Zxy!, as a default for camera and objects!

[newmeler]

Is there a way to make maya set the axis to zxy for cameras and all objects created at zxy to avoid gimbal lock?

thanks

mh88

[iyaiya]

in my understanding, gimbal lock is caused by ambiguity in the Euler Angle representation of angles. As far as I know, the only thing you can do about it is switch to using quaternions, but MEL doesn't support them.

How is changing the order of the basis vectors going to help?

[joojaa]

Well it helps so far as to in wich order the things are exacuted wich means gimbal lock occours at a different angle tahan in the other order.

One can ameke a scriptjob to do this, but as said it realy doesnt help with gibal lock at all.

quarteronions are used trough euler angles in maya (eulers get converted to querteronions offcourse you loose the ability to controll individuasl curves), se eulers have not got so much problems pointing in all directions but rather the problem is how to interpolate between them.

[bradon]

I have a simmilar problem where I need to change the rotation order, however in doing so I get a screwy camera movement. I think this may be an example of Gimbal Lock?

I worked up an example as this is the best way I can think of to describe what is happening. I made two cameras and animated them. One has a rotation order XYZ that one is fine, the other one has a rotation order of ZXY and it seems to flip at a certain point.

It seems like this locking or flipping is more consistent with ZXY order. I need to export camera data in ZXY order but I can't seem to animate a camera very well or at all when the order is ZXY.

can anyone tell me why this is? If it is Gimbal lock? why does the camera not lock up as often or at all with the XYZ rotaton order? and what I can do to avoid it?

many thanks.

here is the link
http://www.paraportable.net/files/screwy\_camera.ma

[joojaa]

better set the rotate tool to gimbal ant manipilet some object it tells youll see whats going on. se when you rotate the fist angle a certaim n amount youll end up in aposition where the impact ofg the ones that come behind are dimished.

Gimbal lock is whan two of the rotations are the same way, the jump yoru experiencing is not a lock per see its just a interpolation issue.

Eulers are prone fror this becausd the rottations depend on the retation that came before it. Theres no way around this. See euler angles can describe any angle in 3d space thats true. BUT it cant interpolate between themall that well atkeast if you animate the rotation channels.

So whie you can easily set positions in eulers if you can sometimes its usefull to interpolate in quats isntead.

[bradon]

so then does quats not use a rotational order? I found something about the rotational order is always XYZ with a third W variable. basically I would need this to be in some form or ZXY order because I need to export the camera data with that order.

most importantly though I can't figure out how to change the rotational interpretation (to quats) of an aimed constricted camera. because the rotate channels are locked by some hidden driven expression. Maybe if I could figure out how to change the rotate interp to quat and then reconvert back into eruler with ZXY order this flipping problem would be solved? I would really like to use aimed cameras and seem to have no problem in the XYZ order but in this instance where I need to export the camera, it must be in ZXY.

I am currently trying to constrain orientation and translation of a ZXY camera to the origional XYZ camera and we'll see if that works although for workflow it would be nice to figure out a more fluid way of working in ZXY order.

[joojaa]

QUOTE(bradon @ 02/15/06, 11:23 PM) [snapback]228338[/snapback]
so then does quats not use a rotational order? I found something about the rotational order is always XYZ with a third W variable. basically I would need this to be in some form or ZXY order because I need to export the camera data with that order.

No they have no order you better read this http://www.euclideanspace.com/maths/geomet...tions/index.htm

Anyway, thinkof quats as somesort of WIERD MAthematical way of representing axis angle. With computational benefits.

Id be amazed if anyobody in their sane mind used eulers for other than user input.

[bradon]

Thanks thats a great link

QUOTE(Joojaa @ 02/15/06, 03:11 PM) [snapback]228350[/snapback]
Id be amazed if anyobody in their sane mind used eulers for other than user input.

are you saying that you never use eulers?

[joojaa]

QUOTE(bradon @ 02/16/06, 02:06 AM) [snapback]228356[/snapback]
are you saying that you never use eulers?

Only in cases where i operate a mechanical rotation sucha s a real joint. Offcourse the input is still in euler angles but the interpolation never is.

And no for external applications i wouldnt use eulers for interpolation id just convert everything internaly to quats ist so much easier and intuitive to do the code. Offcourse as input numbers quats are horrible.

[bradon]

how do you convert the interpretation of a camera rotation from eulers to quats?

I would normally just select the rotate keyframes in the graph editor and change the interpretation on them, but when I have a camera with an aim I cannot select the rotate keyframes.

[joojaa]

Well with aim theres no point relay! Since the eulers controll nothing*, therefore theres no interprtation/interpolation issue. Having aim and up defined is just like using axis angles to start with!

However on the downside thers no eulers to export!

• That is to say each frames rotation is sloved form the aim routine. IN many ways aim has the xact sma properties as quats. HOwever if you eulerize it later for export youre probably in world of trouble.

[bradon]

So in your earlier post about having an interpretation issue with the maya camera file I posted, what kind of interpretation failure am I getting by using an aim constricted camera? and how would I fix that? for some reason I get the interpretation issues only in ZXY order.

[joojaa]

The cameras are copies right!? Big mistake, you cant just go ahead and change the rotation order of the camera and expect it to work, it will not se cameras have a value called constraint rotate order. NOw triuck hgere is its not connected to yoru cam SO it expects your camera to be oriented like it was when you first called the create menu, but you changed it.

run

connectAttr -f camera1.rotateOrder camera1_group.constraintRotateOrder;

No its not a gimbal lock its a stupid user/scripter error (50% you not udnerstanding 50% some lazy ass not chekking the connection so that if the user cahnnges the option wich he shouldnt once theres anim in place).

its like yu changed the cork opener screw into a lefthanded one without bothering to inform people

[bradon]

wow, that was amazing, Thanks!

p.s. How do you seem to know the answer everytime?

[joojaa]

QUOTE(bradon @ 02/17/06, 07:43 PM) [snapback]228487[/snapback]
p.s. How do you seem to know the answer everytime?

Becaus i dont aswer if i am not right,

No seriously, its about knowing what i do. You know a beginning artist can take into account one or two things of what they do. But as a matter of fact I consider about a 1000- 10,000 possinle things about each and every one thing i do (the implications and what comes next, and after that and after that), it comes naturaly because theres nothing i do is by mistake.

for this answer i considered well above 30 differnt things taht i can ACTUALY anticipated (including whats your knowlege base), not to caount all other things i automaticaly account for, wich is just what my brain does. And i did this by reading the ma file because i had no maya at my disposal. so im not just one step ahead of most of you in genaral im 10- 100. But im worng sometimes and i do admit whan ! am.

i suggest calculus for everyone and trig and .... oh ande reading teh friggin manual.

Ask any of the other old timers and they aklk have read the manual cover to cover! And ask abot the accounting factor too, imean most of them had 2 or 3 things consid3ered whanb they started now its too much toi describe. Bateman, Duncan (ok now theres a REAL guru!), everybody, failure and the bunch testify.