Skip Navigation

InitialsDiceBearhttps://github.com/dicebear/dicebearhttps://creativecommons.org/publicdomain/zero/1.0/„Initials” (https://github.com/dicebear/dicebear) by „DiceBear”, licensed under „CC0 1.0” (https://creativecommons.org/publicdomain/zero/1.0/)C
Posts
18
Comments
48
Joined
3 yr. ago

  • Nice take on this one!

  • The space race should win. These are hilariously good and creative.

  • Some mysterious things are happening here. I’m curious though, why did you set it to come in on step 21 when you only have 20 steps? Wouldn’t that prevent it from running? I’m missing something.

  • These are amazing. Nice work!

  • I’m using a MacBook Pro M2 Max with 32gb ram.

  • With Tinkerbell? I think Peter Pan and Rufus were into her? Good question. Sounds like I need to go watch it again.

  • Cheers!

  • Good point, haha!

  • NSFW

    Beach party

    Jump
  • Prompters need to add (21 years old:1.3) in the positive prompt and “child, childlike” in the negative prompt. They also need to judge better whether the content the AI produces is too young looking.

    If you make something and think “is that too young?” Just assume it is and regenerate with the above tips. 21 is a good starting point because AI doesn’t seem to be completely accurate with ages yet.

  • Let me know what you find and if it’s useful! Cheers.

  • These took awhile to get right. The model is a blend of Real Vision 4, LazyMix+, and URPM. The top three were done with a ControlNet (I can't remember if it was Canny or Depth) to aid the pose and the face was improved with InPaint. The bottom one uses the "openblouse" lora, I think. I am finding that DPM++ SDE Karras is producing the best results, but I haven't tried them all systematically to prove that.

  • I thought it was an arm morphing into a Dyson. Gotta vacuum the deck, ya know?

  • For simple stuff, it's best to just go seek out the real photos I would say.

  • Sure, I’ll give it a go!

  • Have you looked at this LoRa managing extension? I saw it mentioned on the regional prompt GitHub readme.

    This extension summary:

    By associating LoRA's insertion position in the prompt with "AND" syntax, LoRA's scope of influence is limited to a specific subprompt.

  • For me, yes. It's been an experiment to generate a foreground and background. I'm not sure it's a trend though, but I have posted two versions of this idea. Are you noticing it more widespread?

    With AI generated stuff, I am trying to make more fantastical stuff that you wouldn't see in the real world (generally).

  • While I generally agree, there is some "artistry" in making prompts that work and figuring out all of the tricks to get better results. My own prompts have gotten a lot more interesting with custom checkpoint merging and dynamic and region prompting; these all took time to learn. I think posting the prompt/sources should be optional and people should request it if they want it, perhaps in a private message.

    I understand that "AI prompter" is becoming a job for some people at some companies and I can now start to see why. As you try to generate more realistic and interesting images, you have to know more about how stable diffusion and the many extensions work. It goes pretty deep—I feel like I am only scratching the surface.

  • I agree with the anime versus realistic stuff. I think posts should have maybe a "realistic" "anime" "artistic" or other tag. These should be defined.