Page 1 of 1

Dall-E 3 (OpenAI) and Unstable Diffusion

Posted: Thu Nov 30, 2023 8:25 am
by freezefantasy
Tried using the new OpenAI Image generation feature and while you can't do anything adult, with smart word choices you can get some interesting scenarios:
BEST 1.png

Also Unstable Diffusion has gotten a lot better now with the community figuring out how to prompt to get photorealistic but still fantastical scenarios.
00011-1115884126.png
00047.png
00165-1177625579.png
00314-3073529972_cyberrealistic_v30BETA.png

Re: Dall-E 3 (OpenAI) and Unstable Diffusion

Posted: Thu Nov 30, 2023 3:26 pm
by C7373
What prompts do you use on unstable dillusion for the photorealistic images, interested in trying it out

Re: Dall-E 3 (OpenAI) and Unstable Diffusion

Posted: Thu Nov 30, 2023 3:32 pm
by Basilisk
Something else to look at is the InstructPix2Pix functionality built into Automatic1111 stable diffusion. You have to download a model trained in pix2pix but you can basically use it to photomanip an existing picture into plastic or what have you. I've just begun experimenting with it but have had some pretty good results so far.

Re: Dall-E 3 (OpenAI) and Unstable Diffusion

Posted: Thu Nov 30, 2023 9:35 pm
by freezefantasy
I get my prompt inspirations from the Unstable Diffusion discord channel.

https://discord.gg/5qGXehPh

You can react to the NSFW images others share to see the prompt and LoRAs and controlnet they used