youtube-opus-gpt-benchmarks
Installation
SKILL.md
YouTube Claude Opus 5.5 & GPT-6.1 Sol Motion Graphics Benchmarks (youtube-opus-gpt-benchmarks)
Codifies the highest-performing programmatic motion graphics patterns demonstrated in Claude Opus 5.5, GPT-6.1 Sol, and Claude Code + Remotion/HyperFrames showcases.
1. The 5 Benchmark Archetypes (Pre-Built in remotion/src/engine/threeSceneStudio.tsx)
Archetype 1 — 3D Cylindrical Rotating Icon / Card Carousel (<CylindricalIconCarousel3D />)
- Distributes $N$ cards around a 3D cylinder of radius $R$: $$x_i = R \sin(\theta_i + \omega f), \quad z_i = R \cos(\theta_i + \omega f)$$
- Sorts cards back-to-front by $z_i$ (Painter's Algorithm) on every frame.
- Modulates depth-of-field blur (
blurPx) and specular border glow by the facing normal $\frac{\cos(\theta_i + \omega f) + 1}{2}$.
Archetype 2 — 3D Sliding Photo / Metric Stack (<SlidingPhotoStack3D />)
- Stacks cards in 3D depth (
translateY(-depth * 22px) translateZ(-depth * 45px) scale(1 - depth * 0.055)). - Every
cycleIntervalFrames, the front card peels off on a spring arc (Math.sin(p * Math.PI) * 280px) and settles at the back of the stack while all remaining cards advance one slot forward.