⚠️ ALPHA RELEASE WARNING: This LoRA is experimental. Expect quirks, and “rough edges” while I iterate on the dataset and training settings.<h3 id="what-it-does:">What it does:</h3><ul><li>Captures CivChan’s facial identity and overall look</li><li>Works for portraits and full-body prompts</li><li>Handles wardrobe changes</li></ul>Known issues / limitations (alpha)<ul><li>Not fully tested (you are the testers)</li></ul><h3 id="recommended-settings">Recommended settings</h3><ul><li>Model: FLUX.2 Klein 4B</li><li>LoRA strength: start at 0.6–0.8, go up to 1.0 if you want stronger likeness (its probably underbaked) </li><li>Steps: 50 (base Klein often needs more steps)</li><li>Guidance/CFG: around 4–6</li><li>Resolution: 1024x1024 recommended; higher resolutions work 🤷</li></ul>

CivChan  - Unofficial

I recaptioned the Civchan dataset and trained it at a linear rank of 64. I baked in her physical appearance, but kept everything else—hair, eye color, and outfit—fully configurable.

Files with SHA256: bc50b88197b9d92f657e3dfe44645c669aa2091c90b8df6fb1852d3caf542ee9