Skill-based 3D generation: 68k to 153k tokens but far more realistic leaves — skills set expectations, not teach
chaseleantj · x · 2026-09-13
Key numbers from the author's 3D scene generation test: with the same prompt, a one-shot run without a skill took 18 minutes and 68k output tokens; with a skill, 77 minutes and 153k tokens — but the leaves look far more realistic.
He explains the point of a skill today: it sets expectations rather than teaching the model how to do things — e.g. targeting Awwwards-level website quality, using the strongest appropriate rendering techniques. More on writing skills in future posts.
Related event: Setting expectations via skills key to better AI 3D generation(2 posts)→
More from coding & agent
- Element Forge: open-source editable AI VFX for real footage in ComfyUI — zitroffell · 2026-10-30
- Gym-Anything Adds Audit Agents That Try to Break Simulations With Legal Robot Actions — wellecks · 2026-10-30
- CMU's Gym-Anything Auto-Builds 200+ Agent Environments from Any Software, Selected by GDP — wellecks · 2026-10-30
- durable-object-migrator: zero-downtime exports for Cloudflare Durable Objects — samgoodwin89 · 2026-10-30
- This 5-question interview prompt makes Grok design your first two custom bots in minutes — nateliason · 2026-10-30
- Cursor and Harvey swap general models for task-specific brains, but most niches can't afford to — MartinGTobias · 2026-10-30