this model consistently generates the same beautiful voluptuous blond woman. Her dataset was built on images from Alive AI, Higgsfield Angles and Photopea editing software,

When using only the trigger word, at the standard strength of 1, the results tend to resemble the character with some artistic changes. The sweet spot seems to be between 1.3 and 1.5. Beyond that, details from the training data begin to overwrite the prompt and resemble the training data. Her face remains undeformed and accurate up to the maximum strength of 2.0.

When including the phrases "braided blond hair, off to one side," a strength of 1.1 is sufficient.