i think about this cut by ryo-timo from The Girl Who Leapt Through Time a bunch, since it’s a really brilliant bit of character acting on top of a run cycle. I haven’t seen the film (though I plan to screen it on a future Animation Night!) but regardless I thought it would be helpful to try and break it down and figure out how it does what it does.
so first of all: the character animation overlays a run, which is what I’m interested in at the moment. the run cycle is shot from the shoulders up, so we’re selling the run through the motion of the shoulders. in a classic run cycle, the shoulders mirror the hips, so when the left hip is backwards, the left shoulder is forwards, and vice versa. in this case we don’t know what the legs are doing but because we’re familiar with this motion, we can easily read a run from the motion of the shoulders and hands, the effect of wind on the hair, and the scrolling background (representing a tracking shot).
according to Williams’s book, a run cycle typically sits in the range of 6-8 frames per step, shot on 1s. ryo-timo’s animation actually has a couple of hold frames in there but otherwise it is indeed mostly on 1s. if this is frame 1:
then we get to the mirror image of this position on frame 9:
frames 3-4 and frames 6-7 are held:
however they do not obviously seem like held frames because the shadow continues to move! there’s some pretty abrupt movement from frame 4 to frame 5:
and likewise from frame 7 to frame 8:
rather than drawing smears, there’s a bit of motion blur on the hands.
so there’s actually very little movement during the passing position (when the arms are crossing over the body). the shoulders almost have a three frame loop! with minor variations inside that. yet in practice, it reads extremely clearly as a lively run.
the motion of the arms would probably look very exaggerated on a full figure, but it works here because we’re shooting from the shoulders up so it has to actually get into the frame. it’s also noticeable that the drawings suggest she’s swinging her arms inwards as much as she is swinging them forwards, angling the hand towards/away from the camera, which I think helps to sell the shoulder motion. (might be worth borrowing that).
anyway over the first couple of seconds, she’s doing this loop, allowing us to see a gradual change each time she hits the key pose. here’s frame 1:
here’s frame 162 from around the 6 second mark. the arm at the back is going up higher, but the arm at the front is lower. the arc of her eyebrow has inverted, her neck is slightly less tilted, and the angle of her collar suggests she’s leaning less far forwards. the first frame is driving forwards, this one is pulling back a bit:
by frame 180 (another few steps) she’s gotten tired, her eyes droop:
everything about her face has closed up here, her hand is rising even less high on the extremes (which actually implies quite a bit of side to side motion! but it looks like it’s sinking down). even the folds in her uniform seem a little less energetic. is she gonna give up?
no way! here’s frame 196, where she’s sprung back up to a pose much like frame 1, but with her teeth gritted. she’s actually sunk her shoulders down a bit in the frame here: her run has gotten tighter and more focused. the drawing of the clothes has also gotten looser.
she starts to rally, takes a breath, and her arm swings get much more energetic again. here’s frame 306, about 12 seconds in:
her hand’s almost going up to her nose, and the arm at the back is going similarly high! I notice that now, her hands are in shot for more frames, and the passing positions go by in just a couple of frames. thus while it still takes about 8 frames per step (down 1 frame from the start), the arm movement feels faster because it takes up less of the cycle.
but she can’t keep up this pace for long. this time, rather than closing her eyes, she starts to wobble around a bit: this is less of a controlled run than a desperate scramble. here’s frame 429, where you catch a glimpse of her right eye as her face turns towards the camera:
at the same time, she’s sliding backwards in the frame. she continues to make these big motions, sweating heavily, but she can’t keep pace with the camera and disappears off to the right. she’s gone from frame 521 to about frame 585 (about three seconds).
but then she comes back in, running all the more furiously than before. here’s frame 653:
what’s different here? first of all, she’s leaning into the run much more. secondly, she’s grimacing with the strain: her eyes are squeezed shut, her teeth are gritted and her mouth is really wide. thirdly, she’s now balled her fists. fourthly, the sweat droplets are now rolling over her face and dropping behind.
what about timing? still 8 frames! she’s not actually making more steps per second, but it feels faster because of the pose. it’s all in the acting! (I would need to make some measurements to figure out if the camera has sped up at this point).
then around frame 780, her pose starts to gradually change again: she’s leaning backwards, still in that snarl expression, but I think it communicates that she’s hit the pace she wants. at the same time, she passes out from the buildings in the background and hits the open air. because the background is much further away, it’s scrolling less fast. (there are two layers of parallax: the trees and the city).
as she continues to run, she ungrits her teeth and takes a deep breath, and opens her eyes, though still looking very Determined:
she basically continues to hold this expression for the rest of the animation as she overtakes the camera. it feels exhilarating.
here’s frame 1 for contrast. you can see there’s actually a subtle zoom out over the course of the animation.
a lot of these frames are very similar, but by these subtle variations in the poses, and that really clever use of the camera motion, ryo-timo is able to pull off a really powerful bit of wordless storytelling. hope the rest of the movie can live up to this lol
one thing I notice is that there’s basically no use of smear frames, although some of the drawings towards the middle are noticeably rougher. instead, there’s a small amount of blur applied to the hands on frames outside the key poses. since the focus is on the face acting, this is probably where most of the attention has gone - we’re not staring at the hand unless we’re making a long tumblr post picking it apart :p
also, regarding the background: it’s very blurred to help sell the motion, but it goes through a succession of dark and light shapes, with occasional figures visible. less often, foreground objects such as cars pass by as well. she starts the sequence in shadow, and by the end of it she’s in bright light. another gradual transition which helps sell the story.