Introduction into the Secrets of Cutscenes
skizzo

24 Sep 2007 | 2 Comments

"Einweihung in die Geheimnisse der Cutscenes"
Originally written by Peter Kohlwein on September 1st, 2005.

This article will provide you with a peek into the range of the cutscenes and explain to you some of the basics of 3D-graphics.

In the beginning it was primitive...

The first step is to create basic models of 3D objects. These objects are created in several ways. Every object in our sequences, whether it is a plane or George’s head, exists of numerous polygons. A polygon is a plane figure that is enclosed on at least three sides. One side is exactly between two vertices (singular: Vertex). A vertex is a point in 3-dimensional space. But where does one start? I start my objects very primitively. A 3D-program can create a so called primitive (this can be for example a dice, a sphere, a cylinder, but also a simple surface). It creates a support for the modeller to design from this simple cube a complex and interesting object. The more detailed an object becomes, the more polygons it gets. The more polygons there are in a scene, the longer it takes to render the scene. It is therefore always important to find the right balance.

Bring out the wallpaper

Because every polygon object exists solely of simple grey materials (1), but since we don’t want to create a black-and-white movie, we have to change the surface of the objects accordingly. This we accomplish by applying textures. One can imagine a texture as a picture, which we glue onto the object. Textures are not only capable of changing the colours of an object (2), but it can also change the gleam (3), the transparency (4) and the drop (5) (the so called bump maps) of an object. Only with a combination of these effects will one get the desired effect.

Render until you drop

When all models are shaped, have textures and have lighting effects, then the both easiest and most frustrating aspect of the cutscenes commences: the rendering. Rendering is the creation of a 3D-scene from programming language, which requires many complex calculations. The result can be a 3D-scene, but also a 3D video. When all rendering components are set (for example: resolution, quality or file size) then one only has to push the ‘render’ button and wait… and wait, and wait. For one image of a Broken-Sword-2.5-scene the average rendering time is around 15 minutes. That may not sound like that long, but when you realize that one single second of video consists of 24 single images (frames) and that a sequence lasts for approximately 2 minutes, then you can quickly calculate the total rendering times: ¼ * 24 * 120 = 720 hours (which is 30 full days)! Often we cannot use options like ‘Final Gathering’ or ‘Global Illumination’ because that would greatly add to the rendering time. The biggest problem in this context is the fact that we have only one or two PC’s available which can handle the calculations for this. Large FX-studio’s are able to create render times of several thousands of hours per frame, because they connect an enormous number of PC’s to another to create so called ‘rendering farms’. In this way, each PC individually has a lot less calculations to make. It would be an immense job to render every single scene separately into a playable video sequence. This would take several hundreds of hours per scene.
Therefore every single frame is rendered into an image and all these are built together in a compositing program (we use After Effects). In After Effects we can then add several more effects, or adjust colour, scripting and accents.

The subject of 3D-graphics is so ample, that I haven’t mentioned everything by far. If this article has interested you, then look out for future articles which may talk about particle systems, effects, rigging, blend shapes, compositing, paint effects, UV-mapping, NURBS vs. polys, sub-Ds, soft and rigid bodies, etcetera, etcetera, etcetera…