this post was submitted on 26 Oct 2023
57 points (96.7% liked)

AI Generated Images

7195 readers
230 users here now

Community for AI image generation. Any models are allowed. Creativity is valuable! It is recommended to post the model used for reference, but not a rule.

No explicit violence, gore, or nudity.

This is not a NSFW community although exceptions are sometimes made. Any NSFW posts must be marked as NSFW and may be removed at any moderator's discretion. Any suggestive imagery may be removed at any time.

Refer to https://lemmynsfw.com/ for any NSFW imagery.

No misconduct: Harassment, Abuse or assault, Bullying, Illegal activity, Discrimination, Racism, Trolling, Bigotry.

AI Generated Videos are allowed under the same rules. Photosensitivity warning required for any flashing videos.

To embed images type:

“![](put image url in here)”

Follow all sh.itjust.works rules.


Community Challenge Past Entries

Related communities:

founded 1 year ago
MODERATORS
 
you are viewing a single comment's thread
view the rest of the comments
[–] Shelena@feddit.nl 0 points 1 year ago (1 children)

Really nice! I like most of them, actually. I think the comic book ones are the best. Thank you and also thanks for the explanation. I have not that much experience with generating images yet. This definitely helps!

[–] tal@lemmy.today 1 points 1 year ago* (last edited 1 year ago)

You're probably not seeing a particularly broad set of examples of what could be done; I happen to be a real sucker for intense light-on-dark hand-drawn stuff, so it's just a random sprinkling of some of the presets that I happen to have sitting around for my own use and like the look of.

Another suggestion: I'm not familiar with StableDiffusionWeb, which is apparently what you're using, but I would assume that it's internally using Stable Diffusion and some model. If you look at civitai.com, you can see a metric ton of images that people have generated...and also the prompts that they used to achieve those effects. Some of those won't be using the standard model -- they'll be trained on other images -- so you won't get exactly the same impact, but usually the prompts are still a solid place to start.

So, to take a random example:

https://civitai.com/images/2680672

Fairy tale fairy tale|gothic, natural skin texture, (full body:1.4), enchanted glade, lighting by flying fireflies, tree branches alive with dancing neon leaves, a beautiful woman wearing a white chiffon dress, crystal shoes, jewelry, lips, medium, open, sensitive, sleeveless, solo, sundress, teeth, thighs, wind, a swirling tornado around her, paint explosion splatter, <lora:SDXLPaintSplash:0.7> paint splatter <lora:DetailedEyes_V3:1> <lora:add-detail-xl_2:1> . Magical, fantastical, enchanting, storybook style, highly detailed

Okay, that's using a nonstandard model and some LoRA models -- which are like little models, that add on effects. I haven't downloaded those LoRA models, and I assume you won't have them available. So we aren't going to get exactly the same effect. But...let's try tacking the non-LoRA prompt stuff onto your username. Maybe pull off a couple of parameters...maybe take off one or two to make it just a bit less revealing...

Shelena, Fairy tale fairy tale|gothic, natural skin texture, (full body:1.4), enchanted glade, lighting by flying fireflies, tree branches alive with dancing neon leaves, a beautiful woman wearing a white chiffon dress, crystal shoes, jewelry, medium, open, sensitive, sleeveless, solo, sundress, wind, a swirling tornado around her, paint explosion splatter, paint splatter . Magical, fantastical, enchanting, storybook style, highly detailed

Steps: 20, Sampler: Euler a, CFG scale: 7, Seed: 2, Size: 1024x1024, Model hash: ebf42d1fae, Model: realmixXL_v15, Token merging ratio: 0.5, Version: v1.5.1

And that required me I dunno, maybe three minutes for looking at the websites and trying a couple images and copying the prompt, and at one point in there my web browser crashed, which counted towards those three minutes. I'm not particularly advocating for that as an end image, just pointing out the knobs that are available.