Making of: "Frucosiadas"
As promised, here is the "behind the scenes" special post of "Frucosiadas" - enjoy!
Idea...
I have been toying with the idea of 3D character animation for quite some time now. Lately I started using my 3D application of choice (Hash's Animation:Master) a bit more often again - initially intended to provide some backdrops for an upcoming pawpet project (which is still in the works, don't fret!) but the idea of moving and animating a character became too prominent to ignore. So I sat back one day and thought about how I could create something that is somewhat new, easy to animate and doesn't take up weeks of render time... little did I know...
My original spec was something around this: "Simple geometry for characters, restricted movement, little to no facial features yet cute to look at when animated. Scenery non-existant, i.e. almost pure white background and only key props or items important to identify certain characters are present in a clear cut color scheme. Done in episodes of about one minute with one key joke each."
A little youtube video - shown on the Funday Pawpet Show - brought with it the idea for the first episode: a hamster running in a wheel while a second hamster was just holding on to the wheel and spinning around in it...
I introduced the idea to yamavu at that time and for lack of a better description I referred to the "simple shape" as "something like a cough drop"...
Only marginally later (two days at most) I started modelling and discovered that a circular or square shape wouldn't do for almost any of the planned movements... that's when the "starfish" shape came into play.
Modelling
That part was actually as fast and easy as planned... I did have some experience with the modeller of Animation:Master and so the starfish was finished in a few hours. Including several slight changes in proportion.
Rigging
For those of you who do not know the term: "rigging" is when you put the animation "skeleton" into your character and tie it to the modelled geometry...
Following advice that's been repeatedly given on the Animation:Master forum I didn't create my own "rig" but used a pre-made skeleton... The setup was quite complicated and didn't work at all at first. So: redo. It worked a bit better but after animating the "walk up to the wheel and surprised bounce" I had to admit: THAT rig wasn't going to cope with the weird geometry. The legs couldn't be moved in a reliable way and every slight little change resulted in a complete repositioning in all three dimensions...
So, again: re-do. A much simpler skeleton was my next aproach and it did work like a charm from the first try on. Woot! But I had to scratch and re-do all the animation too. Awwww.
Animation
I quickly came to realize that the most complicated of all things in animation is the correct timing. Sure it's not that hard to move a few keyframes around but in one continuous shot every slight change on one end means lots of problems on the other. So I shot some reference footage of myself walking, in fact acting out the part of the "active" drops - well, without the wheel of course :P...
Another stupidity in 3D animation: no collision detecteion whatsoever. I had to render the last few seconds three times over because either the drop in the wheel would move through the spokes or the pillow would stick in the ground... These things are hard to see in wireframe/shaded mode and did in fact only turn out in the fully rendered output.
To give you an idea of how many keyframes we are talking about, look at this image...
Lighting
Seriously: I need to learn and experiment more with lighting... both in 3D and RL. 'nuff said!
Rendering
Finally: let's create images! But wait! What options to make my little project look nice? Let's try from top down, I thought. So: Use an OpenEXR sequence as output format and include every buffer that Animation:Master would provide (Light/Shadow buffers for every light source, Depth, Ambience) and render it in crisp HDTV format: 1920*1080*25p. Ummm... renders on frame every half hour or so? Not good. One frame is over 20MB? Also not good. Trying to combine the EXR files using After Effects proved even more challanging: took almost a minute to create a single frame from all these different buffers...
A combination of OpenEXR with Depth Buffer only gave me a nice compromise between quality and speed. Render times (tripe-pass renders) were between 1 and 2 minutes per frame depending on complexity (the idea-lightbulb was the time killer).
Music
I knew from the get-go that I would need some spiffy music to go with the animation, so I asked a musician I came across only a few weeks earlier if he would be interested to provide a simple and catchy tune. I send preview (flat shaded, tiny resolution) renders to him for timing reference and he came up with that jazzy piece that I am very grateful for! Thanks again, Luchs!
Sound effects
This time I took my digital recorder on a field trip around the house... creating squeaky noises all over the place, ringing my bike's bell and knocking on the plumbing in the cellar instead of using pre-made sound effects... :)
Finishing touches
As I was already editing the final version, I came across a ustream session of nanook123 where she drew some sketch commissions... I thought: Hey, a nice little logo would be fun to go with that project and asked her if she could manage one on such short notice. Kudos for pulling it of: I sent the specs/$$$ the same day and had my logo the next :)
Problems...
As good as Premiere and AfterEffects work, the "nifty feature" called "dynamic link" either is completely useless or it just doesn't like me. Either way, my carefully crafted combination of Premiere for the edit and AfterEffects for the preprocessing of the EXR images (levels mostly) failed repeatedly during render and took again two seconds per frame. Not good.
I dumped "dynamic link" and went back to the old way of pre-rendering the AfterEffects compositions and using that video as footage in Premiere. Worked much faster and - even better - without any crash.
Of course the problems with the dynamic link made the time of day quite early and I completely missed the fact that the "youtube widescreen" preset of the Media Encoder uses 4:3 resolution. D'Oh! Now I have a video online that has an utterly ugly black bar running around the image... But if I want to change that, I would lose all of the (only 70 yet??) views, so I'll keep it like it is. It doesn't look *that* bad.
Package
A nice little Skype chat with yamavu gave me the idea of a little extra that I could put up with the video: a papercraft version of the Frucosiadas' package! The art on the package was finished anyway so it would "only" take a few extra lines to create a package, right? Right! Only "few" proved a "few many" lines. The result is still available here...