Live action reference vs. drawn from memory

[slowtiger]

The mentioning of "Chico and Rita" raised a question: how much "reference" can one use until it's not animation any more? I made a list of "typical" use cases, roughly sorted by amount of usage:

• 1. Studying life action, animate from imagination.
  2. Studying video of life action, frame by frame
  3. Using life action video as reference for animation
  4. Several Disney dance scenes: rotoscoped by hand, either completely, or as reference for animation
  5. "Chico and Rita", life action, rotoscoped by hand (about 2 frames per second) as reference for animation
  6. "Out of the Inkwell", aka Koko the Clown (Fleischer): life action, rotoscoped by hand (each frame)
  7. "Lord of the Rings" (Bakshi): life action, rotoscoped by hand (each frame)
  8. "Alois Nebel": life action, rotoscoped by hand with vector shapes (with software tweening)
  9. "Waking Life" and "A scanner darkly": life action, rotoscoped mostly automatically with vector shapes
  10. "Take on me" (A-HA), numerous music videos and TV ads: life action, rotoscoped by hand, using labour-intensive drawing techniques
  11. Gianluigi Toccafondo: painting/drawing on printouts of life action video, expanding and altering the images
  12. Jeff Scher: rotoscoping with watercolour on paper
  13. "Yellow Submarine" Lucy in the Sky sequence: rough rotoscope of iconic dance scenes

I'm not interested in any official definition of "animation" - the Academy might have to deal with that, but not me. I only care about the resulting film, and wether I like it or not. After 100 years of animation we can choose between lots of different styles of movement, from cartoony to realistic to experimental, so using a naturalistic style should be just one choice among others. But is that so?

I'd like to keep my own comments to the list above (and more) until you had a chance to answer. Any thoughts?

[Paul Fierlinger]

You mean you really want to hear mine?

[slowtiger]

Well of course! At least it will define one extreme point of the scale ...

[idragosani]

I personally have no objection to using live action reference and some rotoscoping techniques. Depends on the effect you are trying to achieve, I guess. Some might just use selected key images to get poses and then do the traditional animation from that (which apparently is how Chico & Rita was done). Bakshi has stated he chose rotoscoping & and his "painted film" look for "Lord of the Rings" because the film was too big to draw completely by hand and would have taken forever, and knowing Tolkien's distaste for Disney, wanted a visual style that was darkly realistic and staying away from the Disney look. If you've seen any documentaries on Bakshi's stuff from the 70s, they show that some exaggeration is added to the animation to give it more life and they aren't doing line-for-line tracing on everything.

I think you must use some live action reference, even if you aren't rotoscoping, unless you are doing something that is really flat and stylized (or doing crappy Flash animation).

[Paul Fierlinger]

Well of course! At least it will define one extreme point of the scale ...

Do you lack that much personal conviction that you must start your own provocative thread without voicing a clear opinion of your own? You are already prejudging me by assuming my view will be on the extreme point of the scale, which is condescending and gives me little motivation to help you out.

[Elodie]

Hey Paul, don't be so bitter ! Markus is only asking for opinions from everybody, not only yours.

He is free to start a debate about a subject without voicing his opinon. Probably Markus started this thread in particular because he doesn't know what to think about and would enjoy to have your feedback (and other ones).

[ematecki]

My opinion is : as long as nobody notice, you're fine.
If not, you just became a human plugin for your favorite software... soon to be replaced by a real plugin !

[Paul Fierlinger]

Probably Markus started this thread in particular because he doesn't know what to think about and would enjoy to have your feedback (and other ones).

Really???

I'd like to keep my own comments to the list above (and more) until you had a chance to answer. Any thoughts?

[slowtiger]

(sorry for delay, back from real life now.) Point taken. Fact is, I feel quite undecided about the topic.

I've used to frown upon #10 mostly because it lacked personal input from the animators. Most times it's just a paint job on moving objcts. And I thought it was mainly a waste of time, possible only with a big budget (as ad agencies tended to have) and, as a job, only good enough to pay the rent. (This includes all examples of time-consuming "animation" techniques like printing or drawing frames on any odd object, or cut them into cardboard, or stop-motion Lego bricks as pixels and so on.)

But #11, 12, 13 are examples I like. Although it's the same technique I find them more satisfying to the eye: with Scher, it's the (mostly) surprising choice of tool and colour in every frame which I find delightful. With Toccafondo, it's obviously the parts he adds while painting.

#9 I felt disappointed about. Linklater's direction reduced all possibilities of animation to those of Photoshop or video-filtered actors.

Strangely enough I liked #8 (I've only seen the trailer so far) although its basically Linklater's approach. It is interesting to compare the choices (b/w, comic-like appearance, a political theme) with "Persepolis" which was animated freely. I like to refer to this as "certain degrees of alienation" (as in "Verfremdungseffekt") which might be necessary to connect an audience to a different level of story/character than it would do without. ("Waltz with Bashir" would be another fine example: it wouldn't work as life action,it would have a totally different impact.)

I have another sweet spot of interest in the range between #3 and #7. Bakshi's attempt was an interesting failure: while he still used the design conventions of 70's animation the introduction of small involuntary movements (breathing, turn of body, hand gestures) was something new. Koko the clown doesn't look rotoscoped at all, instead it features his very own blend of fluid motion, fully rooted in the animation style of its time. Disney's rotoscoped dance scenes don't fit into the surrounding films.

I'm planning something with realistically proportioned characters within a realistic view of the world (this is part of the story), but I feel somehow uncomfortable to do scenes as life action first. Of course my animation's perspective and motion would benefit from that, but I just don't like that workflow. A Disney level animator could produce wunderfully natural looking animation from scratch - but how long will it take? A decision for using life action footage is always an economical as well as artistical: with video footage even a team of not so skilled animators could produce a result with a certain degree of naturalism. Also it will make it easier to incorporate "modern" camera angles into animation which would be hard to do the traditional (read: from imagination and construction) way.

So I wonder which techniques you'll find "legitimate":
- sculpt maquettes of characters to help animation
- models of props, scenes
- BGs constructed in CGI software as reference for drawings
- BGs produced in CGI software
- do blocking in a 3D scenery, to imitate camera/lens attributes
- do blocking with actors
- have actors play difficult scenes, film it to study
- film everything with actors first

I'm really undecided about all this. And "Don't care about naturalism" isn't an answer here, because parts of that project need to be "looking real" in order to make the story (and the other part) work.

[Paul Fierlinger]

I would prefer a technique you haven't listed, which is to use still photography (posing actors or incidental snapshots or both) to emphasize "reality" and separately create good, freehand impressionistic animation of stylized characters to connect and give the still photos emotions (must not be cartoons). But I am not suggesting the animation be superimposed upon the stills -- each element can stand on it's own while maintaining a solid graphics feel for the whole.

[Peter Wassink]

I would prefer a technique you haven't listed, which is to use still photography (posing actors or incidental snapshots or both) to emphasize "reality" and separately create good, freehand impressionistic animation of stylized characters to connect and give the still photos emotions (must not be cartoons). But I am not suggesting the animation be superimposed upon the stills -- each element can stand on it's own while maintaining a solid graphics feel for the whole.

interesting thread.

Paul i can't quite picture what you mean. Would the photos be visible in the end product or only used to create the drawings?

Another great example of using of life action material to create animation is Tilby & Forbis "When the day breaks"
It avoids the pitfall of the slavish copying of source footage.

I suspect that this is what makes most rotoscoping so boring, every movement is copied, both the important meaningful ones but also the accidental trivial ones.
The process of tracing all this motion makes all frames equal in importance, the drawings will resemble realistic people but what you loose in the process is the deliberate 'storytelling' power an animator can put in his poses, dynamics and holds.

to get to your question:

The mentioning of "Chico and Rita" raised a question: how much "reference" can one use until it's not animation any more? "

all of your examples i would still call animation.
You can take the list a lot further nowadays, for instance what about a film like Avatar? i would consider that an animation film too.
Every case where you add elements to the image of a film by hand in a frame by frame method i would call animation.

[slowtiger]

I left out "Avatar" and the like on purpose because motion capture is something completly different: it's 3D data, used in another 3D environment.

But rotoscoping is definitely a 2D technique, used in 2D space. I think this difference is important, as we animators have a completely different approach to indicate depth in a line drawing, and we use all that even when using live action reference. Motion capture and CGI work in a 3D space, and all rendering, how many cartoon shaders you might use, will always be automatic, "perfect" to the algorithm, but often not as good as drawn by a human.

I think this is the core of the (my?) problem: there's an audience who wants to see a certain style of camera angles (or camera angles I like to use because they serve my story), which makes it harder to construct perspectives and volumes by hand. It's not so much about the human poses: these I can draw, given enough time. And I will most likely draw them within a good staging, so the drawing will "read" immediately. Only that this staging may conflict with perspective of background.

(This is not a final opinion of mine. I'm right in the process of thinking this over.)

[Peter Wassink]

I think this is the core of the (my?) problem: there's an audience who wants to see a certain style of camera angles (or camera angles I like to use because they serve my story), which makes it harder to construct perspectives and volumes by hand. It's not so much about the human poses: these I can draw, given enough time. And I will most likely draw them within a good staging, so the drawing will "read" immediately. Only that this staging may conflict with perspective of background.

So what is your problem?
Drawing realistic figures in perspective?

[Paul Fierlinger]

Peter,
Sandra and I are at the moment producing 20 minutes of animation to be inserted in various parts of a 90 minute feature documentary, which we are rendering in the way I tried to describe above. I am not permitted to show any of our work but basically the live action is combined with animation by cutting the live stuff in the usual way you would if there was no animation, but then you insert between two clips a piece of animation (every once in awhile; not exactly checker-boarding the edit).

But here's how it works: Firstly, the live action is shot in a flat, 2D sort of way (1 and 1/2 D?) by locking the camera in place with no camera moves and no motion in the scene with the exception of small things like a leaf twirling in the wind or a distant piece of trash moving down a street. It's like a still photo shot with video; very mesmerizing, by the way. Then this clip is followed by our animation of the same sort of setup, but in our own way. I look at the live action scene once, put it out of the way and draw it as I felt it and Sandra paints it as she sees fit, which is quite removed from reality. What this creates is two views of the same experience; real and perceived.

One more important aspect of this setup is that there are very few camera moves in anything. When the live action script calls for a dolly shot of a woman walking down a street, the dolly shot is replaced by jump cuts; each cut jumps a little further down the street and there are no zooms in the cuts. In between those jump cuts is inserted our animation clips, again depicting the same street with no camera moves but in ours, there is a woman walking away from the camera as if shot from afar with a long lens -- she's walking but not discernibly making progress.

The overall visual effect is that of a very flat view of the world; like a series of pictures, whether photographed or painted. It makes the combination of live action with animation work together because the live action parts don't overpower the animation, they are shot "flat" to match the animation yet there is always motion in both.

I see that slowtiger posted before I could finish typing this and I believe that we are talking about the same problem with my addition of having solved it.