The Pixel Crush

-------------------------------------------|Digital Animation & Game Criticism|-------------------------------------------

Thursday, 28 February 2013

Pixel Knitting: I

Which came first, the shader or the model? One of pixelosophy's most enduring conundrums.
Because without the shader the model is just points in space, but without the model the shader is merely material properties- yet to take form. I wonder if there's a niche market for poetry whose target audience is the philosophical CG artist?
-----X-----

When I started working for Aardman Digital on championsheeps I had just finished herosquad with them, championsheeps was in its prototype phase, development proper was beginning and there was a demand for assets. There was a desire to push the look of the games to a different/'nother level by using CG and rendering to 2D images, otherwise known as "sprites". Much the same as had been proven to work well before but on a much larger scale and spread over five games in a suite.
The first thing to be done was to take the existing Shaun model, ready it for rigging by making it bipdel (two legged), and begin look development on the fleece. I knew I wanted the wool to absorb the light the way it did in the real stop motion model. So I started with this ugly looking combination of an SSS shader and Vray's fur plugin.

A really quick way to completely slow up Vray, our renderer of choice, is to ask it to scatter some soft shadows through thousands of small hairs of geometry.
I casually string these renders together as though this process took a few tweaks and I spent no more than a few minutes on it, but it in fact kept me busy for a good few days. This is partly because in order for a project's look to be consistent all the shaders must be tested in the same lighting set up so as to ensure they will react to any given scenes lighting in a predictable way that allows all the objects to sit together. This helps prevent different materials over or under exposing, or appearing more or less reflective than they actually are.
Click this image for super enormous extra fluffy vision
I later revisited the wool shader and more closely matched the hue of the real Shaun, you can also see below that the lighting has been refined and the cove more fully surrounds the subject, as well as not appearing overly bright. I had some great reference photos from the model making department (real glitter and pritt stick [those are professional tools right?] model making) that showed their cove was a slightly creamy colour so I mimicked that.
In order for Shaun to bleet he needed some basic mouth shapes with different poses and positions. One of the show's creators even came and explained how they stuck on the mouths and lent us his personal Shaun model. Which we definitely didn't examine with slightly too much enthusiasm, with alarming consequences.
I had also by  this point added a slight bumpiness to the plasticine shader to get the little dirt indents and fingerprint marks.
Next up was Generic Sheep, who was basically Shaun with different proportions, though simply scaling the body had some undesirable effects on the contours of the fleece.
This weird fleece was flagged in one of our daily stand up sessions that happened every morning, we each brought the team up to speed on our accomplishments from the previous day, then outlined our intentions for the current day, and what any problems might be.
I actually didn't think this solution would work because its so simple. Make a sphere, extrude all the faces (turning the "keep faces together" option off in the channel box, and cover in wool easy. Except at this point I wasn't aware of this option until Nathan revealed its existence. So, like a fool, I selected alternating faces across the entire mesh like a chessboard and extruded those before inverting the selection and doing the same.
After some more shader tweaking and adding in a rim light I managed to perfect that glowey halo of fuzz round the edge of the sheep's fleece, its my favourite thing :) So fluffy.
With the sheep's look coming together it was time to prepare them for rigging. A back and forth process that gets the model just right for the skeleton, deformers, and various controls to be applied. My first taste of a process that Id better understand after completing my first character from scratch, and then later understand even more on the commercial we're currently working on. Still learning new parts of that process. A lot of lattice and blend shape stuff specifically.

Its nice to finally be able to post this stuff now that the games are out. Play it HERE! I finished my part of these games months ago, November I think.
I'm proud of the final games, though the strength of their design and playability has nothing to do with me, they look pretty great and the designers did an awesome job of making the CG, flash, and 2D backgrounds look cohesive and charming. This post is the first of several, hence the numerical title.

The digital team are an utterly fantastic bunch of people, one example of this being the personalised badges we were all given. Me and Jake had a stab at coining the trendy new term "custom", hence my badge declaring that I was just that! I wore this amazing thing every day for ages.
"custom" spread the word
 And then I sneakily crept into that magic part of the company photo where
the lens distortion makes me look around 200% more beefy than in reality.
rubbing shoulders
As these blog posts go on and I learn more I feel less and less qualified to tell people "how this thing should be done" or "how this technique works". Maybe that'll change over over the course of the next few posts, on this project.

No comments:

Post a Comment

Let The Discussion Begin: