Siggraph starts

siggraph_16.jpg

We are in San Diego !
Siggraph started for us today, with a brilliant presentation called “Anyone Can Cook Inside Ratatouille’s Kitchen”, by the nice folks from Pixar.

What made this entire morning session so interesting was it’s focused discussion on just producing the food of Pixar’s Ratatouille. There was no real discussion about animation, or character rigging, – it was all about how the team made the fine (and not so fine) food in the latest Pixar film.

While there were a few people talking the two speaker who just shone in our humble opinion were Athena Xenakis and Stefan Gronsky who talked about Shading and Lighting.

This was the exceptional sort of high quality presentation that makes Siggraph the must attend it event it is for serious cg and film professionals. It was just brilliant.

Xenakis walked through the use of Sub-surface scattering or just ‘Scatting’. Many papers have been presented on Sub surface scattering, but this paper was on the application of it to produce beautiful renderings of food.
The irradiance point cloud generated by the scattering algorithm is stored vai way of a voxel grid. What was interesting was also the way the team used a weighted vertex cluster to shape the scatter. In the case of Bread this was applied to a volumetric hyper texture density function – which provided the air bubble structure of freshly baked bread. We will be learning more when we interview the senior renderman team this week here at siggraph for fxguidetv.

Gronsky’s following talk on lighting was even more fascinating. It build on the shaders by examining the lighting used in the film. Gronsky started by detailed analysis of real food photography and what makes a photograph convey taste. This discussion alone could have made a not half bad photography course, but it was of course followed by applying these principles to 3D rendering.
The discussion centered on 4 points:
1. The quality of light – soft area lights
2. Translucency – back lighting and rim lighting
3. Surface texture – how specular highlights are broken up by textures, and how vital reflections are to moist/wetness.
4. Colour – which denotes ripeness and freshness – especially the level of saturated colour

He then discussed in brilliant detail how each of these were address, how the team used fake global illumination, and built on the work of the shader team.
Gronsky’s team of 38 lighters lit 1700 shots.
Some of the most interesting technical discussion was around the relationship between Scattering and Pixar’s Gummy lights. Pixar adopted Scatter lights some time ago, after Pixar software engineer Brad West saw Wann Jensen’s 2001 SIGGRAPH demo. Gummy lights were developed around the time of Finding Nemo and build on some concepts dating back to Siggraph 2000 Deep Shadow paper on self shadowing in clouds and hair. Gummy lights effective are lights which are transmissive but affect different wavelengths of light differently. In others white light falling on and thus in to a block of cheese, shifts to yellow then orange, giving the otherwise colourless cheese its CG colour. While Scattering gives it a luminance quality, literally as if it is subtly glowing inside. While Scattering has a blur value the Gummy lights really pushed the CG from good to great.

Using the example of Cheese, bread and wine, Gronsky showed the same plate of food with and without Scattering and Gummy lights. This was followed by backlighting vs rim lighting on grapes and even how negatively playing with the principles discussed were used to make soup look undesirable for a scene before a recipe is saved. And then once fixed how the lighting team made the second version of the soup look so much more inviting. This particular discussion extended to the floor and pot lighting and was truly illuminating !!!