March 12, 2026
Cos? Sin? Nah—bring the drama
Avoiding Trigonometry (2013)
Bye-bye angles? Coders say use vectors; comments spark math wars
TLDR: A veteran dev says 3D graphics should avoid angle math and use vector methods, claiming cleaner, less error-prone code. Comments split: some cheer the simplicity, others say it’s old news (Rodrigues’ formula), while a few go full radical with “no angles ever.” It matters for faster, sturdier graphics code.
A 2013 hot take just resurfaced: a graphics guru says if you see sine, cosine, or angle math deep inside a 3D engine, it’s a red flag—use vector tricks instead. Translation for non-nerds: stop juggling angles, start moving arrows. The crowd went wild. One fan, srean, cheered, calling it a personal “moment of enlightenment,” while others joked that every acos() call was “kitten murder” (yes, the post said that). Meanwhile, chriswarbo pushed the needle further with Rational Trigonometry, a math philosophy that ditches angles entirely and sticks to ratios—cue the “Angles are canceled” memes and links to Divine Proportions. Not everyone was impressed. xeonmc shrugged, saying this is basically Rodrigues’ rotation formula—as in, nothing new to see here. storus chimed in with a cool-headed “wouldn’t a change of basis fix this?” sparking a mini war between Team Angle and Team Vector. And djmips dropped a vibe-check link—finish your derivations—calling for less hand-wavy math and more show-your-work. The mood: half “finally!” and half “we already had that,” with memes yelling “sin/cos are sus” and purists clutching their unit circles.
Key Points
- •The article argues against using trigonometric functions deep inside 3D graphics algorithms, except at input stages.
- •Dot and cross products encode the necessary orientation information typically accessed via sine and cosine.
- •An example axis-angle rotation routine demonstrates a common pattern relying on sin and cos to build a rotation matrix.
- •A typical alignment workflow (using acos of the dot product, cross for axis, and normalization) is presented as suboptimal.
- •The author contends that staying in vector space avoids unnecessary complexity and numerical issues, though alternatives are not fully detailed in the excerpt.