stuff, but hopefully that script will help.

One inverts a rotation by dividing the ZERO_ROTATION by it. ("Dividing" is short-hand for "applying the '/' operator; one must never assume that all properties of those operators apply to quaternions as they do to real numbers.)

And because you're switching to the child's coordinate system, you'll need to express the root's position by inverting (subtracting from ZERO_VECTOR) that child's offset from the root rotated by that inverted rotation.

I'm pretty sure this all helps not one whit, so just attribute this post to a spell of OCD.

[Beowulf Zessinthal]

You have it! Thanks!

.. you and your ilk are amazing to have around.

I know partially some of the things you described, but trying to keep it in my head and derive the correct permutation of the formula was too much for me!

OK .. thanks so much Qie Dirac (!) .. the inverse calc is correct, now i will see if it can sort out my problem (!)