What 3D Renderers Solve

A 3D renderer turns a scene description (geometry + materials + lights + camera) into a 2D image.

Scene + Camera → Renderer → 2D Image

Two main approaches:

1. Rasterization (real-time graphics):

• For each triangle, figure out which pixels it covers.
• Compute color per pixel.
• Standard for games, GPUs.

2. Ray tracing (offline / modern hybrid):

• For each pixel, cast a ray into the scene.
• Find what it hits.
• Compute color (recursive for reflections).
• Standard for films, modern AAA hybrid (RTX).

Real engines:

• OpenGL / Vulkan / DirectX / Metal: GPU APIs.
• Unity / Unreal: full engine, abstracts API.
• Blender Cycles: production ray tracer.
• PBRT (Physically Based Ray Tracing): research engine + textbook.
• Three.js: browser 3D.
• Bevy: Rust + WGSL.

We'll cover the math + key algorithms. Note: full image rendering is hard to test via stdin/stdout (binary pixel comparison fragile). We test:

• Vector + matrix math.
• Ray-shape intersections.
• Triangle setup.
• Per-ray color calculations.
• Geometric transformations.

You can build the renderer alongside; tests verify the building blocks.

Reference: Real-Time Rendering (Akenine-Möller), Physically Based Rendering (Pharr), Scratchapixel.com.

Key concepts:

• Vertex: a 3D point.
• Triangle: 3 vertices.
• Mesh: collection of triangles.
• Material: how surfaces respond to light.
• Light: a light source.
• Camera: viewpoint + projection.
• Frame buffer: 2D pixel array.