The Walk and the Whip
by Andrew Jaremko
(copyright 2005 Andrew Jaremko)
In the summer of 2004 I took the Introduction to Character Animation course at Quickdraw. One of the assignments was an animated walk. I (of course) immediately grabbed my copy of Richard Williams's "The Animator's Survival Kit" and started work. My walking robot eventually looked pretty good (I thought), but of course the exercise also made me see how much I didn't know.
I started watching people walking. Williams makes a big point later in his book about the importance of the "whip" action, though he doesn't really define a "whip". I started to think I could see some whip action in walks, specifically when the rear leg swings forward. I thought it was worth checking further, but didn't want to just trust my impressions.
I was working with some of my super-8 movies (a Disneyland trip in 1976) and saw that I had a shot of some people walking at right angles to the camera, with nothing in front of them. I grabbed the frames and made loops that I could look at.
By coincidence, this walk loops on 24 frames - a magic number for animators. It's a real coincidence that it does. My super-8 film is at 18 frames per second, so a normal two step per second walk would be 18 frames. People just stroll at Disneyland, so my subject took 2/3 of a second for a step - 12 frames. A full left-right cycle came out to 24 frames.
One of the first things I saw was that weight isn't transferred from one foot to the other instantly. Williams's "normal" walk drawings imply that it is. He starts with a drawing like this one, showing full weight on the back toes and no weight on the front heel.
I didn't rotoscope my frames, but I see the weight transfer starting with this position
and finishing, about 3 frames later, with this position.
It's very clear in the movie frames.
Alan Ferguson, looking at these frames with me at Quickdraw, pointed out that this is just one example walk and is slightly slower than normal. The weight transfer could take proportionately less time in a normal walk. We could doubtless find out a lot more if we just asked the sports physiology professors at the University of Calgary. Or looked in a technical library.
After that little sidetrack, I got back to looking for the whip action, and confirmed what I thought I was seeing.
While the toes are still planted and the rear heel is coming up, the rear leg is bending and the rear thigh is already moving forward. Once the leg has bent to an angle of about 120 degrees, the toes lift off. The thigh keeps swinging forward, but the knee stays at about 120 degrees until the thigh stops, like this:
The lower leg and foot swing forward and out once the thigh has stopped rotating.
This is the part of the walk that I see as a whip action. The thigh moves before the lower leg and foot, and drags them along. The thigh stops, and the lower leg keeps going until it "snaps" into position. Williams doesn't define a whip action, but leaves it to our intuition. I think whip actions happen with flexible objects (jointed or rope-like) that are anchored at one end and driven by forces that work from the point of attachment. Inertia delays motion in parts of the object further from the attachment. The details of the delays should give more or less feeling of weight.
And also (contrary to Williams) my film doesn't show the leading knee bending after it contacts the ground and starts receiving weight. Legs seem to stay straight all the time they carry weight.
Those last four drawings look suspiciously like a walk cycle. Here they are, in motion:
This was actually a pleasant surprise, since I hadn't planned these four drawings as a step cycle. The walk shows no weight whatsoever - it looks to me like the glide of a geisha.
Williams makes the point that, as animators, we are caricaturing motion and giving characters their own motions. And also that to do this, to show what's different about an individual, we need to know what the ordinary, common motion is. It seemed to me that even Williams's "normal" walk is a caricature. My film, of course, shows a "character-less" walk. It's an ordinary, casual stroll.
I decided to make a chart of the timing of what I could see in the walk. For the weight transfers, the chart looks like this:
A similar chart for Williams's "normal" walk might look like this:
This is a walk on the very edge of becoming a run. In a run, there are intervals when neither foot is in contact with the ground.
A different walk (say Daffy's as he enters the saloon in "Drip-Along Daffy") might look like this:
Here, both feet bear weight most of the time. Each leg moves forward for the next step very quickly. This is also the pattern I used for the robot walk I mentioned earlier.
This is giving me another way of thinking about animating walks, and might even help me "get my brains in my pencil" (to quote from Williams) once it really sinks in. I could start by putting a grid onto the walk chart. The grid would contain the number of frames I want to draw - say 8, for example.
This grid gives me the four positions I used in the example cycle. Because the leg timing is symmetrical, the leg positions in the second four frames are the same as in the first four. Things are different when I want an odd total number of frames. (Williams doesn't discuss odd total numbers at all.)
In this case, the leg positions in the second half of the walk cycle aren't the same as in the first. I tried drawing this cycle and came up with these leg positions.
And the walk looks like this.
A few comments about this one. I'm still learning where to put the body relative to the legs so it balances properly. I actually had to lean the body a little forward in these drawings to get it to look better. There's a bit of a limp - the body does come down a bit as the right foot lands but not when the left foot does. And with an odd number of frames, there's ambiguity about just where the weight goes in the overall cycle. All of which just reminds me how much more I have to learn.
I charted more than just the weight transfers, hoping it'll help me learn more about animating walks. Click on the link for a PDF file of my walk chart.
I also grabbed some leg bones from the CorelDraw clipart collection and animated them according to what I saw in my film. I think the bone walk turned out pretty well. Doing it helped me understand what goes on in the bones and joints, even though I didn't do anything with motion in the toes.
Click on the link for a PDF file
of the 24 frames in this walk.
Next time, I think I'll start with where the weight goes and then put the feet and legs in to support it. Happy walkies!
by Andrew Jaremko
copyright 2005 Andrew Jaremko