|
|
Rotate and new Animation Editor Problem
About Truespace Archives
These pages are a copy of the official truespace forums prior to their removal somewhere around 2011.
They are retained here for archive purposes only.
Rotate and new Animation Editor Problem // Bugs
Post by ed_baker // Jul 8, 2007, 4:44am
|
ed_baker
Total Posts: 355
|
Trying to keyframe a shape made from a cylinder, and I run into the following problem.
I rotate using the widget on the Y axis, set the keyframe. Rotate back using the widget to the other side, set the keyframe and then back again. Similar to a pendulum. Now, the problem is that the translation is being written based I can only guess from the previous keyframe and now the object is floating all over the place when the animation plays back. Serious problem here. I tried to fix it by deleting translation keyframes from the dope sheet, only problem is that now,the axis for the object is at the bottom of the object. Make any sense? Anyone else having this problem?
This problem doesn't seem to happen for the Cube or Torus shapes, but it happend for every other one. I can fix it by grouping (in modeler) a cylinder to a cube asa child, but I still have to delete the translation KFs in the workspace. Then it rotates propery.
See below for screen shots and a copy of the file. The blue cylinder has had the translation KFs deleted, but now rotates as if the axis is at the bottom of the object or world z=0 even though the axis is still at the object center.
Any help would be great, or is this a bug that has to be fixed?! No problem with old KFE when trying to rotate these objects.
Ed |
Post by Jack Edwards // Jul 8, 2007, 6:16am
|
Jack Edwards
Total Posts: 4062
|
Ed have you checked your FCurves? TS uses splines so that might be causing the floatiness in the animation. If I have time later I'll give your scene a look. Animation's not my strong point though. :p
-Jack. |
Post by ed_baker // Jul 8, 2007, 6:58am
|
ed_baker
Total Posts: 355
|
Ed have you checked your FCurves? TS uses splines so that might be causing the floatiness in the animation. If I have time later I'll give your scene a look. Animation's not my strong point though. :p
-Jack.
Hey Jack,
That's exactly where the problem is. See below for the FCurves for the translation KFs for the cylinder. There are values where there shouldn't be giving the cylinder a nice ride around the screen. Next after that you see the KFs for the tralslation values in the FCurves for the cube. No values other than the postition it is supposed to be in.
When you try to even out the KFs in the the curves to one level you end up with the problem of the axis rotating from the floor or base of the object like the blue cylinder from the example above. |
Post by ed_baker // Jul 9, 2007, 3:01pm
|
ed_baker
Total Posts: 355
|
Gotta Bump this thread. Any one else able to recreate the problems above? Norm, Tom, anyone?
DL the scene above or try the same steps and see if you have the same problem please?
Ed |
Post by ed_baker // Jul 11, 2007, 6:19am
|
ed_baker
Total Posts: 355
|
Can anyone re-create this problem?
I'm surprised no one is answering.
ANYONE?!
Ed |
Post by Steinie // Jul 11, 2007, 6:24am
|
Steinie
Total Posts: 3667
|
I would help you in a second but not my expertise! |
Post by ed_baker // Jul 11, 2007, 6:26am
|
ed_baker
Total Posts: 355
|
I would help you in a second but not my expertise!
Thanks Steinie,
No biggy. I was pointing out a possible bug that is getting in the way of something I'm working on and wanted to make sure it wasn't just me.
Ed |
Post by frootee // Jul 11, 2007, 6:41am
|
frootee
Total Posts: 2667
|
Ed,
I think I read about this in the animation chapter.
Change the F Curves from spline to Linear. With spline curves even if it's flat you'll still get movement.
So set the values you want at the start and end keyframe. I think you want these to be the identical.
That should solve the problem.
Frootee |
Post by ed_baker // Jul 11, 2007, 6:47am
|
ed_baker
Total Posts: 355
|
Ed,
I think I read about this in the animation chapter.
Change the F Curves from spline to Linear. With spline curves even if it's flat you'll still get movement.
So set the values you want at the start and end keyframe. I think you want these to be the identical.
That should solve the problem.
Frootee
Hey Frootee,
I had already tried that. No difference. Try it yourself if you can real quick. Rotate one way, then goes past dead center to rotate the other way, and back. Like a pendulaum. This has to be on either x or y axis. Z doesn't have this problem. Like I pointed out above, this doesn't happen on cubes or toruses. I want to see if anyone out there can re-create the problem.
Ed |
Post by frootee // Jul 11, 2007, 6:53am
|
frootee
Total Posts: 2667
|
OK. Since the cube rotates properly, did you look at its F curve properties for comparison?
This may help, and it may not. Look at Chapter 9, pages 87 and 88. I know you've already tried this but it may shed some light on the problem.
I don't have truespace here at work so I can't load it.
To clarify, leveling the bezier splines between keyframes is not the same thing as settnig the interpolation to Linear between the two keyframes.
Sorry I can't load it up right now because I would like to help more!
Frootee |
Post by ed_baker // Jul 11, 2007, 7:00am
|
ed_baker
Total Posts: 355
|
OK. Since the cube rotates properly, did you look at its F curve properties for comparison?
This may help, and it may not. Look at Chapter 9, pages 87 and 88. I know you've already tried this but it may shed some light on the problem.
I don't have truespace here at work so I can't load it.
Frootee
Thanks Frootee,
Those pages don't really apply, but I see the point you are making. If you scroll up to my second post, you can see the translation FCurves from the example above for both a cube and a cylinder. Cube is perfectly flat as it should be. Cylinder KFs are taking their information from somewhere else.
See what I mean?
Ed |
Post by frootee // Jul 11, 2007, 7:04am
|
frootee
Total Posts: 2667
|
Exactly. Please reread my last post; I just edited it; probably after you were typing this.
The cube looks like it is set for linear interpolation. Hence the straight line. The cylinder is set for bezier interpolation. The symptoms you are describing sound like this is occurring, but since I can't load truespace (dangit), I can't confirm. I will tonight.
But, you should be able to select each of these F curves to verify what type of interpolation is being used, for the cube, and for the cylinder. Could you do that for the x, y, and z axes for both the cube and cylinder, and post the results here?
BTW I am having some strange issues with physics and jerky fixation points; my issue is somewhat related.
Thanks Frootee,
Those pages don't really apply, but I see the point you are making. If you scroll up to my second post, you can see the translation FCurves from the example above for both a cube and a cylinder. Cube is perfectly flat as it should be. Cylinder KFs are taking their information from somewhere else.
See what I mean?
Ed |
Post by ed_baker // Jul 11, 2007, 7:52am
|
ed_baker
Total Posts: 355
|
Exactly. Please reread my last post; I just edited it; probably after you were typing this.
The cube looks like it is set for linear interpolation. Hence the straight line. The cylinder is set for bezier interpolation. The symptoms you are describing sound like this is occurring, but since I can't load truespace (dangit), I can't confirm. I will tonight.
But, you should be able to select each of these F curves to verify what type of interpolation is being used, for the cube, and for the cylinder. Could you do that for the x, y, and z axes for both the cube and cylinder, and post the results here?
BTW I am having some strange issues with physics and jerky fixation points; my issue is somewhat related.
I see what you mean, but I'm not leveling the KFs. Mainly because you can't set them to exact values, so there can still be some drift, or the weird result of having it rotate at base of the object.
Setting the curves to linear either before or after the fact isn't the problem. If you look at my first example, you see the cube's translation KFs are exactly flat as they should be because the cube has not moved anywhere, just rotated in place. Doing the same for a cylinder does not have the same results.
I even tried setting beginning and ending keyframes before setting the others, and then I went into the curve editor and selected all of the keyframes and set them to linear. Next I set the two rotations inbetween the starting and ending KFs. Below is what happened. A very nice "linear" curve, but the KFs have values that drift from the starting and end values resulting in the cylinder floating around instead of just rotating in place like the cube and torus do.
Ed |
Post by frootee // Jul 11, 2007, 8:27am
|
frootee
Total Posts: 2667
|
Man. What a pain huh?
Although you may not be able to set the keyframe values exactly in the KFE, you can go to that keyframe, then right click the object selector tool, and manually set the values in the popup window. If you do this and you have linear interpolation for translation, and the cylinder still pulls this, then I'd say it's a bug.
Try this then, to see if this will work for you.
Add a bone to the cylinder, and add a positional (Translation) lock to both joints. Then add an IK handle to one of the joints. If necessary, configure the settings for either the LMB or RMB for rotation. When you grab the handle, the cylinder should rotate as expected. You can then set keyframes for rotation, without the undesired translation.
I hope that helps.
I will look at your scenario tonight.
My issue was with physics. I created a cylinder, and added fixation points. The cylinder not only rotated but also slowly translated towards the ground, as if the fixation points were attached by a rubber band. |
Post by frootee // Jul 11, 2007, 8:40am
|
frootee
Total Posts: 2667
|
another thing to try. Go into the link editor, select the cylinder, export its matrix, and for tx, ty, and tz, plug them into a constant. That oughta keep the cylinder still...
Frootee |
Post by ed_baker // Jul 11, 2007, 8:50am
|
ed_baker
Total Posts: 355
|
Man. What a pain huh?
Although you may not be able to set the keyframe values exactly in the KFE, you can go to that keyframe, then right click the object selector tool, and manually set the values in the popup window. If you do this and you have linear interpolation for translation, and the cylinder still pulls this, then I'd say it's a bug.
Try this then, to see if this will work for you.
Add a bone to the cylinder, and add a positional (Translation) lock to both joints. Then add an IK handle to one of the joints. If necessary, configure the settings for either the LMB or RMB for rotation. When you grab the handle, the cylinder should rotate as expected. You can then set keyframes for rotation, without the undesired translation.
I hope that helps.
I will look at your scenario tonight.
My issue was with physics. I created a cylinder, and added fixation points. The cylinder not only rotated but also slowly translated towards the ground, as if the fixation points were attached by a rubber band.
That's funny about the skeleton. I thought about that this morning, but don't have time to mess with it right now either. Wouldn't be too much of a pain, just unnecessary extra work. :(
Regarding your physics problem, I haven't had that problem, and when I've worked with just single constraints, and they worked fine. Have you double checked that the fixation points are at the same z value?
Ed |
Post by ed_baker // Jul 11, 2007, 8:51am
|
ed_baker
Total Posts: 355
|
another thing to try. Go into the link editor, select the cylinder, export its matrix, and for tx, ty, and tz, plug them into a constant. That oughta keep the cylinder still...
Frootee
I'll have to try that. What could I use as a constant? I haven't gotten that deep into the LE
Ed |
Post by frootee // Jul 11, 2007, 9:07am
|
frootee
Total Posts: 2667
|
Zero comes to mind.
Basically, whatever x, y, or z value you want to hold your cylinder at. For starters, I would just use 0 for all three, just to use a simple number.
I will do that also tonight and upload the RsScn file if it works.
Also make sure your axes are where you want them to be. You may have to do this:
1. Turn on axis visibility for the cylinder using the Axes/Location icons
2. Are the axes in the center of the cylinder? If not, use the axes/location icons again (forgot the name of the correct one) to center the axes inside the cylinder
3. TUrn off axis visibility
4. Make Sure the Object Move Tool is Off
5. Try to rotate object.
6. If you cannot rotate the object, then turn on the Object Rotate Tool.
4. try rotating and keyframing
I'll have to try that. What could I use as a constant? I haven't gotten that deep into the LE
Ed |
Post by ed_baker // Jul 11, 2007, 9:47am
|
ed_baker
Total Posts: 355
|
Also make sure your axes are where you want them to be. You may have to do this:
1. Turn on axis visibility for the cylinder using the Axes/Location icons
2. Are the axes in the center of the cylinder? If not, use the axes/location icons again (forgot the name of the correct one) to center the axes inside the cylinder
3. TUrn off axis visibility
4. Make Sure the Object Move Tool is Off
5. Try to rotate object.
6. If you cannot rotate the object, then turn on the Object Rotate Tool.
4. try rotating and keyframing
Tried it this way too. I'd like to look at your example of the constant. That would be great if you could put that together.
Thanks again for all your help.
Ed |
Post by frootee // Jul 11, 2007, 11:12am
|
frootee
Total Posts: 2667
|
will do. You know, in baseball, I'd already be on the seat. :p
Frootee |
Post by SteveBe // Jul 11, 2007, 11:56am
|
SteveBe
Total Posts: 282
|
Hi Ed,
I have reported this issue and problems I've been having to developers.
My suggestion for now would be to use model side to set axis poiints.;) |
Post by frootee // Jul 11, 2007, 3:59pm
|
frootee
Total Posts: 2667
|
Hi Ed.
I looked at the FCurve editor. This is what I did:
1. for tx, starting with the last keyframe, I set the Interpolation to: Constant End Value.
2. Then I went back to each keyframe and repeated the process, selecting
the interpolation to Constant End Value.
3. This left some spikes, so I just moved those spikes to be inline with the
rest of the curve.
Here's the result.
I fixed the cylinder.
Try doing this with the cone to see if you have any luck with it.
You can also try, Constant Start Value, and start from the left.
Note: On close inspection I still see a little bump after frame 10. |
Post by frootee // Jul 11, 2007, 4:50pm
|
frootee
Total Posts: 2667
|
UPDATE:
This is improved. The jerk after frame 10 is gone. This is what I did:
For tx and tz:
1. In FCurve editor, I started at the first keyframe and Set the Interpolation method to Linear.
2. I repeated this for the other keyframes.
3. I then set the keyframe values as close to zero as possible
4. Then, I adjusted the vertical scale of the FCurve editor:
In the upper right corner of the FCurve editor, there are two values, one on top of another. I changed these from their current values to -1 (on top), and +1 (on bottom).
This zoomed into the data at the scale I needed in order to set all tx and tz keyframes to ZERO.
Here's a screen shot, and the fixed file: |
Post by ed_baker // Jul 13, 2007, 5:38pm
|
ed_baker
Total Posts: 355
|
Hi Ed,
I have reported this issue and problems I've been having to developers.
My suggestion for now would be to use model side to set axis poiints.;)
AH-HAH!!!!
That's It SteveBe!! You are a genius! Only problem is that you have to flip back and forth in order to set everything, but that's really no big deal. At least now it works. :jumpy:
Thanks again!!!!
ED |
Post by ed_baker // Jul 13, 2007, 5:44pm
|
ed_baker
Total Posts: 355
|
UPDATE:
This is improved. The jerk after frame 10 is gone. This is what I did:
For tx and tz:
1. In FCurve editor, I started at the first keyframe and Set the Interpolation method to Linear.
2. I repeated this for the other keyframes.
3. I then set the keyframe values as close to zero as possible
4. Then, I adjusted the vertical scale of the FCurve editor:
In the upper right corner of the FCurve editor, there are two values, one on top of another. I changed these from their current values to -1 (on top), and +1 (on bottom).
This zoomed into the data at the scale I needed in order to set all tx and tz keyframes to ZERO.
Here's a screen shot, and the fixed file:
Close but still not quite there Frootee. I tried this before, and while it gets rid of the jumping around, it moves the axis to the bottom of the object. The cylinder is now rotating properly, but now from the wrong place.
SteveBe's suggestion of setting the axis in the modeler side worked immediately. You'd have to re-set all the keyframes for any animation, or else it appears it will move the object to where the axis is and not the other way around when you switch back from the modeler.
Ed |
Post by frootee // Jul 14, 2007, 3:12am
|
frootee
Total Posts: 2667
|
I see your point ed. I created a simple scene from scratch using just a simple cylinder and came up with the same results.
Well, at least I got some more practice in.
EDIT: One thing I would like to point out is, the axis locations in Workspace are not the same as the axis locations in Modelview when you first create the
cylinder:
In Workspace, when you make the axes visible, they are in the center of the object.
Go to Modelview, make the axes visible, and they are in the Center of the object.
Do note that this occurs only if the object is created in Workspace.
If the object is created in ModelView, the axes are automically centered in the object. So when you go to Workspace, they are still correctly centered, and
there is no translation.
But then again the idea is to move away from Modelview, since it's going away right?
Bug. Good call SteveBe.
Froo |
Post by ed_baker // Jul 14, 2007, 4:52am
|
ed_baker
Total Posts: 355
|
I see your point ed. I created a simple scene from scratch using just a simple cylinder and came up with the same results.
Well, at least I got some more practice in.
EDIT: One thing I would like to point out is, the axis locations in Workspace are not the same as the axis locations in Modelview when you first create the
cylinder:
In Workspace, when you make the axes visible, they are in the center of the object.
Go to Modelview, make the axes visible, and they are in the Center of the object.
Do note that this occurs only if the object is created in Workspace.
If the object is created in ModelView, the axes are automically centered in the object. So when you go to Workspace, they are still correctly centered, and
there is no translation.
But then again the idea is to move away from Modelview, since it's going away right?
Bug. Good call SteveBe.
Froo
Yeah, I'm now able to move on in the scene I'm working on. It's really a simple thing and was planning on the bones workaround, but just han't gotten to it yet. Probably wouldn't have noticed it if I hadn't been trying to do it. But at least there is a work around for now.
Ed |
Post by frootee // Jul 14, 2007, 4:57am
|
frootee
Total Posts: 2667
|
this is true; but, I noticed I had other strange problems with boning a character.
Maybe I was doing something wrong but, once I boned a character (maybe before...), when I tried rotating it, it appeared to be rotating about the origin. :confused: |
Post by DigitalSoapbox // Jul 14, 2007, 6:33am
|
DigitalSoapbox
Total Posts: 71
|
I've recreated the steps you've mentioned and I'm running into the exact same problem, so it's at least been confirmed. |
Post by ed_baker // Jul 15, 2007, 3:00am
|
ed_baker
Total Posts: 355
|
this is true; but, I noticed I had other strange problems with boning a character.
Maybe I was doing something wrong but, once I boned a character (maybe before...), when I tried rotating it, it appeared to be rotating about the origin. :confused:
I had one where I was able to move the axis, and successfully rotated from a place outside the center of the object, but couldn't change the axis after that. Might also have been becaus I had already KF'd the boned object,but can't remember if I deleted the KF's or not.
Ed |
|
