When using the prompt “pregnant”, neural network draws characters at advanced stages of pregnancy, the AI ​​creates pictures with defects in appearance: a third hand on the stomach, two navels, etc. This LoRA was trained on selfies of pregnant women, posted on social networks under the tag #pregnant.

LoRA's task is to add the effect of pregnancy to characters, avoiding the appearance of specific effects. But due to the small size of the dataset and the specificity of the sample, the pictures may have features typical of selfie photos in front of a mirror.

Characters tend to show bare midriffs - try indicating clothing in the prompt.