It cannot guarantee a specific face
A text description may suggest age, hair, expression, and styling, but identity and facial details can vary between generations.
Workaround
Add a clear reference image when the workflow supports one, then review several outputs instead of expecting one exact likeness.