For more LoRAs and updates, Join my Discord:
https://discord.gg/ZVWVhT43GW

Trained using AI-Toolkit on 859 images extracted from videos.

Showcase images generated with the mxfp8 turbo model.

This LoRA was captioned in a structured way, so prompts work best when you mirror that format:

HMNSFW <position>, ANGLE_<angle>, <plain-English description>

1. Position token (right after the trigger): cowgirl, missionary, doggy, handjob.
Add anal for anal variants (e.g. cowgirl_anal).

2. Angle token (optional but strong): ANGLE_POV_ABOVE, ANGLE_BEHIND, ANGLE_POV_FRONT, ANGLE_SIDE_PROFILE, ANGLE_CLOSEUP, ANGLE_OVERHEAD, ANGLE_LOW_ANGLE.

3. Description: write a natural, detailed paragraph — shot type & setting, the woman's

appearance (skin tone, hair, build), pose, clothing/nudity, where contact/penetration

is, lighting, and background. Longer, descriptive prompts (~60–120 words) match the

training captions and give the best results.

4. For cum/finish shots use the cum register and describe it plainly as

"cum / thick white liquid" on the face or body.

Example:

HMNSFW cowgirl, ANGLE_POV_ABOVE, a low-angle POV of a slim woman with long blonde hair

riding a man on a bed, bright diffused daylight, her hands on his chest, erect penis

visible at the point of contact, neutral wall and headboard behind, shallow depth of field.