Z- Image Update:

Seems to have desired effect on images but I see really low quality on texture and details compared to qwen. Up to user's choice...

Eyes rolled, tongue out, mind blown. This lora intended to produce that result...

Trained on carefully generated, edited and tagged dataset completely anonymized and by me.

My main focus train on such a way that it only affects the related parts of the image, so feel free to describe the rest. Just ensure the trigger word ahegao face.

Here is example prompt to create similar images:

ahegao face. A medium shot captures a young woman with long, wavy, deep red hair that cascades over her shoulders and partially obscures her upper chest. Her head is tilted back and to her right, causing her eyes to look upward and to her left. Her mouth is open wide, and her tongue is extended outward, showing its pinkish-red underside. Her eyes are large and light-colored, with prominent, dark, winged eyeliner extending from the outer corners. A small, silver-colored nose ring is visible on her left nostril. Her skin tone is light, and her neck and upper chest are bare.  She is seated, with the top of a brown, textured fabric couch visible behind her left shoulder. To her right, the large, dark green, veined leaf of a plant is visible. The background consists of a dark green, vertically pleated curtain. The lighting is direct and frontal, illuminating her face and upper body.

I used official comfy workflow. Didn't test them with lightning loras, usual settings are:

Euler/beta
CFG 3.5 - 4.0
30+ steps