Codex built and runs a local anime pipeline: SD + LoRA → Wan 2.2 on RTX 5060 Ti

Wonderful_Sample6291 · reddit · 2026-09-07

A Redditor gave Codex one objective — build a local pipeline producing a short anime sequence on an RTX 5060 Ti 16GB — then only reviewed outputs and adjusted strategy.

Pipeline: Codex (production supervisor) → ComfyUI → Stable Diffusion + character LoRA → image QC → Wan 2.2 clips → QC/retry → editing.

Key lessons:

Current output is a 30s sequence, visibly imperfect but a reproducible proof of concept for agent-driven video production.

Original post →

More from coding & agent

coding & agent channel →