Turning my Warhawk BattleMech

[bubba]

I have moved to the next stage of development of my Warhawk - making it turn. But I just can't seem to keep it from sliding or looking awkward. I sure could use some help.

 

LowerhalfWarhawk.prj

 

Warhawkturning1.mov

 

 

Thanks.

[John Bigboote]

It looks as if it is mechanically not built to move laterally... so thus the unnatural look. You need to add some sort of rotation-ability to the legs... not only in your rigging but you need to sell it visually as well with some sort of gear or swivel in each lower leg...

[robcat2075]

As Matt says, you'll need the leg or foot to be able to turn a bit independent of the body direction.

 

You really can't do good turns while using a repeating walk cycle.

 

This will work when you are animating each step separately as I showed you before:

 

RobotTurn.mov

 

(The audio may not play if this movie is played in a browser. In general, t's best to download all QuickTimes and play them in the QT Player.)

[johnl3d]

Wonder if the neighbors enjoyed watching that ..nicely explained

[robcat2075]

Wonder if the neighbors enjoyed watching that ..nicely explained

 

About seven years ago they took down their six-foot fence and put up a nine-foot fence.

[Meowx]

About seven years ago they took down their six-foot fence and put up a nine-foot fence.

[NancyGormezano]

Wonder if the neighbors enjoyed watching that ..nicely explained

 

About seven years ago they took down their six-foot fence and put up a nine-foot fence.

 

Umm...There wouldn't be any razor wire on top as well? perhaps? And is it surrounding a large concrete "mansion"?

[robcat2075]

Wonder if the neighbors enjoyed watching that ..nicely explained

 

About seven years ago they took down their six-foot fence and put up a nine-foot fence.

 

Umm...There wouldn't be any razor wire on top as well? perhaps? And is it surrounding a large concrete "mansion"?

 

Interesting true fact... in 15 years I have never seen them put out any garbage.

[mouseman]

Interesting true fact... in 15 years I have never seen them put out any garbage.

A grow op, maybe?

[NancyGormezano]

I vote for zombies or hoarders

[Guest carlosgf]

About seven years ago they took down their six-foot fence and put up a nine-foot fence.

 

 

It was funny

 

Your video is well explained. I thoguhht at the begining that you just picked up this video from internet... but than I saw your avatar on your shirt.

[Meowx]

Razor wire aside, acting things out yourself for reference is absolutely one of the best things you can do for your animation. It's amazing how much life you can breath into a motion - or even just a static pose - by standing up and acting it out yourself. Plus, sometimes it's just fun to get up and stomp around like a dinosaur!

[bubba]

Here is a second try (after "listening" to all the advice.) For some reason, it took the longest time to get the steps coming out of the turn correct. I don't know what it was but the feet slipped and slid for many frames.

 

Warhawkturning2.mov

 

 

LowerhalfWarhawk.prj

 

Thanks.

[robcat2075]

Here is a second try (after "listening" to all the advice.) For some reason, it took the longest time to get the steps coming out of the turn correct. I don't know what it was but the feet slipped and slid for many frames.

 

Warhawkturning2.mov

 

 

LowerhalfWarhawk.prj

 

I detect a tiny bit of sliding there. Make sure the keyframes at the beginning and end of a foot-on-the-ground portion are zero-sloped so the foot isn't wandering in between. And of course make sure the beginning and end are really in the same spot or the foot will still be moving as it tries to transition from one to the other.

[robcat2075]

I tried to take a look at the PRJ but it doesn't have the model in it.

 

None-the-less, I think zero-sloping your channels is the main to-do.

[bubba]

Sorry about not embedding the model.

 

Here is number 5. I hope I have worked out all the bugs. I did change to zero-slope. I had it set on Linear.

 

Warhawkturning5.mov

 

LowerhalfWarhawk.prj

[robcat2075]

If the start and end keys are the same, zero-sloped and linear interpolation will give the same result, no change between the keys, that you want.

 

However, as an example, take a look at the left foot from 2:12 to 2:27 where it is on the ground. The beginning and end are a bit different, so the foot will creep between. You could manually edit the second key to match the first or, easier, copy the first one onto the second.

 

[bubba]

I see what you mean. But when I change one parameter it seems to affect all and I have to go back through each step and look at it from each angle.

 

Question: What is the "W" dimension? X,Y,Z defines three-dimensional space, so what is "W" space?

[NancyGormezano]

Question: What is the "W" dimension? X,Y,Z defines three-dimensional space, so what is "W" space?

 

A:M defaults to use quaternions to compute orientation (rotation) as opposed to Euler angles (roll, pitch, yaw). You can change the default method tools/options/Action/Interpolation.

 

Quaternions use 4 coefficients, versus 3 for eulers. You can choose to change the interpolation method of existing channels by right clicking on the rotate/convert driver/euler. It is easier for humans to understand euler angles, versus quaternion values. Quaternions are traditionally preferred for computation because they handle gimbal lock.

 

The math to convert between quaternions and eulers can be googled. Quaternions are not easy to describe but here is something:

 

http://www.j3d.org/matrix_faq/matrfaq_latest.html#Q60

 

QUATERNIONS

===========

 

Q47. What are quaternions?

--------------------------

Quaternions extend the concept of rotation in three dimensions to

rotation in four dimensions. This avoids the problem of "gimbal-lock"

and allows for the implementation of smooth and continuous rotation.

In effect, they may be considered to add a additional rotation angle

to spherical coordinates ie. Longitude, Latitude and Rotation angles

A Quaternion is defined using four floating point values |x y z w|.

These are calculated from the combination of the three coordinates

of the rotation axis and the rotation angle.

 

Q48. How do quaternions relate to 3D animation?

-----------------------------------------------

As mentioned before, Euler angles have the disadvantage of being

susceptible to "Gimbal lock" where attempts to rotate an

object fail to appear as expected, due to the order in which the

rotations are performed.

 

Quaternions are a solution to this problem. Instead of rotating an

object through a series of successive rotations, quaternions allow

the programmer to rotate an object through an arbitary rotation axis

and angle.

 

The rotation is still performed using matrix mathematics. However,

instead of multiplying matrices together, quaternions representing

the axii of rotation are multiplied together. The final resulting

quaternion is then converted to the desired rotation matrix.

Because the rotation axis is specifed as a unit direction vector,

it may also be calculated through vector mathematics or from

spherical coordinates ie (longitude/latitude).

Quaternions offer another advantage in that they be interpolated.

This allows for smooth and predictable rotation effects.

Edited May 19, 2011 by NancyGormezano

[robcat2075]

I see what you mean. But when I change one parameter it seems to affect all and I have to go back through each step and look at it from each angle.

 

Question: What is the "W" dimension? X,Y,Z defines three-dimensional space, so what is "W" space?

 

nancy explained it but you don't need to really know exactly what the values mean. You just have to see that the values are obviously not staying constant while that foot should be stationary.

 

 

Set the Key Bone filter

Set the Rotate and Translate filters

Set the time to the first key

Select the bone

CTRL-C

Set the time to the second keyframe

CTRL-V

 

now the first and second keyframes match and the bone will not wander

[bubba]

Question: What is the "W" dimension? X,Y,Z defines three-dimensional space, so what is "W" space?

 

A:M defaults to use quaternions to compute orientation (rotation) as opposed to Euler angles (roll, pitch, yaw). You can change the default method tools/options/Action/Interpolation.

 

Quaternions use 4 coefficients, versus 3 for eulers. You can choose to change the interpolation method of existing channels by right clicking on the rotate/convert driver/euler. It is easier for humans to understand euler angles, versus quaternion values. Quaternions are traditionally preferred for computation because they handle gimbal lock.

 

The math to convert between quaternions and eulers can be googled. Quaternions are not easy to describe but here is something:

 

 

 

Thanks Nancy. I remember the antenna and solar array engineers talking about quaternion equations and gimbal lock. I had no idea what they meant then, nor do I now. Can I change the preferences in Tools>Options>Action from Quaternion to Euler and delete the W curves?

[NancyGormezano]

Can I change the preferences in Tools>Options>Action from Quaternion to Euler and delete the W curves?

 

That would only change the default method for new curves I believe.

 

You can not just delete the w curve from existing curves as there is not a 1 to 1 correspondence between the x,y,z of quaternions and x,y,z channels of euler angles - but for existing curves, you can right click on the rotation property of the model or of the bone and choose convert driver to/euler. Then it will be setup properly, and you can deal with only 3 channels.

[robcat2075]

If you change a channel from Quat to Euler A:M will delete the W on its own as it is converting the values to Euler which has no W.

 

But for character animation you really want to keep Quat. Euler is going to get you weird gimbal lock problems.

 

Do the copy/paste I explained. It works, I've done it a hundred times in the same situation. If it's not working, you're doing something wrong.

[bubba]

Once again into the valley rode the 600 ....

 

If you go real slow, in the project chor window, there is a wobble in the upper leg bone of the left leg. I don't know what is causing it. I did covert all the quaternions to eulers, that I could find. Thanks Nancy.

 

Warhawkturning6.mov

 

 

LowerhalfWarhawk.prj

[robcat2075]

Once again into the valley rode the 600 ....

 

If you go real slow, in the project chor window, there is a wobble in the upper leg bone of the left leg. I don't know what is causing it.

 

This is really a problem with the "knee" trying to "aim at" the knee target which is placed rather low. Flopping to the side was the "best" it could do given that the lower leg will prevent it from getting any closer.

 

At the start of your scene, move the Left Knee Target up to about the level of the Pelvis bone. That seems to be a better overall place for the knee to try to point to without having to move sideways.

 

Since legs and arms move quite a bit in animation, elbow and knee targets often have to be managed during an animation to get the best result.

 

 

 

 

I did covert all the quaternions to eulers, that I could find.

 

Go back to your Quat version. The only time you really need Euler is for things like propellers that need to spin more than 180°.

[robcat2075]

btw, there are a bunch of "Bone Position" channels in your chor. These are caused by a bug in copy/paste. You can and will want to delete all those whenever they occur.

 

For example, expand "Pelvis" to see some of them.

[bubba]

This is really a problem with the "knee" trying to "aim at" the knee target which is placed rather low. Flopping to the side was the "best" it could do given that the lower leg will prevent it from getting any closer.

 

At the start of your scene, move the Left Knee Target up to about the level of the Pelvis bone. That seems to be a better overall place for the knee to try to point to without having to move sideways.

 

Since legs and arms move quite a bit in animation, elbow and knee targets often have to be managed during an animation to get the best result.

 

But this is the same model I used in my other animation and it did not have the leg wobble (straight walk.)

 

 

Go back to your Quat version. The only time you really need Euler is for things like propellers that need to spin more than 180°.

 

Why should I go back. Three dimensional space seems easier to understand. Also, if the W driver is to make things smoother, do I really need that for my machine?

[bubba]

btw, there are a bunch of "Bone Position" channels in your chor. These are caused by a bug in copy/paste. You can and will want to delete all those whenever they occur.

 

For example, expand "Pelvis" to see some of them.

 

I saw them and wondered where the hell they came from. Then in the Reports section I saw you bug report. I have added a "monitor".

[robcat2075]

This is really a problem with the "knee" trying to "aim at" the knee target which is placed rather low. Flopping to the side was the "best" it could do given that the lower leg will prevent it from getting any closer.

 

At the start of your scene, move the Left Knee Target up to about the level of the Pelvis bone. That seems to be a better overall place for the knee to try to point to without having to move sideways.

 

Since legs and arms move quite a bit in animation, elbow and knee targets often have to be managed during an animation to get the best result.

 

But this is the same model I used in my other animation and it did not have the leg wobble (straight walk.)

 

Maybe you crouched a bit lower or higher or lefter or righter than before. It's like a cliff, it's not a problem until you go too far.

 

 

 

Go back to your Quat version. The only time you really need Euler is for things like propellers that need to spin more than 180°.

 

Why should I go back. Three dimensional space seems easier to understand. Also, if the W driver is to make things smoother, do I really need that for my machine?

 

3D space... you got that either way. But we're talking about angles. Quat will interpolate from angle to angle far better and more intuitively than Euler. You don't need to know what the W means. All anyone really does with those curves is look at them and ascertain whether they are changing as fast or as slow or as not-at-all as you want.

[robcat2075]

btw, there are a bunch of "Bone Position" channels in your chor. These are caused by a bug in copy/paste. You can and will want to delete all those whenever they occur.

 

For example, expand "Pelvis" to see some of them.

 

I saw them and wondered where the hell they came from. Then in the Reports section I saw you bug report. I have added a "monitor".

 

TIP: when shift-selecting several bones for copy/pasting keys...

 

Copy while the KeyBone filter is on and Paste with the KeyModel filter on.

 

That prevents Bone Position Keys from being made.

[bubba]

I think you might be right about the model moving slightly different. I don't have the body "bobbing and weaving" this time, so I might have "fallen of the cliff."

 

Ok, I will go back to Quaternion (as long as you say I don't need to understand W, or how to adjust it .) Steffen also said that it did not make sense to delete the W curve (in a bug report) because it is created internally anyway.

 

One thing I did notice in working with this particular animation was that sometimes the values in the CP Properties window did not match the values in the Object Properties panel. I would change one but the change would not be reflected in the other.

[robcat2075]

While researching I found this anti-Quaternion article:

 

http://www.gamedev.net/page/resources/_/do...aternions-r1199

 

It's a lot of math i don't understand but most interesting is the Editor's note at the top disavowing it, in light of an apparent flame war revolving around it.

 

The writer's apparent proposal to ditch both Euler and Quat for "Vector" interpolation may have merit... in my brief test in A:M it seems quite functional. I'll test further when i have some time.

[Gerry]

I started reading the article and linked thread and I was doing fine, right up until the math started.

[robcat2075]

My "Editing Quaternions" video, created for users of other programs who say Quaternions are impossible to use.