March 11sec Part 2/4 – PreProduction

So this is Part 2/4, focusing on my Pre-Production work for the 11second club March competition. You can catch part 1 here.
Or just to remind you what we’re talking about, you can check out the final product again.

Character:
This project consisted of two characters – the Director and the Rock ‘n’ Roll Guy (who I will call the Rocker). For both characters I used the Norman Rig. Norman is not a free rig, but if you don’t have access to him there are plenty of free rigs out there that are highly customizable as well. I would recommend the resource page of the 11 second club, or the Animation Buffet blog, that is meant to keep track of ALL the free rigs out there… man are there a lot!

The Rocker required a bit of extra customization.
Here are the sketches again for a rough idea of where I was headed:

So first things first – the makeup. I quickly painted some black makeup in photoshop according to the UV layout from Maya . If you remember from part 1, I wanted the top part of the mask to ride the brow line. This would really accentuate the brow shapes, and also allow for the ‘pointy’ parts of the makeup to be angled outward and downward (as opposed to up like KISS), giving him an overall sad and pathetic look.

When I was going to make a first attempt at applying this texture, it wouldn’t have done me much good to check it on only the defualt neutral face pose. I need to see how it deforms across the face when pushing and pulling his controls, because that’s the way the audience is going to see him. So I made a quick and dirty facial test, and then applied the rough texture to it:


I was super happy that I did this test, because I noticed something that needed to be fixed right away. When the character closes his eyes you loose all connection with him. The whole middle of his face becomes a giant black puddle with no geography markers. I checked out how Mr. Incredible’s mask was handled, and it seemed to me that either they never fully closed the eyes (leaving a little sliver of lighter color showing through), or a tiny rim of skin showed up on the lower lids when he blinked or shut his eyes. Either way, you never loose the geography in the face. So I went back in and added a sliver of skin tone on the lower lids (if I used white like the rest of the face it would blend into the whites of his eyes, and also feel pretty fake). Now when I reapply that to the face test, it becomes much more readable during the last pose:

On that test I had also refined and cleaned up the black brow lines and the lips to get a clearer readability. Admittedly, during the test animation I closed the lids too much (the geometry was crashing), and the slice of skin color ALMOST disappeared, but I took note of that for how much to close the lids for the actual animation.Now that I had gotten the face pretty close to what I wanted, I moved onto the hair. For modeling I roughed it out with a cube, then just started wildly pulling vertices everywhere. Then I parented it to the head and this is what I got:


Personally, I think the hair feels way too rigid. It basically feels like it’s just stiff geometry parented to the head (and it is). Every rotation the head makes, the hair follows perfectly. I wanted to change that, so I built a very quick rig to help break up the rigidity of the hair. I don’t want to turn this into a post specifically about rigging, but I do want to cover it briefly – so I’ll try to be quick.

Hair Rig: I made a new joint tree centered from the head, with joints outside on the left, right, and back of the head. I parent-constrained the root of that to the head joint. So far (if I bound the hair to this joint chain) I would get the exact same effect as before, but this is a means to a better end. Next, I added 3 separated ‘dangling’ chains at each outside joint, and only POINT CONSTRAINED them to their respective counterparts. This way only their translation, not their rotation, is defined by what the skull is doing, and they will stay ‘dangling.’ Also, their rotation channels are free in case I want to animate them for overlap on the hair. (side note: I would never animate rotation values on the joints themselves, its harder to break the rig – and much cleaner – if you make null groups that control those for you… but thats a whole different discussion). I made a quick diagram of this added hair rig to try to explain this better for those that are more visually oriented:

(I left out the chain for the hair on the back of the head for the sake of clarity)

After this was set up, I bound the hair geometry to the new joint chains and weight-painted accordingly. Now there is a slightly looser feeling to the hair in the facial test. It’s subtle, but its there. Pay special attention to the first head move, and you’ll see the hair doesn’t rigidly follow the skull anymore.


A different type of rigging setup was called for when it came to the guitar – getting it to follow the body, and the left hand to stay on the neck, while keeping all my channels open for layering animation on top of everything. But I won’t go into that. That’s enough rigging talk for now, right?


The Location:
According to my plan (storyboards can be seen in part 1), I need a couple of main things to make this scenario work. Mainly a stage and curtains, a staircase to the stage, a music stand and a sheet of music. I’m not really going to go into modeling talk. Mostly because I suck at modeling, but I am going to touch on a few things I think are important when it comes to set pieces in general.

Most Importantly, I only put in what was absolutely necessary. So why the drums? I felt it was needed to add some weight to the left side composition i n shot 1 (and right side weight in shot 3). Why the mic stand? Again, for another compositional purpose because without it the Music stand becomes too much of a focal point in shot 3. If it had tons of negative space surrounding it, it would stand out far too much, and you would be drawn to it.

But here’s the thing – I didn’t spend any unnecessary time with the set pieces. The music stand is simply a bunch of cubes, stretched, scaled, and rotated to make the general shape. The mic stand is just a cylinder with a half-flattened sphere at the base. The guitar and drum kit models I got from my friend Blake, and I actually spent a little bit of time REMOVING detail from them. Yes, removing detail. There were little nuts and bolts on the drum kit, all the tiny parts of the guitar, etc. If you let too much detail get on your set-pieces, you run a real risk of your shots becoming incredibly cluttered and your props will compete for the audience’s attention. Not good! (unless that’s your intention for the shot). So don’t waste all your time modeling useless details into your background set pieces.

Another thing I think really helps keep your set believable is asymmetry – even if it’s subtle. It’s very easy in the computer to keep things perfectly parallel and perpendicular. It just feels incredibly… well.. computer generated. Fake. Put things a little off-axis. Angle things a bit more or less than 90 degrees. Keep it interesting. Check out the Kitchen Floor in Ratatouille. It’s a tile floor, so it would be easy to make all the angles absolutely perfect. But none of them are. There are NO perfect squares (from what I saw). They are all slightly askew, tiny imperfections in the angles, small inconsistencies which make it feel less perfect and less computer generated.

I tried to incorporate that theory in my set. For example, lets take the stairs (staircases, like tile floors, can very easily be created too perfectly). None of the steps are parallel, nor the railings, and neither of those things are exactly perpendicular to the other.

==============================================================

Well that’s pretty much it for my pre-production work. Part 3 will focus on Layout and Camera-work. I got lots of wonderful and supportive comments about the ‘documentary’ feel to the piece, and I’d love to share some concepts that helped me achieve the look.

-Jacob