cupi3.0

詳細

ファイルをダウンロード (1)

モデル説明

✨ cupi3.0 — Character Anchor LoRA

Built because I got tired of redheads with glasses forgetting who they are mid-render.


🔥 What This Is

cupi3.0 is a character-anchor LoRA focused on consistency, personality, and control.

It stabilizes:

  • rich red / red-orange hair with depth and highlights

  • glasses that actually sit right and stay put

  • expressive faces that don’t drift across angles

  • soft, intentional curves that keep their proportions

This isn’t a costume or pose LoRA.
It’s the character backbone you build everything else on.


🎯 Purpose

The purpose of cupi3.0 is simple:

👉 Same character. Different scenes. No surprises.

Portraits, cozy interiors, dramatic lighting, playful or spicy vibes — the character stays recognizable so you can stop babysitting the model and start composing shots.


🤔 Why I Made This

If you’ve ever had:

  • a face subtly change every batch

  • red hair slowly turn orange → blonde → ???

  • glasses decide they’re optional

  • bodies exaggerate or flatten depending on camera angle

…yeah. Same.

cupi3.0 exists to shut that down and give you predictable, repeatable results.


🛠️ How to Use

Recommended weight

  • 0.8 – 1.1 (sweet spot for most setups)

Works best for

  • portraits & busts

  • cozy indoor scenes

  • bedroom / lifestyle lighting

  • over-shoulder & looking-back angles

Prompting advice

  • Let cupi3.0 handle face, hair, glasses, and skin

  • Use your prompt for camera, pose, outfit, and environment

  • Don’t over-describe facial features — it already knows

Skin finish

  • Slightly polished / glossy by default

  • Easy to matte down with negatives if you prefer softer skin


🧠 Workflow Friendly (On Purpose)

This LoRA plugs cleanly into system-based workflows:

  • base character

  • pose / camera blocks

  • outfit swaps

  • reusable environments

Drop it in. Get consistency. Move on.


📸 Example Images

All images on this page were generated using cupi3.0 as the character base, with different poses, lighting, and outfits — same face, same vibe.

(Yes, it behaves in ComfyUI.)


If you use cupi3.0, post your results.
I’m genuinely curious what people push it toward — wholesome, cozy, cinematic, spicy, or absolute chaos.

このモデルで生成された画像