Skip to content
Olivier Wervers edited this page May 9, 2021 · 2 revisions

There's a lot of terminology in discussions that talk about aspects that are very specific to RCT. This page aims to explain some of them to allow others to better understand what exactly is being discussed.

Sprite Rendering

Rendering Engines/Methods

There are a lot of different methods to render 3D models into 2D sprites. Some utilize ray-tracing or path-tracing, while others use rasterization or a hybrid between the two. The original graphics were rendered using the default scanline renderer present in 3ds Max.

Anti-aliasing

Some rendering methods leave jagged edges between transitions from one shape or surface to the other. Anti-aliasing aims to smooth out those edges either using post-processing or by sampling pixels in a larger radius than a single pixel during the rendering process.

Black on white line showing the difference between alias, and anti-alias

Anti-aliasing throughout the RCT2 graphics is inconsistent. Some graphics utilize it where others don't. Due to limitations at the time, it wasn't feasible to have an alpha channel. As a result, anti-aliasing on the outside of an object would sometimes result in a darker edge that looks a bit like an outline.

Hotdog stand clearly showing the dark outline created by anti-aliasing

Lighting Rig/Setup

The lights that are set up in the scene. The lighting rig is to some degree relatively consistent across the RCT graphics. From the shadows on some objects, we can discover the location of the main light. A second shadow can sometimes be found on large objects.

Quantization/Dithering

The RCT graphics use a small palette that the graphics adhere to. A quantization algorithm is used to turn an image with a less limited amount of colors into an image that only uses colors from the palette. There're different quantization algorithms available like Riemersma and Floyd Steinberg.

See this link to see how quantization algorithms produce different results: https://imgur.com/a/fFTF8q3

Isometric Projection

The sprites are rendered using an orthographic camera that is positioned in a way that shows the left and right sides of an object equally. The projection is close to an isometric projection where each side of a cube would have the exact same size when rendered to a 2D image. The RCT graphics use a projection that is optimized for computer graphics resulting in less jagged lines on the horizontal plane that we too call an isometric projection.