Convolutional Layers in Tile Generation

[BenGhahramaniUQ]

Description

Task: Convolutional Layers in Tile Generation. Feature: Procedural Map Generation with Convolutional Layers and Chunk-Based Spawning The goal of this task is to integrate convolutional layers to enhance map tile generation. This approach will serve as a more advanced alternative to the wave function collapse algorithm. The convolutional layers will assess adjacent tiles and use a weighted sum approach to determine the most appropriate tile for each position, ensuring smoother transitions and cohesion across tiles.

Objectives

• Implement convolutional layers for map generation.
• Consider surrounding tiles in the decision-making process for selecting the next tile.
• Use a weighted sum algorithm to guarantee that adjacent tiles align correctly, leading to seamless tile placement.

Example

• Map Expansion: When the map is expanded or a new section is loaded, the convolutional layer algorithm will run, ensuring that new tiles align with previously loaded tiles.

Dependencies

This task will require the below tasks to also be completed.

• [x] Dependency 1: Chunk based spawning (#231)
• [x] Dependency 2: Wave function generation (#81 )

Milestones

List of steps that need to be completed for this task.

• [x] Step 1: Implement initial skeleton. (Sep. 6)
• [x] Step 2: Code imlplemented with some functionality. (Sep. 9)
• [x] Step 3: Final testing. (Sep. 11)

Completion Deadline: Sep. 12

Documentation

• Main description of feature
• [User documentation](https://github.com/UQcsse3200/2024-studio-2/wiki/Map)

Members

BenGhahramaniUQ (@BenGhahramaniUQ) Howie (@Howard-12) Praneet (@praneetdhoolia) JonnaHerman (@JonnaHerman) Chloe Catane (@chloemicah)

[praneetdhoolia]

Convolutional layers not fully incorporated (change in work plan).