A softer light for long learning sessions.
Loading your preference…
Light uses Latte. Your choices stay in this browser.
The complete Parallax studio / six volumes
From your first pixel to the systems behind a renderer. Original lessons, small experiments, and six projects with a finish line.
50 chapters. Around two years at 4–6 hours a week, with room to repeat, rest, and finish.
Return to my current course ↗You only need one active project.CPU · macOS & Linux
Bring your existing C++ knowledge into modern practice through ownership, value semantics, memory budgets, geometry, light, triangles, and shaders. Small experiments become one finished rendering study.
OpenGL 3.3 · AMD Linux primary
Learn OpenGL directly in Parallax: context and resource lifetime, vertex input, shader interfaces, cameras, textures, lighting, render targets, and a scene of your own.
Portable C++ · macOS & Linux
Build a CPU path tracer by understanding probability, material scattering, light sampling, acceleration, and convergence. Measure error before chasing a faster image.
C++ concurrency · macOS & Linux
Give a renderer dependable boundaries: bounded work, ownership, cancellation, asset generations, allocation budgets, profiling, and reproducible releases.
Vulkan · AMD CachyOS
Learn Vulkan through a small explicit renderer: devices, queues, presentation, memory, descriptors, pipelines, synchronization, compute, and evidence from a captured frame.
Metal · M4 Pro macOS
Build the Metal path in C++ with metal-cpp and a narrow platform boundary. Study MSL, storage modes, passes, completion, compute, and Xcode GPU tools.
Each chapter teaches the idea here, then asks you to predict, build, investigate a failure, and explain a decision. References such as LearnOpenGL are further reading, not a replacement for your Parallax lesson.
Professional habits run through every course: explicit ownership, bounded work, tested assumptions, clear failure paths, and measured performance. Native companions are focused references; your course projects grow through your own implementations.
Try the frame-lifetime experiment ↗
