thought I'd try

[mtpeak2]

I thought I'd try some real animating. Started with a simple walk cycle.

walk.mov

[TheSpleen]

looks smooth. A+

[itsjustme]

Looks good, Mark!

[kwhitaker]

I thought I'd try some real animating. Started with a simple walk cycle.

 

looks good, Mark

[rusty]

I thought I'd try some real animating. Started with a simple walk cycle.

 

It is not a bad walk cycle -- especially if it is your first attempt (in this case I say A+ too). I believe that a person 'could' walk like that but to me it doesn't look very real. I can see the subtle secondary action -- good work. It's not bad though I can pick it pieces but I wouldn't dare because despite my many years of study on this, I'm lousy at it. In case you haven't noticed, walk cycles are a bitch. I develop mine by trying, tossing, trying, tossing (repeat half a dozen more times) then accidentally getting it right.

 

Okay... I lied. I will pick at it a bit. There's no sense of weight (quickly bringing the foot down after the heel touches helps as well a the knee bending a tad for a frame or two when it takes the weight). The back foot does not 'push off' enough. After it pushes off it speeds up and comes forward a little faster. Right after the push off is the highest point in the cycle (for the character or head), the cross over is the second highest. It doesn't seem to move up and down enough. Walking is basically falling forward and catching yourself over and over... the body should seem to lean forward after the rear push off. I can't see the front but the shoulders and hips move in a figure 8 opposite from each other, that is to say, they both rotate from a top view and also from a front view and do so opposite from each other (I really can't tell if this is happening for sure). The hand moves in a subtle figure 8 as well from a side view. Again I can't see the front (or a birds-eye) but... from the front view... the center of gravity shifts back and forth over the leg supporting the weight and, the foot bends outward as it comes forward... and, the hands come inward when in front and also when in back to a lessor degree.

 

The human walk is a very complex affair... some people can get them right every time in just a few seconds and... I hate each and every one of those people LOL. ;-)

 

Rusty

[nino banano]

hi mtpeak2...I´m agree with rusty, nice start...I think it needs to move a little the head as well ...cheers

[robcat2075]

Hi Mark, well it's about time you were getting around to animating!

 

If I were to pick out one crucial missing element it would be the compression pose right after the heel contacts the ground. Basically the body falls into each new step and doesn't immediately rebound up at the moment the heel touches the ground. It takes a *few* frames for the leg force to undo the downward momentum of the body and get it going back up again.

 

Richard Williams talks about this in his book on p. 103 (and I show it in my vid on posing walks. In the tut link in my sig.)

 

I think that is the hardest part of walks to get working right.

[John Bigboote]

A great start. I find animating to be very 'additive'...in that you start out with something (like this) then you clear your mind, take a break... come back to it with 'fresh eyes' and see what else it wants. My 1st impression is that it 'wants' a little hip sway added, maybe a little hip up and down as well. The hands are good, but try offsetting them to get a little 'lag' on them. Then- take a break...

[mtpeak2]

Thanks everybody.

 

I tried to adjust everyones comments, head movement and lean forward/weight. I'm sure I screwed everything up.

 

I think I could rig a character faster than it took me to animate and tweak this, which still isn't right. Now I now why I rig than animate.

walk_side.mov

walk_front.mov

walk_birdseye.mov

[phatso]

That's a pretty jive-ey walk. Prolly not what you were after, but I often get things I wasn't after, and they turn out useful so I save them. If you were to put a big smile on the guy's face and have him walking up to a hot woman, it would be perfect.

 

I 'specially like the secondary motion of the hands.

[Animus]

Looks good Mark!

 

You took a realistic model for a start, those are hard, our eyes are experts with those guys.

Do you have 0 slope on the first and last frame, there seems to be a break in the cycle.

Is he walking along a path, or just cycling the action?, since there are no shadows or other reference

object, it's hard to tell. Anyway, I think it is nice to try on a path, this way it is easy to try a cycle at different speeds.

We build cycles with 24 frames most of the time, but everything is not happening in 1 second.

I think with this action, he could move faster since he seems to be putting some effort with his arms forward

and head rotating. Our forearms don't go very far in a relaxed walk.

 

I wonder, since you have a rig handling BVH, is it possible from there to bake a cycle, get rid of the Bvh link,

clean it up, match the first and last frame match and create a cycling walk action from there?

I am very interested in BVH, but had very poor success with linking the skeleton.

 

Michel

[Paul Forwood]

Looks pretty good to me, Mark, though I agree with Michel that for such a relaxed walk his arms could could relax more too.

[rusty]

Looks good Mark!

 

You took a realistic model for a start, those are hard, our eyes are experts with those guys.

Do you have 0 slope on the first and last frame, there seems to be a break in the cycle.

Is he walking along a path, or just cycling the action?, since there are no shadows or other reference

object, it's hard to tell. Anyway, I think it is nice to try on a path, this way it is easy to try a cycle at different speeds.

We build cycles with 24 frames most of the time, but everything is not happening in 1 second.

I think with this action, he could move faster since he seems to be putting some effort with his arms forward

and head rotating. Our forearms don't go very far in a relaxed walk.

 

I wonder, since you have a rig handling BVH, is it possible from there to bake a cycle, get rid of the Bvh link,

clean it up, match the first and last frame match and create a cycling walk action from there?

I am very interested in BVH, but had very poor success with linking the skeleton.

 

Michel

 

Michel,

 

BTW, I'm using the TS2 rig (but I've also linked the Squetch rig) and I have had very good luck in linking BVH files but this is for another post.

 

Rusty

[rusty]

mtpeak2,

 

Much better! Do check the zero slope on the first and last key frames in the cycle. There is a little too much jerk somewhere there.

 

Now, I think you've neglected to move the shoulders down and forward from the 'T' modeling pose the character is modeled and rigged in. The shoulders look good for the back swing but look unnatural for the rest of it. Does this make sense?

 

Also, as your foot/lower-leg swings forward, it does so in slow motion. Look at it. Don't you think so? The foot swings forward faster which gives you time for the following. Just after the push off (when the foot is back and which pushes the body forward) the foot/lower-leg rises upward before coming forward... and at the end of the swing, just before the heel touches down, the foot/lower-leg kicks forward. All of this is somewhat subtle and allows you to move the foot/lower-leg forward with more gusto.

 

Along with the above is some toe action which I can't get back to your videos to check on right now... you may or may not have it already. It is logical stuff though... in the back position before the push off the toe is bent lying flat on the ground ready to spring the body forward. As the foot lifts from the ground and rises upward the toe returns to straight again. Likewise in the forward position when the heel hits the ground, as the foot slams down the toe can raise up then flatten upon hitting the ground... this adds to the slam down effect which adds weight to the character.

 

My wife is attempting her first walk cycle right now and, at this point, is very discouraged. She is following the tutorial from the 'Character Performance' CD from Barry Zundel which is very good. I also have the tut on the Angevin CD Animate! also very good. In fact I have and have studied every walk cycle tut in existence (well OK, probably not, but every one I could find)! I've come to believe that I have accumulated too much on the subject LOL.

 

My two cents anyway.

 

Rusty

[agep]

Very nice progress Mark

[robcat2075]

One of the mentors at AnimationMentor said that when he ran out of ways to make a walk look weightier he would increase the time that both feet are on the ground.

 

I notice that you have the trailing foot lifting off right when the leading foot contacts the ground. That's a bit unlikely. If it stayed behind for a few more frames, pushing the body forward, I think that would help.

[mtpeak2]

Thanks everybody.

 

Rusty and Robert, thanks for the advice, I tried to apply them. I think I'm getting close.

 

Did I mention that I could probably rig faster than I can animate?

walk_side2.mov

walk_front2.mov

walk_birdseye2.mov

[nino banano]

Hi mtpeak2...I like the progress here , keep walking

[rusty]

Great job... a few tips to improve it... maybe LOL.

 

When the back foot pushes off it should go up and back (or, if not 'back', just a pause) -- you need to have the 'pushing off'. Also this push off causes the body to start to fall forward. You need a little more -- you're almost falling back in that part.

 

Also, before the forward foot's heel touches the floor, the foot should raise up a little before touching down.

 

Something many people miss is when the forward foot's heel hits the ground, there is no ease -- like the old bouncing ball it doesn't stop or slow down at all until it hits the ground.

 

The shoulders need to come down in the passing pose.

 

Also it sometimes helps if the hand swinging forward moves faster in the last half of the swing.

 

All of this stuff is subtile but needs to be there or at least a little more 'there' then it is now.

 

Finally, your character needs to move up and down a little more; high point is just after the push off and just before the passing pose (the 2nd highest)... low point is where the weight is transferred to the other foot (the knee should bend here as well as it takes the body weight).

 

Rusty

[Gaijin]

Something many people miss is when the forward foot's heel hits the ground, there is no ease -- like the old bouncing ball it doesn't stop or slow down at all until it hits the ground.

Another thing easily missed is that there's a 10% overlap between stride (foot in the air) and stand (foot touching the ground) - and that once a foot is in stand it doesn't move off the spot. This is especially difficult to do using the cyclic action in A:M.

[rusty]

Hi Gaijin,

 

Another thing easily missed is that there's a 10% overlap between stride (foot in the air) and stand (foot touching the ground)

 

The 10% overlap you mention is not clear to me and I don't think I know about this. Can you elaborate?

 

...and that once a foot is in stand it doesn't move off the spot. This is especially difficult to do using the cyclic action in A:M.

 

As far as I know, this is pretty easy to do in AM using the curves editor.

 

Rusty

[Gaijin]

The 10% overlap you mention is not clear to me and I don't think I know about this. Can you elaborate?

Human walk cycles are usually broken down into several distinct phases, with the top-level division being between when a foot is in contact with the ground and supporting weight, and when a foot is in the air and progressing to the next contact point. We'll label these

 

Stand (or Support) = foot in contact with ground

Stride (or Swing) = foot in the air, progressing to next contact.

 

In the process of one complete cycle, each leg progresses from Stand to Stride to Stand and does so in an alternating manner between left and right. All basic stuff - nothing new here.

 

The overlap I was referring to was the point when a leg switches from Stride to Stand, and the heel of the foot makes contact with the ground. At this point the other leg hasn't yet transitioned from Stand to Stride, so you have a section of each half-cycle where both legs are in Stand and neither leg is in Stride. This segment serves to transfer support weight from one side to the other and comprises approximately 10% to 12% of the overall cycle (if memory serves me correctly). The measurement of respective heel strikes gives you the length of the gait.

 

Once the foot contacts the ground on a "normal surface" - we'll ignore ice-covered lakes, sandy beaches, and other slippery or unstable walking surfaces - it does not "slide" along the ground relative to the bottom plane of the foot. Weight transference cause the Stand leg to "rock" from heel-contact to toe-pushoff in progression, but the overall foot position relative to its landing point does not change. Any amount of "slipage" in the foot's "grounded" position which isn't explained by the walking surface or overall circumstances is immediately recognizable and is what gives an animated character that unrealistic "video game glide" appearance.

 

In addition, the human walk is distinctly non-linear in its forward velocity, although inertia of the torso mass tends to dampen the variations to a degree. A:M's cyclic action was designed around repeatable movement with a CLV (constant linear velocity) along the path axis, and determines the current and any interpolated path position by using the primary model bone as a reference point. Since a walk cycle's reference really needs to be the Stand contact point and any interpolation should be performed using NLV motions defined by keyframes, this makes it very dificult - at least for me - to construct a decent walk cycle using the cyclic action.

[steve392]

Mark I found this very usefull ,I allso had a good walk one by a Woman but lost the link

cycles

edited found it ,the walk cycle tut

wa;k cycle tut

[rusty]

The 10% overlap you mention is not clear to me and I don't think I know about this. Can you elaborate?

Human walk cycles are usually broken down into several distinct phases, with the top-level division being between when a foot is in contact with the ground and supporting weight, and when a foot is in the air and progressing to the next contact point. We'll label these

 

Stand (or Support) = foot in contact with ground

Stride (or Swing) = foot in the air, progressing to next contact.

 

In the process of one complete cycle, each leg progresses from Stand to Stride to Stand and does so in an alternating manner between left and right. All basic stuff - nothing new here.

 

The overlap I was referring to was the point when a leg switches from Stride to Stand, and the heel of the foot makes contact with the ground. At this point the other leg hasn't yet transitioned from Stand to Stride, so you have a section of each half-cycle where both legs are in Stand and neither leg is in Stride. This segment serves to transfer support weight from one side to the other and comprises approximately 10% to 12% of the overall cycle (if memory serves me correctly). The measurement of respective heel strikes gives you the length of the gait.

 

Once the foot contacts the ground on a "normal surface" - we'll ignore ice-covered lakes, sandy beaches, and other slippery or unstable walking surfaces - it does not "slide" along the ground relative to the bottom plane of the foot. Weight transference cause the Stand leg to "rock" from heel-contact to toe-pushoff in progression, but the overall foot position relative to its landing point does not change. Any amount of "slipage" in the foot's "grounded" position which isn't explained by the walking surface or overall circumstances is immediately recognizable and is what gives an animated character that unrealistic "video game glide" appearance.

 

In addition, the human walk is distinctly non-linear in its forward velocity, although inertia of the torso mass tends to dampen the variations to a degree. A:M's cyclic action was designed around repeatable movement with a CLV (constant linear velocity) along the path axis, and determines the current and any interpolated path position by using the primary model bone as a reference point. Since a walk cycle's reference really needs to be the Stand contact point and any interpolation should be performed using NLV motions defined by keyframes, this makes it very dificult - at least for me - to construct a decent walk cycle using the cyclic action.

 

Thanks! I get it. Is there more information anywhere on 'human walk is distinctly non-linear in its forward velocity'. I'm not entirely convinced that AM can't handle this in an action cycle/stride-length that follows a path but I can imagine some potential problems. Anyway I'd like to know where this variation(s) in forward speed takes place and to what degree(s). Of course I can deduce this from video of walks by some number of people (male/female) but I'd rather not take the time right now.

 

Rusty

[Gaijin]

We're starting to delve into biometrics - here's a couple of "starter sites" which deal in gait analysis:

 

Introduction to the study of human walking

 

Clinical Gait Analysis

[rusty]

We're starting to delve into biometrics - here's a couple of "starter sites" which deal in gait analysis:

 

Introduction to the study of human walking

 

Clinical Gait Analysis

 

'We'? A class?

 

Thanks for the links (going to make me look for it huh... well if I have time)!

 

Cheers,

Rusty

 

Edit: Despite the fact that my work is realistic, this is more than I need (I think). If going this extra mile bought you a more realistic walk cycle I might get excited but I see no evidence to indicate such a thing and even if it were true, I do 'trailers' and these offer mostly quick glimpses that go by too quick for subtle tweaks to be noticed. BTW, do you have a walk cycle (or straight ahead walk) you can post -- I'd really like to see something.