Premium character LoRAs
Original synthetic characters, bundles and AI tools are in my shop:
Join the community on Discord: https://discord.gg/nB79wRNtMT
Downblouse
A concept LoRA for Krea 2 focused on downblouse shots: loose tops and low necklines seen from above while she leans forward or bends, showing cleavage with or without a bra, at home, in kitchens and offices. It gives an amateur photo look.
Usage
Start with the trigger word DownblouseRK, then describe the woman (skin, hair), her top (neckline, fabric, color, how loose it is), whether she wears a bra and what it reveals, the camera angle looking down into the top, her pose and the place, and end with the light. Naming the loose top and the downward angle brings out the effect the most.
DownblouseRK, Photograph of a woman with <skin> and <hair>, leaning forward in a loose <color> <top> that reveals <...>, seen from above. She is <pose> <place>. The lighting is <light>.
Recommended weight: 1. Made for Krea 2 Turbo (8 steps, CFG 1).
Prompt Examples
"DownblouseRK, Photograph of a woman with fair skin and long, wavy red hair. She is wearing a light blue, slightly loose, button-up shirt that is unbuttoned to reveal a black bra and her cleavage. She is bending forward slightly, with her left hand resting on her thigh and her right hand out of frame."
"DownblouseRK, Photograph of a woman with fair skin and light brown, wavy hair. She is wearing large, round glasses with a black frame, a loose, beige knit sweater with a deep neckline that reveals the top of her breasts and a hint of a white bra underneath. Her pose is bent forward, leaning slightly to the side, with her hands resting on her jeans."
"DownblouseRK, Photograph of a woman with fair skin and long, wavy, blonde hair. She is leaning forward, looking directly at the camera with a slight smile."
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.