About this model:

This "lora" was trained with 80 images and a batch size = 4. The total training steps were 2000. I am the author of several models around here that are published as community models. It's been about 80 attempts or more to make a better model of Gal Gadot. This is my last attempt; This model would be the fourth attempt with the same 80 images and different captions. The truth is, I consider this model to be superior to all the ones I have made before, for the simple fact that it has a great ability to adhere to the indication, as well as excellent visual quality; I recommend that you try it for yourself, I would appreciate if you can use the model and post an image that you have generated and liked.

About generating images with this model:

I recommend this configuration to generate:

IMAGE SIZE:

Images above 1024x1024. I have used 1280x1280 for several tests.

SAMPLER:

DPMPP_2S_ANCESTRAL

SCHEDULER:

BETA

STEPS:

+20

STRENGTH:

0.9-1.0 (I recommend 0.9)

STRENGTH:

2.5-3.5 (I recommend 3)