FCurve Rotations Bug

Post Reply
clintonman
Captain
Posts: 2260
Joined: 21 May 2009, 21:08
Type the number ten into the box: 0
Location: California
Contact:

FCurve Rotations Bug

Post by clintonman » 19 Sep 2014, 06:15

tS_FCurve_NotGood.jpg
This is a major bug in the graph editor. The rotation curves don't match the reality of the animation. Picture shows the matrix value at the start and end of the animation. Only rz(yaw) is animated, but the curves shows rx and ry as well. This makes it impossible to use the custom bezier handles for rotational aniamtion, so I'll be using only linear interpolation ( no handles). I think the FCurve rotations are only useful in the simple case where the animated object has zero rotations to start and only rotates in one direction - x, y or z.
The translation and scale fcurves are ok as far as I know.
Clinton Reese

http://clintons3d.com

User avatar
Draise
Captain
Posts: 2955
Joined: 21 Sep 2009, 19:33
Type the number ten into the box: 0
Location: Bogota, Colombia
Contact:

Re: FCurve Rotations Bug

Post by Draise » 19 Sep 2014, 15:22

... Rotation on local or global matrix? And if so, is it euler or additive?

In softimage the fcurves also can give different values to its actual rotation. Me often than not you have local rotation and global on separate key frames.

clintonman
Captain
Posts: 2260
Joined: 21 May 2009, 21:08
Type the number ten into the box: 0
Location: California
Contact:

Re: FCurve Rotations Bug

Post by clintonman » 20 Sep 2014, 02:03

Draise wrote:... Rotation on local or global matrix? And if so, is it euler or additive?

In softimage the fcurves also can give different values to its actual rotation. Me often than not you have local rotation and global on separate key frames.
There is no option in tS to key different rotation spaces. I think that ability might be unique to Softimage. The FCurve says "Matrix" so I assume it's keying the local instead of world values. Also the object has no parenting or grouping so world and local should be the same and they do give the same values at the Matrix and WldMatrix connectors.
Truespace looks euler to me with Z/yaw being the last rotation so it should have no effect on the x and y rotations.
Clinton Reese

http://clintons3d.com

User avatar
Draise
Captain
Posts: 2955
Joined: 21 Sep 2009, 19:33
Type the number ten into the box: 0
Location: Bogota, Colombia
Contact:

Re: FCurve Rotations Bug

Post by Draise » 08 Oct 2014, 16:27

Yeah, local and world I think is it all...

User avatar
Draise
Captain
Posts: 2955
Joined: 21 Sep 2009, 19:33
Type the number ten into the box: 0
Location: Bogota, Colombia
Contact:

Re: FCurve Rotations Bug

Post by Draise » 26 Oct 2014, 23:40

Sure enough, I found your bug. I was animating a simple rotation and also got the error. I found that it initially worked fine, the fcurves reading as planned, but then when I used custom bezier or dropped one of the values beyond 360 degrees, I got the random ondulate swivels. I could not fix it. I need to test more to see what actually triggers it.... Cause it did work fine initially.... Well for me..



I will keep testing it..

User avatar
Draise
Captain
Posts: 2955
Joined: 21 Sep 2009, 19:33
Type the number ten into the box: 0
Location: Bogota, Colombia
Contact:

Re: FCurve Rotations Bug

Post by Draise » 27 Oct 2014, 15:02

Ok, found that the extremities are not limited to just Euler, as things can rotate beyond 360 degrees. So it must be additive.

Post Reply