Premium character LoRAs
Original synthetic characters, bundles and AI tools are in my shop:
Join the community on Discord: https://discord.gg/nB79wRNtMT
Changing Rooms
A concept LoRA for Krea 2 focused on fitting room selfies: women trying on clothes, lingerie and swimwear in store changing rooms, partly undressed, topless or nude, with the mirror, curtain, hooks and bench around them. It gives an amateur photo look.
Usage
Start with the trigger word ChangingRoomRK, then describe the fitting room (mirror, curtain or door, hooks, bench, clothes hanging), how she holds the phone, what she is trying on or that she is partly undressed or nude, her body, skin and hair, and end with the light. The more fitting-room details you give, the stronger the effect.
ChangingRoomRK, Mirror selfie of a woman with <hair> in a store fitting room, <trying on ... / topless / nude>, <curtain, hooks, clothes hanging>. The lighting is <light>.
Recommended weight: 1. Made for Krea 2 Turbo (8 steps, CFG 1).
Prompt Examples
"ChangingRoomRK, Photograph of an adult woman with light skin and long brown hair in a ponytail. She is taking a mirror selfie with an iPhone, holding the phone in her right hand, looking over her shoulder. The background is a gray fitting room with a bench, a brown coat, a red card, and a water bottle on the floor."
"ChangingRoomRK, Photograph of an adult woman with light skin and long brown hair, wearing loose blue jeans with a white drawstring and a black ribbed long-sleeve top, which she is lifting to expose her small, perky breasts and flat stomach. She holds a smartphone in her right hand, taking a mirror selfie, with her left hand lifting her top. The background is a fitting room with a black curtain, a gray puffer jacket hanging on a hook to the left, and a black handbag with a chain strap on the floor."
"ChangingRoomRK, Photograph of a woman with fair skin and long, straight brown hair. Her left hand holds up a black shirt with white trim, while her right hand holds an iPhone with a clear case, taking a mirror selfie. The background shows a gray curtain, a white wall, and a tiled floor."
Training
- Base model: Krea 2
- Type: Concept LoRA
- Steps: 4000
- Rank: 32
- Optimizer: adamw, constant learning rate
- Engine: ai-toolkit
Generation settings
Trained on Krea 2, and works on Krea 2 Turbo for faster generation.
- Raw: ~30 steps, guidance 3.5
- Turbo: ~8 steps, guidance 1.0
The gallery images were generated with krea2_turbo_mxfp8.safetensors, at 8 steps, guidance 1, LoRA weight 1.