Two 45-minute loops with a critic: how GPT-6 Astra iterates on 3D reconstruction
bilawalsidhu · x · 2026-09-05
Bilawal Sidhu shared workflow details of his GPT-6 Astra 3D reconstruction run: the task ran for two loops of roughly 45 minutes each, with a critic agent evaluating outputs against multi-view imagery of the input 3D scan. He planned to let it run overnight to see how much closer it could get to the source scan.
The post is a concrete look at a generate-and-critique multi-loop agent workflow applied to 3D scene reconstruction.
Related event: GPT-6 Astra Rebuilds a Real 3D Scan into a Blender Scene from Scratch(6 posts)→
More from coding & agent
- Why LLMs dissolve the long-standing barriers to formally verified software — teortaxesTex · 2026-10-23
- Row-Bot 5.1 connects local AI to 100+ apps and 40,000 MCP servers — Acceptable-Object390 · 2026-10-23
- Agent evals should test metric definition drift, argues dev in new post — jylusdev · 2026-10-23
- AI agents drove this dev's GitHub Actions bill from $738 to ~$2,400/month — CellistOk6782 · 2026-10-23
- How to keep AI-built apps from looking vibe coded: designer themes plus a design MD file for agents — Roger_M_Taylor · 2026-10-23
- freeCodeCamp tutorial: how to build production-safe agent loops after a Claude Code run burned $50,000 — Roger_M_Taylor · 2026-10-23