Each drawing was sliced into hundreds of pieces which were moved in relation to one another, thus creating the illusion of movement. The film was first shot in a sound studio as a 90-minute video and then transferred to a storyboard. From there 2,300 original illustrations were drawn based on the storyboard, which together formed the actual film scenes using Flash animation, classic animation, and 3D technologies
Ive posted an example of how a a shot is animated using Flash CS4
An extremely thorough & explanative interview with Director of Animation Yoni Goodman about how waltz with Bashir was created
How long was the animation process and how many people worked on the animation?
The animation process took 2-2.5 years. We began with six animators (including myself) and got up to 10 animators. So, the whole crew consisted of about 25 people more or less
What animation techniques did you employ to create such vivid imagery?
We used a Flash cutouts animation technique specially designed for this project for all parts of the animation, sometimes slightly adding traditional animation to the process. we also used very few shots using 3d, but only for camera movements and such. There was absolutely no Rotoscope animation throughout the entire project. Every time someone mentions rotoscope I get angry calls from my animators in the middle of the night.
Tell us about the whole rotoscoping confusion…
Well, Waltz with Bashir has 0% rotoscoping in it. We keep fighting the rumor that we rotoscoped, as many people compared our process to Linklater’s A Scanner Darkly and Waking life, but the whole movie is Flash cutouts. We did have live reference of the interviewees, but we just used it to learn the movements. There was absolutely no tracing over a video.
David has this great dark, graphic novel, contoured look which we really wanted to use, so I had to come up with a technique that we could work with, and still refrain from doing it in traditional animation. Traditional animation features typically employ 200-300 people. There are the keyframers and inbetweeners and colorists and cleanup artists, but we just don’t have enough experienced manpower in Israel, and no budget to outsource it, so we used cutouts instead.
Every drawing made by David was sliced into many pieces in Flash, sometimes counting to 400 pieces per character, and they were set up in different levels of hierarchy. So like normal cutouts in Flash, we split a character into primary pieces – head, torso, limbs etc.. Each of these was symbolized, and inside each symbol we had another set of layers containing the parts it was made of. This complexity didn’t allow us to draw much of anything from scratch, as we were maintaining David’s original line. Eventually was done using 10 animators (including myself).
Once an illustrator had finished the artwork for a particular scene, explain the process of preparing that layout for animation.
Once an illustration was complete (mostly by David, he illustrated something like 80% of the movie), the image was imported to Flash, traced and broken into pieces. The sliced pieces would then be converted to symbols and set in their appropriate hierarchies. Then we would place the sound and the animatic inside the file (after the animatic was done, we sliced the sound according to the shot’s length, that way we knew when it was supposed to end in the FLA). The animatic was placed as reference for timing and acting in a guide layer, and then the animation process began. The first part, moving the main symbols, requires some skill, because things never look completely right – joints are usually breaking and the movement looks very stiff.
Once the main symbols are done, we went into the secondary hierarchy to animate each symbol and compensate for the stiffness, creating a smoother look for the animation. Usually you work in and out, back and forth – sometimes correcting the main symbols after you are done with the inner hierarchy, then going inside and adjusting the new positions. It sounds terribly complicated, and in a way it really is, but still, to get that very fine clean movement, it’s a lot faster (and easier to correct) then traditional animation.
At the end of it all, I think there were less than 10 percent of the film was traditionally animated. The only scene that was fully created using traditional animation was the waltz scene.
It sounds like very little CGI animation was employed in the film, but many sequences appear to have so much depth. Do people often mistake the animation for CGI?
A few month ago I went to Annecy film festival, where Waltz with Bashir was the opening movie, and in my speech, as part of my ongoing campaign to deny the rotoscope rumors, I said “this movie has no rotoscoping in it” (my animators asked me to say it). Later on, someone told me there was a rumour that it was all done in CGI. So much for fighting rumours. We had very few shots (10 or so) in which we used Maya, but that was mainly for camera movements. There was no actual modeling, except for a ship and a few houses in the snow which you can see with a very good telescope.
I think most combinations between 2D and 3D animation don’t work so well, and we really tried to make the 3d environment as flat looking as we possibly could. We just placed the Flash animated sequences on flat planes in the 3D environment and moved a camera through them so we could get a depth-of-field feeling. We also got to a point where people mistook very complex Flash cutouts for CGI. For instance, the scene where the tanks are squashing cars, but these parts were all done in Flash. We didn’t do animation in any other software (unless you regard the explosions as animation. These were done in After Effects).
How else did you use After Effects on the project?
Every shot in the movie was exported as PNG files and rebuilt in Adobe After Effects. The smoke, explosions and simple camera movements in the movie are all done in AFX. We also used grains and filters to avoid the vector look Flash usually gives.
With all of the various pieces that went into each element, did the software ever strain under the weight of your production?
As we used highly-detailed symbols with lots of inner-hierarchies, the more you got into the nested timelines, the heavier the FLA would become. Some work files got to a point where they could not be opened on a computer with less than 4 GB of RAM. This is actually the main reason why we only worked with one shot per FLA, and no shot over 40 seconds per FLA, otherwise it would crash the software or make it impossible to work with.
Flash is a sort of a blessing and a curse all at once. It’s a format I find very handy, very easy to produce high quality animation, and it’s very easy to correct and fix animations (once you learn how to build the file correctly), but at the same time, it still has the “internet animation software” attitude to it. I always feel like I’m abusing the software by doing things it wasn’t designed to do, and it’s a shame, because I feel it can do so much more. I’ve been using Flash since version 4, and it doesn’t seem like there have been many improvements in the broadcast field. I actually think Flash MX was much better at handling heavy files and high-rez exports than any other version that came after it.
I have many ideas, and I also read many thoughts (in your site as well) how to improve Flash in that aspect. I think Flash needs to evolve into a broadcast-quality-specific animation software, perhaps as a separate “Flash-based” software that uses all of it’s features but is strong enough to handle the strain. Ari Folman once tried to approach Macromedia (they were the owners then) after we finished “the material that love is made of” and the pitch to present some of our thoughts but they sort of snubbed him out of the building. I would be really happy if they decided to improve the software in that direction.
No comments:
Post a Comment