!dream happy people, enjoying life, romantic style painting --image external-content.duckduckgo.com/iu/?u=http%3A%2F%2F1.bp.blogspot.com%2F_XCCzcMw9CYM%2FTUdxoE9CGaI%2FAAAAAAAAL0Q%2FRb3-r4pRgGQ%2Fs1600%2Ftumblr_lfrdmpIQZe1qe3pa4o1_500.jpg&f=1&nofb=1 --seed 3741993460

NSFW

!dream large diamond encrusted golden crown (schiaparelli diamonds crystals swarovski and jewelry iridescent) on head of a powerful lion, animals in distant background, lion prowls and looks at other animals, savanna, South Africa, sun overhead, 8k, masterpiece, extremely detailed, very wide angle, long shot --seed 4168771319

!dream Rooney Mara wearing Cookie Monster pyjamas at a red carpet movie premiere --seed 1253678807

yea all these AIs default to white people gotta add race

StableDiffusion founder Emad has mentioned making models trained on languages besides english tho

The A.I. has a hard time rendering black women. This was the best I could do

!dream Kanye West as the Joker, Batman, movie still, photograph, 8k, cinematic, dramatic, ambient lighting, wide shot --n 3 --seed 3941293449

!dream Kid Cudi as Batman, movie still, photograph, 8k, cinematic, dramatic, ambient lighting, wide shot --seed 1134351968

!dream Kanye West in the Joker make-up, movie still from Batman film, photograph, 8k, cinematic, dramatic, ambient lighting, wide shot --n 3 --seed 1078688473

!dream Kanye West as Patrick Bateman in American Psycho, 4k, cinematic, cinema --seed 4167601487

!dream Depressed man sitting alone on Mars 4k, photorealistic --seed 3976486306

!dream New York City 100 Years in the future 8k --seed 2813760099

Lmao

That looks incredible

It’s gotta get better with rendering people in general. It can barely do more than portraits, and even those aren’t great. In time.

!dream galactic city on the moon with glass domes, epic scene, hyper realistic, extreme detail, artstation, concept art, futuristic biodomes, cinematic lighting, wide angle lens —n 4 --seed 2261466133

do y’all think any gaming companies will use ai art for concept art ? feel like they could tbh

!dream Travis Scott but handsome --seed 3040016448

don't see why not. even if not as a company directive, individual creatives across all genres gonna be using this for inspiration i feel like

If they aren’t they are losing. Look at the stuff we are making and I have only been doing it for two days. At the very least they need to get ideas and expand on them from the AI art.

slightly tweaking scaling/steps on images can make such a difference to the image quality

i'm highkey heated that we didn't have the seeds on the earlier renders so i can't go back and touch them up

it does portraits perfectly in my experience. using —up upscale can fix faces rn. I’ll add a strength option to it soon

the sd model is still being updated and we’ll have a newer version running next week(?). It makes much better faces/generally better images overall

also going to be trying out some custom pipelines to improve base image quality, including a finetuned language model for preprocessing of prompts, multiple image models, fine tuning etc. so ktt/lowkey’s image generation will be unique.

just watching and trying to figure out the demand so it can be prioritized properly. Also curious about the demand for generating animations… but this might be going too far into Lowkey Labs territory.