Arterra

Arterra

Algorithms, Ideas, and Thoughts

Noise Generation

Overview

At the core of nearly all procedural terrain generation is noise. Not to be confused with audio/signal noise, noise refers to the randomness, or more specifically pseudorandomness, that allows for the generation of new content. Whereas most functions adhere to generating predictable patterns expected by users, randomness purposefully attempts to subvert expectations by creating unpredictable outcomes. While achieving unpredictability may simply involve increasing layers of arbitrary complexity, natural terrain isn’t defined by absolute randomness. Rather, a level of predictability is commonplace; one would expect rough rocky terrain along mountain ridges and smooth sandy deserts. To capture such patterns without predefining terrain requires a balance between randomness and predictability.

Structure Placement

Overview

Procedural generation implies the production of new content based off of certain factors not unilaterally explicitly indicated by any individual. As computers are inherently deterministic systems, it follows that the sole way to produce undefined outcomes is to introduce randomness or noise into generation. This law is ubiquitous, notwithstanding of procedural terrain generation but found in all computer-related fields, notably machine-learning. As such, procedural terrain generation most commonly refers to the derivation of surface-terrain from a series of noise maps which dictate specific features of production.

Structure Pruning

Overview

Procedural generation implies the production of new content based off of certain factors not unilaterally explicitly indicated by any individual. As computers are inherently deterministic systems, it follows that the sole way to produce undefined outcomes is to introduce randomness or noise into generation. This law is ubiquitous, notwithstanding of procedural terrain generation but found in all computer-related fields, notably machine-learning. As such, procedural terrain generation most commonly refers to the derivation of surface-terrain from a series of noise maps which dictate specific features of production.

Structure Planning

Overview

Procedural generation implies the production of new content based off of certain factors not unilaterally explicitly indicated by any individual. As computers are inherently deterministic systems, it follows that the sole way to produce undefined outcomes is to introduce randomness or noise into generation. This law is ubiquitous, notwithstanding of procedural terrain generation but found in all computer-related fields, notably machine-learning. As such, procedural terrain generation most commonly refers to the derivation of surface-terrain from a series of noise maps which dictate specific features of production.

Mesh Generation

Overview

On-demand 3D terrain generation in games is traditionally accomplished through several methods. One is voxelization which quantifies terrain as regularly sized voxels(cubes) and the other is polygonization which quantifies terrain as combinations of 2D polygons.

avatar
Blackmagic919
Jonathan Liu