Pawel Witkowski's picture

AI generated

This is one of my first attempts at AI Generated photos. I'm so surprised of the results that I had to add this here just as a notice to all of us that makes photos...

It's generated using Stable Diffusion, enlarged 3x from typical 512x768 and then added some analog noise and small fixes in Photoshop

Log in or register to post comments

Only strange thing is her thumbs. I wouldn't notice that if I wasn't looking.

Good catch! Second one is a leafs in left bottom corner are too sharp (like messed up depth of field). I did fixed those two mistakes manually in Phtotoshop already, but skipped updating this version on Fstoppers thou :) Those mistakes are only way to figure out that it's still not real photo. Without that it would be damn hard I think.

it's all about the prompts

It Is! But man... It's hard to go out do some shoots around if you just can type in some prompts and got somewhat similar results....

Don't let it demotivate you, this is just a tool not a replacement for your work. How many prompts did you have to go through before this result. That's your expertise coming through. A regular person doesn't know the questions to ask. (myself included) It's a great piece for showing clients what you're thinking and time saver in that capacity. I think Post processing will be faster with AI too. The key is to adapt.

Strange times incoming for sure. I'm not yet that demotivated, and for now I use already a bit of AI tools to enhance my work and I will continue so. Eventually I enjoy process of doing the work with models so prompts are not that fun in that area, but as it comes for results well... I'm just excited to see where it goes :)