The Lora is useful to represent examples of the use of science and technology in daily life. It is useful for examples of applied citizen science. It is not used for sexual purposes.

Works better with the realisticVision V60B1_v51VAE.ckpt Checkpoint as well as the epicrealism_naturalSinRC1VAE.safetensors Checkpoint. Both work great!

Recomended:

Positive Prompt

middle shot, crowed big city, (simetry:1.25), photorealistic photo, sharpen, natural lighting, extradetailed, brown hair young slim american woman (wearing bold glasses:1.15), she is (paying attention to her cellphone:1.05), we see a slight movement of her hair. The camera stays infront of her.

Negative Prompt

(bad eyes:1.2), text, watermark, asian girl, formal clothing, artificial ligh, fake, toon, anime, (bad morphology:1.1), overlapping, (eye missmatch:1.25), old, night, crazy fingers, short hair, man, closeup, lost look, (close eyes:1.05),old woman, bad reflexion, (eldery:1.2), (2 cellphones:1.05)

Steps: 20

cfg: 5.0

sampler_name: dpmpp_2m

scheduler: karras

Once you add the Load Lora node, I recomend a value of 0.1 for the strengh_model and a 4.0 for the strengh_clip.

The best images are the ones with the girl looking straight ahead. I guess it's because I used a smaller amount of images where she appears from a three-quarter view for the AI training.