@whitewhoadie — 12x breakout selling Snowflow

Loading video…

🚨 9 HOURS + 4M CLAUDE TOKENS = A FULL WEBGPU SNOW ENGINE

@whitewhoadie

33.8K followers · usually 21.9K views a post

Views

263.7K

Likes

7.0K

Comments

275

667

View velocity

+39.4K views over 28.8d · ~1.4K/day · 7 observations

Promoted product

Snowflow

A real-time procedural snow rendering demo built with WebGPU, Babylon.js and hand-written WGSL. Features GPU-generated terrain, snow deformation, procedural characters, cloth, surf wakes, water spe...

Why it worked

The video likely performed well due to its impressive visual demonstration of a complex technical achievement (a full WebGPU snow engine) created with AI, combined with a clear and concise explanation of the process and a direct link to the project's repository.

Summary

A developer spent 9 hours and 4 million Claude tokens to create "Snowflow," a browser-based Web GPU demo featuring realistic snow physics and terrain deformation. This project, which evolved from a technical design document rather than a prompt, showcases the increasing importance of detailed specifications in AI-driven development.

Structure

1Introduction of Snowflow and its creation process
2Explanation of the technical details and resources used
3Comparison to previous plugin versions
4

Original caption

🚨 9 HOURS + 4M CLAUDE TOKENS = A FULL WEBGPU SNOW ENGINE SNOWFLOW is a browser demo with persistent deformable snow, waterbending spells that reshape terrain, and snow-surfing with dynamic wakes. Ships with: • Custom snow shaders (subsurface scattering, glints, compression) • 90 FPS targets, zero allocations in the render loop • Cloth, particles, atmospheric lighting The crazy part: it started with a full technical design document, not a vague prompt. Visual quality was the product. 👉 https://github.com/Noniv/snowflow_demo Who’s writing specs instead of prompts today? Drop a 🔥