I like Aku no Hana. I think it’s good*.
That asterisk is what you’ve come here to read about, isn’t it?
Our immediate reaction to pressing ‘Play’ is a question: are we watching an anime? We establish with our first shot, which provides us with a view of power lines and green hills. So far, so good. Birds chirp, the music swells; and yet, the artificial and high-pitched “anime voice” is nowhere to be found. Where is it, along with the voiceover narration? It’s just then that we cut, and to this:
What is that? We know that it’s human, but is it anime? It shambles forward. You recoil, but submit yourself to watching. You start skipping ahead to see if this isn’t some cruel joke. You close your media player. Whatever you do, you follow it up by making a comment on the public forum of your choice.
Ladies and gentlemen, welcome to the strange world of rotoscoping.
We recall that rotoscoping is a technique of animation whereby an animator traces over a recording of something’s movement in real life. This may be a person, or an object. The actual tracing can be accomplished in a number of ways — tracing over a magnified strip of film, tracing from the film itself, digitally tracing, allowing programs to run traces, et cetera. This technique is rare to encounter in Japanese animation; the last significant and excellent uses of it were in 2009’s Trapeze or the concert scene from Sakamichi no Apollon. Now Aku no Hana has brought this technique to the forefront of discussion.
What I found in Aku no Hana were two things I rarely feel watching anime, and never together: fear and disgust. For the longest time, we hear nothing but footsteps and piano as we watch our protagonist walk his way to school. Unconventional camera angles, the uncanny physical features of the characters, and extended close-ups strain the atmosphere. Carefully placed sounds and music push eeriness into dread. We can’t help but feel disgust at the characters — so close to being human, but monstrosities in their physical shortcomings. The poor technical quality of animation actually helps in this regard. They even try and sound like humans. One of these monsters has the gall to declare intellectual superiority over his fellow beasts, simply for his choice in literature. Disgusting. We’ve stepped fully into the uncanny valley.
I loved every second of both episode one and two. A beautifully subdued, naturally conveyed unnatural museum of base creatures not too different from us. Unexpectedly ominous.
I stand by the artistic decision to rotoscope. What I do not support, however, is how poorly the rotoscoping was executed. It shows a distinct lack of experience and laziness in the use of the technique. This post hopes to explain exactly where and why the show’s animation fails on a technical level. This will revolve around a technical examination of the principles of motion in animation.
Let’s address the question of frame rates. Contrary to the common misconception, every anime television show in recent memory has been shot in the digital standard of 24 frames per second (FPS). The only difference is in how those frames are used. Anime, given their limited budgets, tend to be animated on threes whenever possible. This means that each image is used three times consecutively every second, or that the image changes on every third frame. This produces an effective 8 frames per second.
This may seem to be an exercise in semantics, but this distinction is of phenomenal importance. If anime were truly shot in 8FPS, then fast motion would be almost impossible to portray. A car moving quickly against a road would appear to stutter. It’s because anime is shot in 24FPS that animators can switch between animation on threes, twos, and ones depending on intended movement. Threes are most often used for still scenes or cuts with little movement, such as plain dialogue; twos are used for moderate movement to fast movement; and ones, while extremely rare, are generally used for key cuts and movements requiring a great deal of fluidity. There are other, more artistic, uses of these varying ‘speeds’ of animation, but those are for another time.
Now, the question of artistic intent. I wrote a while back on the nature of the ‘camera’ in animation. Application of the anicamera lens does not reveal much. Largely so because the rotoscoping is almost certainly done via a computer program. This makes a judgment of the animation based on the inclusiveness of the frame unproductive. For example, take a look at this sequence. Can we, in good faith, say that the effect of the faceless individuals in the background are the results of good rotoscoping? Considering that facial features pop in at arbitrary distances, and that the popping-in may simply be an unintentional effect of the shader settings employed, may simply be nothing more than the product of sloppy and lazy work.
With some of this tangential information and observation out of the way, we can begin talking about the actual failure of Aku no Hana: the animation.
A point of notice about animated bodies is that they do not move unless they are willed (and drawn) to move. This is why having something animated on threes is an effective technique for anime: because movement can be dictated to fluidly coincide with frames. Remember: all things in the frame are intentionally created and intentionally manipulated.
The movement of living human beings isn’t nearly as simple. As still as one may believe they are, there’s likely a tiny portion of the body — a finger, a foot, a cheek, a shoulder — in motion. When human beings move, especially with clothes on, a large number of movements take place. For example, if one raises his/her arm while wearing a shirt: muscles on the neck and shoulder activate, the shoulder hem of the shirt crumples, the side of the shirt rises, and shoulder levels change. All of this happens smoothly. Aku no Hana‘s animation chooses not to take all this into account, instead opting to persistently animate on threes. The rotoscoping exacerbates this problem: our minds expect to see those secondary effects. This saves money, but results in a choppy and poor quality of animation as we skip the delicate motions in-between:
Had more effort been put into the rotoscoping, it would be possible to present a cleaner animation. In the case of the minutiae of movement required of speech, the results are interesting, on that border between recognizable and alien:
We do see tiny glimpses of hope, however. When we perceive motion, we do not see it in perfect frames. For example, quickly swipe your hand in front of your face. You know that’s your hand, but did you see your hand or a blur? Likely the latter. This blur compensates for the fact that we cannot capture motion perfectly. Animators can manually adjust for this (if animating on a slow count) by utilizing techniques like smears:
Obviously, this technique would not suit the atmosphere Aku no Hana is striving for. It has a set of alternatives, the most suitable of which is motion blur. In fact, the show does utilize the technique; in the second gif above; when the mother offers seconds, around the ten minute mark in the first episode; and when Kasuga stuffs the gym clothes into his shirt in the second episode. Sadly, these moments are sparse, inconsistent (some high movement parts don’t use it), and most likely unintentional. Yet, regardless of their frequency or intention, they’re effective. The frames still stutter and skip, but it gets slightly creeper, as it becomes slightly more relatable.
What we’re left with is a strange creature. On the one hand a tense, atmospheric work with moments of excellence that has so far proven rewarding, and on the other, the product of a wide array of seeming technical incompetence.
Are these so-called failures lazy? Do they end up enhancing the work? To both of these questions, I say, “absolutely.”
——notes and further readings——–
nil has written a post on Aku no Hana’s rotoscope. I highly suggest reading it, here.
1. [^] my guess would be that the recording was subjected to computer rotoscoping. Manually traced rotoscoping tends to produce a lot less movement (most minute movements are adjusted for) and more detail.
2. [^] in Western animation, rotoscoping is not as rare. Many of Disney’s animated features used rotoscoping extensively. There’s a rotoscoped feature-length movie titled A Scanner Darkly that I encourage everyone to watch. Waltz with Bashir takes the effect of rotoscoped animation, and expands upon it using a completely separate technique. These succeed where Aku no Hana does not because of various artistic choices made in their implementation. Also, I’m not sure if I’d say the Sakamichi no Apollon sequence was completely rotoscoped. It was most likely partially rotoscoped.
3. [^] don’t lie. You’ve done this before, probably sometime during high school. You probably hate that period of your life. While this is generally detestable regardless of rotoscoping, rotoscoping brings the animation that much closer to us, and such behavior becomes that much more identifiable. This is a master stroke on the show’s part. If you still do it now, well, shame on you. By the way, the authors I used to justify my horrible faux-intellectualism were William Faulkner and Vladmir Nabokov.
4. [^] a possible trick is elongating certain portions of the track the car is racing upon, to produce the illusion of constant speed.
5. [^] refer to note 1. above
6. [^] I don’t have gifs for these, sadly. People seem to be intent on picking out the worst examples of bad animation from the show.
EDIT1 (2013/04/16): clarified the wording on the paragraph about animation on threes, after recognizing how misleading some of the statements were. Included a link to nil’s post on the topic.
EDIT2 (2013/04/19): changed one of the links in “Contrary to common misconception”.