When generating an image with genAI, we can determ gender age or ethnicity through trigger words. This can also be manipulated with specific trained LORA models. Adding a specific LORA with specific positive or negative value to the prompt itself can lead to interessting results. As this tech is still highly biased and hard to implement in the creation of diverse content, it is still an interesting way to have more control to create human representation of all kind.
Slider test rows
The following test rows show each slider in action based on a pretty simple input. The images are also postcompiled with ReActor, a face swaping tool to get a more comprehensive view to the results. The base prompt is: upper body two third portrait of bald androgyn person in black bomber jacket