Premium character LoRAs
Original synthetic characters, bundles and AI tools are in my shop:
Join the community on Discord: https://discord.gg/nB79wRNtMT
Deepthroat
A concept LoRA for Krea 2 focused on deepthroat oral sex: adult women taking a penis deep into the mouth and throat, kneeling, lying with the head off the bed or on all fours, with saliva, tears and running makeup. It gives an amateur photo look.
Usage
Start with the trigger word DeepthroatRK, then describe how deep the penis is, her expression (eyes, tears, saliva, makeup running), her position and the man's, her body, skin and hair, what she wears or that she is nude, the camera angle and the place, and end with the light. The more precise the position and the expression, the stronger the effect.
DeepthroatRK, Photograph of an adult woman with <hair>, <position>, deepthroating a man's penis, <saliva / tears / makeup running>, wearing <... / nude>, <camera angle> <place>. The lighting is <light>.
Recommended weight: 1. Made for Krea 2 Turbo (8 steps, CFG 1).
Prompt Examples
"DeepthroatRK, Photograph of a nude adult woman with light olive skin and dark brown, wet hair pulled back. She is performing deep oral sex on a man. Her mouth is fully wrapped around the man's erect penis, which is deeply inserted into her throat."
"DeepthroatRK, Photograph of an adult woman with light skin and long, brown hair with blonde highlights. Her lips are wrapped around a large, erect, circumcised penis, which is deeply inserted into her mouth and throat."
"DeepthroatRK, Photograph of a fair-skinned woman with red hair in a loose updo, wearing heavy black eyeliner, mascara, and pink eyeshadow, with a nose ring. Her mouth is wrapped around an erect, uncircumcised penis, which is deeply inserted into her throat."
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.