Very rudimentary model, based off using one image just to see if such low effort could produce decent results. Requires keyword. Side view only.

Best set at 0.50 and letting a style LoRA do the heavy lifting. 0.80 and higher causes blurriness.