Generation of 1024x1024, 768x1024, 1024x768 text to image.

Lora is trained to be photo real without use of trigger words, .5-.75 recommended for generation and up scaling.

Terms like indecent exposure, exposed breast, or 1girl in a park or 1girl in a train, will have the best results when using for text to image.

When up-scaling a lora value of .5-75 is also recommended. Limited use of positives with strong negatives will produce photo real results at 1.5-2x scaling. (Or iterative scaling),