As its name implies this LoRA tries to create images that look like has some depth or meaning, but it is in the end meaningless, is just for the looks.

Trained with realistic and semirealistic

I have found that the model affects incredibly to the outcome of the image, so I advice to test on different models also the strength will create more intense images, and for more generic images I recommend less strength.

On negative strength it creates the oposite to this obviously.

I also advice to experiment with hyper detailed prompts and also with short prompts, as I have found that both generate very different images

Pony version: the strength needed varies depending on the model used, I suggest to experiment till find the desired outcome.