Rotations - once again

[Darkie Minotaur]

Rotations and me will never be friends. It's a simple task: I have an evelope - the body is the root, the flap is the child. On touching the envelope, I want the flap to open (to rotate up and reset the position, so it is still aligned to the envelope body) - somehow, I don't get to do it correctly. Here is what I have (the script is in the root):

 

default {
    touch_end(integer num_detected) {
    	rotation grIncrementX = llEuler2Rot(<160.0,0.0,0.0>*DEG_TO_RAD);
    	list liposrot = llGetLinkPrimitiveParams(2, [PRIM_POS_LOCAL, PRIM_ROT_LOCAL]);
		rotation newrot = grIncrementX * llList2Rot(liposrot, 1);
    	vector newOffset = llList2Vector(liposrot, 0) * grIncrementX;
    	llSetLinkPrimitiveParams(2, [PRIM_POS_LOCAL, newOffset, PRIM_ROT_LOCAL, newrot]);
    }
}

 Where do I go wrong (yet again)?

[Kimm Paulino]

Oh, I still don't get them either ... but it looks like you are rotating the "rot" and "position" in different axis/directions - one is increment * oldrot, the other is oldpos * increment ... shouldn't they both be the same?  in rotations order matters IIRC.  Can't offhand remember which one you need mind

(Everything I know about child rotations is here - but its from the point of view of the child, not the root rotating the child - http://kimmscripts.wordpress.com/2011/05/25/child-prim-rotation/)

ETA: Actually, won't you also have to compensate for the centre of rotation being at the edge of the flap?  Unless you've use path-cut type tricks (as you can do with a door).  That is what I had to do in the script on the above link.

When I'm stuck, I tend to look at Void's rotation page on the wiki: https://wiki.secondlife.com/wiki/User:Void_Singer/Rotations

ETA again: fixed link (mumble, grumble ... auto link recognition ... brackets ... mumble, grumble)

Kimm.

 

[Rolig Loon]

You don't need to change the locl position of the flap.  All you're doing is rotating it.  I can't check this in world, but I think you can do that by just writing

llSetLinkPrimitiveParamsFast( 2, [ PRIM_ROT_LOCAL, newrot ] );

[Dora Gustafson]

A guess: The order of arguments are right?
Did you try:

rotation newrot =  llList2Rot(liposrot, 1)*grIncrementX;

[Rolig Loon]

You might be right, Dora.  I never trust myself completely without testing in world to see what happens, and I usually trust your instincts about rotations before I will trust my own.  My general rule, though, is that operations are applied from right to left.  Normally, that means you start with the current rotation (rightmost), then apply a correcting rotation (to the left of it). 

[Darkie Minotaur]

Thank you for your suggestions.

@Dora: I followed Void's rotation tutorial "To rotate in the Self Frame (prim's own axis):

• (vRotAmount2ChangeBy * vRotCurrentLocal)"

I tried it, too - the result wasn't any better

 

@Rolig: I do reposition it - flap isn't rotated around the center of the flap. 

[Rolig Loon]

Oh.  I didn't see any position offset in your snippet, so I assumed that your flap was just a cut prim that you have twisted 45 degrees on both X and Y so that it could rotate like a door does around its cut edge.  In that case, your position should be

llSetLinkPrimitiveParamsFast( 2, [PRIM_POS_LOCAL, llList2Vector(liposrot,0) + offset_vector*grIncrementX*llList2Rot(liposrot,1) ] );

where you still need to define the offset_vector.  (Again, I' feel more comfortable if I could test this in world, but it feels right.)

[Darkie Minotaur]

The offset of a child in the link set, I assumed, is it's local position - am I mistaken here?

[Rolig Loon]

In this context, it's the amount you want to move the child prim from its current position.  So, if you want to move it 0.1m in the direction of its local X-axis, the offset is <0.1,0.0,0.0> .   The logic of my suggestion was that you want to set the prim's position to its current local position, corrected by moving it some distance that is, itself, adjusted by rotating it by some increment (160.0 degrees) relative to its current local rotation .... if that makes sense.  :smileywink:  

[Rolig Loon]

Darkie --  The easiest way to solve the problem is to avoid it.  I'll send you a demo in world that uses Void's simple hinge script.  It works like a dream.  (I'll still keep poking at your way of doing it, though, because it ought to work, even if it's much harder. )

[Darkie Minotaur]

Thanks a lot! I will have a look tomorrow.

[Rolig Loon]

Oooo... This turns out to be really tricky if you don't use the easy way out.  (The easy way is to use a cut prim and Void's simple hinge.)  If you don't use a cut prim for the envelope flap, you have to calculate an offset based on the width of the flap and the angle that you will be rotating it through.  Here's what works, including the code necessary to close the flap again on a second touch.  Notice that I was wrong earlier.  You do not need to apply any rotation at all to the position offset, once you calculate how much the offset is.  [Note:  The flap on my envelope is a rectangular prim, flattened on Z and oriented so that it will hinge on its Y-axis, so the offset I need to calculate is on the X-axis.  It's also necessary to offset on the Z axis, because otherwise the leading edge of the envelope flap does not stay level with the envelope as it opens. This was not intuitively obvious to me until I started messing with it, but you'll see what I mean if you leave the Z correction out.]

 

rotation adjust;vector offset;default{    state_entry()    {        adjust = llEuler2Rot(<0.0,-120.0,0.0>*DEG_TO_RAD);        vector Size = llList2Vector(llGetLinkPrimitiveParams(2,[PRIM_SIZE]),0);        offset = <-(1.0-llCos(llRot2Angle(adjust)))*0.5*Size.x,0.0,-llSin(llRot2Angle(adjust))*0.5*Size.x>;    }            touch_start(integer total_number)    {        if (llDetectedLinkNumber(0) == 2)        {            list temp = llGetLinkPrimitiveParams(2,[PRIM_POS_LOCAL,PRIM_ROT_LOCAL]);            vector Lpos = llList2Vector(temp,0);            rotation Lrot = llList2Rot(temp,1);            adjust = ZERO_ROTATION/adjust;            llSetLinkPrimitiveParamsFast(2,[PRIM_POS_LOCAL, Lpos + (offset = -offset), PRIM_ROT_LOCAL,adjust*Lrot]);        }    }}

 

[Kristin Burner]

mew

 

Rotation hurts my prim brain soon to be a meshie brain.

I cheated one one thing i did, i rotated ti manually , recorded the new rotation, then i scripted it to go from both points, that worked for what i was making.

[Darkie Minotaur]

*snickers* that's exactly how I do it at the moment - but for several reasons it is better for my thinggie to calculate pos and rot. If I ever understand what Rolig did, that is

[Innula Zenovka]

This is how I ended up doing it, which rotates on a different edge from yours (I'm not quite sure which edge of the child prim it's supposed to use as the pivot):

 

  rotation grIncrementX;  default {    state_entry(){           grIncrementX = llEuler2Rot(<160.0,0.0,0.0>*DEG_TO_RAD);    }    touch_end(integer num_detected) {                       grIncrementX.s*=-1;        list liposrot = llGetLinkPrimitiveParams(2, [PRIM_POS_LOCAL, PRIM_ROT_LOCAL,PRIM_SIZE]);        vector mySize = llList2Vector(liposrot,2);         float myX = mySize.x;        float MyY = mySize.y;        float myXOffset = myX*0.5;        float myYOffset = MyY *0.5;       vector  vVecOffset = <-myXOffset, myYOffset, 0.0>;            rotation local = llList2Rot(liposrot,1);      llSetLinkPrimitiveParamsFast(2, [PRIM_POS_LOCAL, llList2Vector(liposrot,0)+ (vVecOffset - vVecOffset * grIncrementX) *local,                       PRIM_ROT_LOCAL, grIncrementX * local]);    }}

 

[Darkie Minotaur]

Thank you, Innula - I think that I can mod it so it does what I want!

[Qwalyphi Korpov]

The resident who provided the previous content, if any, has replaced it with this generic statement.

[Kimm Paulino]

---

Thank you for pointing to that item Kimm.

---

You are most welcome!  Of course, I am just basking in the reflected glory actually due to Void for having put the page together in the first place Kimm.