information:
this is a dreambooth test that merge random face from generated images.
training steps: 2500 steps, batch size 4.
recommended weights: 5-7
additional information: a model testing to check better settings. when the model pass the check, additional informateur may be added later.