Based on 60 images from her Instagram and promotional shoots etc.
Much improved, much more detailed than v1.0.
25,000 steps, Learning rate of 0.0015, 12 vectors per token
Example images all run using a standard prompt with outfit and background changes only. Did not generate any images that were not clearly her, nor any eldritch horrors.
Based on 60 images from her Instagram and promotional shoots etc.
Much improved, much more detailed than v1.0.
25,000 steps, Learning rate of 0.0015, 12 vectors per token
Example images all run using a standard prompt with outfit and background changes only. Did not generate any images that were not clearly her, nor any eldritch horrors.