Open-source PixlStash uses face-likeness clustering to streamline character LoRA creation
Infamous_Campaign687 · reddit · 2026-11-17
What it is
PixlStash is an open-source image management app that packages character sheet generation, dataset selection and LoRA training into one workflow, replacing scattered Python scripts.
Key mechanics
- Its database backend computes face likeness pairs across the whole library, making face-likeness clustering nearly free — handy for picking consistent-character LoRA training data.
- Iteratively create and improve LoRAs; outputs flow back into the library and are instantly usable in workflows, powered by ComfyUI and AI toolkit under the hood.
- Ships adapted workflows for Krea 2, MOSS-Image, Flux.2 Klein 9B (t2i) and Qwen Image Edit 2.1 / Flux.2 Klein 9B (reference-image).
Verdict
The author finds Qwen Image Edit 2.1 easily the best model for both tasks.
More from coding & agent
- YC F26 startup OpenScaffold turns past deployments into reusable playbooks for coding agents — ycombinator · 2026-11-17
- Manufact kicks off launch week with MCP publishing to ChatGPT — tensorqt · 2026-11-17
- Learner watches your agent traces and drafts one-tap skill updates for approval — bfrench · 2026-11-17
- Tencent Hy4 Flash scores 5.74 on RuneBench at just $0.03 per run, cheapest on Pareto frontier — mariofilhoml · 2026-11-17
- Together AI boosted agent task success from 67% to 96% with homegrown agent evals — nutlope · 2026-11-17
- Not using coding agents is now irresponsible: software engineering as managing a compiler — PrateekVJoshi · 2026-11-17