I'm tired of coming up with and entering these damn long prompts. So here's a model that generates things I might like based on short prompts.
This LoRA tries to generate what I like in style that I like, and it doesn't positively affect long prompts and may even reduce detail. If you're a fan of using a ChatGPT to create prompt for three pages of detailed text, or, worse, write long head prompts by yourself, then skip and forget about this LoRA.
Working range for LoRA's strength is [-2.5 .. 2.5] for ZIT version and [-0.5..1.5] for Flux version.
Look at these images. It's not about age, it's about aesthetic. You even could find negative strength value makes image more... realistic?
If you're still here, get the secret amplifier for this LoRA:
masterpiece, best quality, cinematic lighting, dramatic lighting, bokeh, 8k, high quality, high detail
Description
With rearranged blocks it's unstable like a god of unstability
FAQ
Comments (3)
What were those blocks you removed? Thank you
Some noisy and redundant parts
I'm not very sure of what this is doing, but adding it at a low level (0.4) may enhance my results.
But it is more like adding noise to the prompt, not sure if it's better, but it may change them in iteration