Here are pictures I made in a colab ( https://youtu.be/XIzQhITfKVc ) with Dynamic Prompting.
The purpose is to make mistakes and accidents, to be very vague, or to use abstract words.
Plenty of trash, but sometimes you can get an idea of light, action, and composition.
I never seek perfection (you can do that with A1111, Comfy, Controlnet, LORAs, and all). I seek happy accidents to get ideas.
Most of these pictures come from a single prompt.
I made a Custome ChatGPT to create vocabulary and ideas for those prompts.
One prompt can be made of 10 of those:
{chiaroscuro | backlighting | bokeh | Rembrandt lighting | low-key | high-key | split lighting | butterfly lighting | softbox | diffuser | reflector | fill light | key light | rim light | ambient light | natural light | artificial light | light falloff | color temperature | light meter | exposure | depth of field | lens flare | vignetting | HDR | high contrast | low contrast | silhouette | shadows | highlights | flat lighting | bounce lighting | beauty dish | snoot | gobo | honeycomb grid | cinematic lighting | three-point lighting | light painting | sidelight | golden hour | blue hour | window light | diffused light | harsh light | soft light | directional light | umbrella light | catchlight | dappled light | graduated filter | neutral density filter | polarizing filter | light trail | light leaks | multiple exposure | motion blur}
http://www.patreon.com/Quick_Eyed_Sky for prompts, tips, and ideas.
https://www.youtube.com/@QuickEyedSky