A third-person sci-fi shooter prototype with enemy AI — built and taught step by step in a complete playlist.
This project is a third-person shooter prototype built entirely in Blueprints with Unreal Engine 4, set in a Martian environment where the player controls a marine and must eliminate all enemies to complete the level. The development process was recorded in a full step-by-step teaching playlist.
You can watch the complete teaching playlist here: YouTube Playlist
You can also watch a demo of the finished prototype here: YouTube
Third-Person Shooter Foundations
The third-person perspective positions the camera behind and above the player character, keeping the character visible throughout gameplay. This perspective is architecturally distinct from first-person: the player character is a visible presence in the scene, which means character animation — movement, combat, hit reactions — is a significant part of the visual experience. A first-person character can get away with minimal body animation since the player rarely sees themselves; a third-person character needs animation for every gameplay state.
The camera setup uses a spring arm component that maintains a configurable distance and angle from the character, automatically adjusting to avoid geometry clipping. The character rotates to face the movement direction — a standard third-person movement model — or the aim direction during combat, depending on whether the player is in exploration or combat mode.
Enemy AI: Clear and Win Condition
The game loop is structured around a clear condition: eliminate all enemies in the level. This is tracked by a counter decremented each time an enemy dies. When the counter reaches zero, the level complete state triggers. The simplicity of this win condition is appropriate for the scope — it’s clean, unambiguous, and creates a progression arc within each level without requiring complex game mode logic.
The enemy AI uses Unreal’s Behavior Tree system for its core behavior loop: patrol waypoints while the player is undetected, switch to pursuit when the player enters detection range, attack when within combat range. The Martian setting gives the enemies a specific visual character — alien or robotic adversaries that move differently from human enemies — which allows the animation set to be more stylized and less constrained by human locomotion conventions.
Detection uses a sphere overlap combined with a line-of-sight check — the same two-stage detection model used across all the AI projects in this series. The enemy can hear the player (wider detection radius, no line of sight required) and see the player (narrower cone, requires clear line of sight). This layered detection produces encounters where approaching from cover and staying quiet are both viable tactics, even in a simple prototype.
Setting and Atmosphere
The Martian setting shapes the environment design: red rock terrain, industrial military structures, low atmospheric haze that reads as the thin Martian atmosphere. The setting also justifies the combat scenario — a marine eliminating threats on a hostile planet — without requiring narrative setup. The environment communicates the premise instantly.
Post-process effects contribute to the Martian atmosphere: a warm red-orange color grade for the exterior terrain, cooler industrial tones inside structures, and a subtle haze that extends the apparent depth of the outdoor areas. These are the same post-process techniques applied in the Cavern Demo and First Person Platformer, applied here to a sci-fi exterior context.
Teaching the Project
The teaching playlist documents the construction of the prototype from empty project to finished game. For a third-person shooter with enemy AI, the teaching sequence naturally follows the dependency chain of the systems: character movement and camera first, then animation, then the weapon system, then the enemy AI that needs all of the previous systems to be in place before it can interact with them.
This sequential structure — building systems in order of dependency, with each system building on the previous — is a good teaching practice that mirrors sound development practice. It prevents the situation where the teacher (or developer) is referencing systems that haven’t been built yet, keeping each step self-contained and testable.
Reflection
The Mars Marine Game is one of the earliest complete game prototypes in this series — a functional third-person shooter with a clear win condition, enemy AI, and a coherent setting, built early in the Unreal learning journey. Comparing it to the more sophisticated combat projects that came later — the Melee Combat System, the Zombie FPS, the Bad Bot prototype — makes the progression visible: the same fundamental systems (character movement, enemy AI, win condition) appear at progressively greater architectural complexity as the technical vocabulary expands.
The sci-fi setting is also worth noting as a design choice. Genre settings aren’t just aesthetic — they define the interaction vocabulary, the visual language, and the player’s expectations. A Martian marine game has different design constraints than a zombie apocalypse or a fantasy dungeon, and those constraints shape decisions about enemy behavior, environment design, and game feel that are embedded in the project from the start.
Nicolás Bertoa 
Leave a comment