Maze Dev log – Week 2

February 20, 2026

Overview

This week focused on pushing the project forward from core systems into a more playable loop and a cleaner front-end experience. I expanded the decal painting system with color selection and improved HUD feedback, while also making major upgrades to the procedural maze flow—start/end placement, spawn handling, and better in-level generation control. Midweek, I moved toward a more modular content pipeline by converting the room template into a Blueprint for future variety. To tie it together, I implemented a start menu level and GameInstance-driven settings flow, including a working maze preview in the menu with a camera rail that adapts to maze size.

Major Accomplishments

Implemented a functional start menu flow (Level_Menu + GM_Menu + HUD_Menu + GI_Maze) to set maze size/seed, start game, quit, and generate random seed.
Added a menu “Preview Maze” feature that generates a preview using chosen settings and adjusts camera rail position based on maze size.
Upgraded the procedural maze system with automatic start placement, randomized exterior exit placement, and a guaranteed solvable path between start/end.
Expanded the decal painting system with a 0–9 color picker and HUD preview that updates based on selected color.
Converted the room template into a Blueprint to support future modular room variation and packaging into PCG modules.

Systems Developed or Updated

Decal Painting System

Allows placing decals with selectable type, rotation, color changes, and removal.
Added a color picker workflow: keys 0–9 map to assigned colors; selecting a color updates both the placed decal and the HUD preview.
HUD preview now auto-updates available decal color indicators (currently hard-coded to a 10-color set).
Added an “activated” boolean gate to simplify testing without removing the component.
Current state: Functional (needs polish for runtime color configurability).

Procedural Maze Generation (PCG Host + Maze Logic)

Generates a maze volume and ensures a valid path from start to exit.
Replaced manual start trigger placement with automatic selection of the cell closest to maze center (with a +350 Z offset to prevent floor clipping).
Exit placement now randomly selects from exterior perimeter points; removed dependence on trigger boxes for start/end logic.
Adjusted endpoint scale to avoid intersection issues that caused multiple exit openings on corners.
Player start now matches the maze start location; player is teleported to a spawned start marker once generation completes.
Implemented the maze as a Blueprint Actor to better accommodate runtime logic and settings.
Current state: Functional (workflow improvements pending).

Spawn / Load Handling (Dark Room Bootstrap)

Uses a “dark room” spawn area while the maze generates, then teleports the player to the generated start marker.
Works for now, but the transition needs improvement (input lock, reset timing, and ideally a fade/loading screen).
Current state: Prototype.

Modular Room Template Blueprint

Rebuilt the room template as a Blueprint to support future room variety (e.g., selectively hiding/removing walls).
Plan is to bake this Blueprint into a packed level and use that as a PCG module.
Current state: In Progress.

Material Color Variation Prototype

Created a proof-of-concept material setup to introduce slight tint variation across meshes sharing the same texture/material.
Intended to reduce repetition (especially on walls) without requiring many unique assets.
Current state: Prototype.

Menu System + Settings Pipeline

Added Level_Menu as the game’s entry map with GM_Menu and HUD_Menu.
Created GI_Maze to store selected maze settings (size, seed, etc.).
Added GM_Maze to pass the chosen seed/size to the maze host for generation.
Set project defaults: Game Default Map = Level_Menu, Default GameMode = GM_Menu, GameInstance = GI_Maze.
Implemented correct input mode switching (menu input vs in-game input) and cursor handling.
Current state: Functional.

Problems Encountered and Solutions

Problem: Decal removal trace wouldn’t hit decals reliably in the maze.
Cause: The custom trace channel defaulted to Block, so too many things (likely player capsule/mesh) blocked the trace before it reached the decal collision sphere.
Solution: Set the DecalPlayer trace channel default to Ignore, then explicitly set Block only on the decal collision sphere.

Problem: Maze PCG host wouldn’t generate after moving logic into a Blueprint Actor.
Cause: The actor lacked a “volume-like” component, so PCG had no usable bounds context (unlike a typical PCG Volume setup).
Solution: Added a collision/volume component to provide bounds so the PCG graph could generate as expected.

Problem: GameInstance casting logic broke construction-script/editor testing for the maze actor.
Cause: Casting to the GameInstance failed in contexts where the GameInstance isn’t created (editor construction script / non-play runtime context).
Solution: Avoided relying on GameInstance casting in construction-time logic and kept runtime-only access patterns for play sessions.

Current Project State

The project is in an early prototype / system implementation phase, but major building blocks are now working together: procedural maze generation with automatic start/end handling, a functional decal painting workflow, and a start menu that drives maze settings into gameplay. The menu preview feature is a strong step toward a polished front-end, and the move toward modular room Blueprints sets up future content variety. Some areas are still clearly prototype (dark-room bootstrapping, transitions, and editor/runtime workflow quirks), but the core loop is becoming more cohesive.

Next Week Plan

Improve the maze load transition: add input lock during generation and implement a fade/loading-style transition instead of relying on the dark room alone.
Refine the menu preview workflow and ensure runtime generation settings behave consistently (especially around partitioning for final builds).
Continue the modular room pipeline: bake Blueprint room templates into packed levels and validate them as PCG modules.
Decide whether decal colors should be runtime configurable (and if so, convert the hard-coded color set into editable data).
Tighten the maze host/editor workflow so construction-time previews don’t depend on runtime-only systems (GameInstance access, play-only data).