Ring of Fire - An Overview, by Michael 
Wolf:
"Ring of Fire for us started in the middle of 1997.The producer Michael Jungfleisch and the artist and director Andreas Hykade had this wild idea in their head, and were looking for ways to create a new look for their cartoon.
So they consulted with me and we made a couple of
preliminary tests that turned out to be just right. I opted to use
Lightwave instead of a 2D compositing solution because I have much more
flexibility animating and the real-time OpenGL preview is a real
time-saver as well.
The film was to be produced in 2K Cinemascope. The workflow was as follows: According to dopesheets traditional (pencil) key-drawings were made and then in-betweened. Once the results on the linetester were satisfying, the paper & pencil drawing went to the artwork department,where every drawing was reworked on to cells using a very work intensive technique by first inking the cell and then scraping of the highlights using a small knife (well, it is a very unique look, and that look was one of the reasons why we couldn't use something like an Animo for animation).
Now we had 30GB full of single frame 8 bit .TGA's, RLE compressed (actually 2 images per cell, on containing the cell and one with the clipmap, which had to be 8bit as well !).. Luckily most of the images compressed really well, but there still a couple that wouldn't go below the original 40MB on disk. The real problem would come at rendering time anyhow, since the images loaded in Lightwave would definitely take the original amount of memory.
So the next step for me was to write a plug-in to help with the image sequences (you can look at more plug-insby the same author at http://www.gadget.de/lw/lwgadget.htm). I got the
dopesheets from the director and soon found out that renaming all the
images so that they would load properly was out of the question. The
plug-in I wrote basically allows me to transfer the dopesheets to a list
of frames and spews out that list as a new sequence of dummy files
containing the image name of the original file. The I programed a loader
that loads the dummy files, retrieves the original image file an loads
that. Since I had this loader anyhow, I modified it so that it would
create smaller proxy files that were loaded instead of the big images.
This saved a lot of time when loading scenes and doing
previews.
The cells were mapped on simple polygons with a matching clipmap, the ground is usually a plane as well and we sometimes used hemispheres for the sky. In one scene we actually built a room out of flat textures that were given to me (A bit like hand-drawn UV mapping). Using key visuals and the dopesheets supplied by the animator / director we were able to layout the scenes up to the first test render (in PAL 720x306).Then I would sit down with the director and tweak the scenes. First we'd finish up the placement of props (some scenes have more than a hundred characters) and then tweak the animation (mainly the position of characters and the camera).
When all shots were complete we went over all of them again to adjust the lighting. (Funny thing, lighting. At the start of the project lighting was only supposed to be used in one or two shots, but the director got so excited about the possibilities that we ended up lighting almost every scene.)
At this point the film was finished in PAL resolution and ready for editing (basically shaving off a couple of frames here and there) and the music / soundfx. We had to start with the rendering (That was this January) since the film was going to be premiered at the end of March.
So we quickly bought 8 dual Celeron 466 512 MB machines as a renderfarm (it was the best price performance ratio at the time) and went away rendering (using LightNet). 2 Months and 22400 frames in 2K (I think it was 1800x1600) later.....
"Making of Parkhaus Werbespot" hier