AIBooru

Comments

Blacklisted:
[hidden]

ewae said:

that looks super cool, do you know what kind of cadence or seed types you used? i wanna try making something like this too

Hello,
Describing the process or the specificity of this would be too long for a single comment.
I had prepared a quick tutorial month ago in parallel, and I'd be happy to share it.
I don't know the policy for external links, so if you want some information feel free to reach me on Pixiv or Discord at : Kairos4463

  • 5
  • Reply
  • [hidden]

    Placido123 said:

    Does somebody know the Author of this one? Like a Pixiv account or something? please

    Hello,
    I do not have a Twitter account, but I've just updated my artist entry with my Pixiv URL.
    Those animations were only posted here and not shared on pixiv as I keep my pixiv account SFW only.
    Also, those animations were extremely time-consuming, and with the lack of feedback, I just stored the idea, and this one was the last one produced.
    But seeing that it started getting some traction recently, I'm considering making them again.

  • 5
  • Reply
  • [hidden]

    Her outfit is really all over the place, isn't it? Got a school uniform tie, seems the mangled skirt is maybe part of a pencil skirt, she's got something that looks like shoulder laurels which make her look like a cop or something...

  • 0
  • Reply
  • [hidden]

    tiki said:

    I wish you would continue to post with metadata, I was learning so much from your prompts! -T

    I just didn't want people to judge me because of how highly weighted all the prepubescent innocent 9 years old girls were in the prompt. I'll add metadata to some new uploads, promise.

  • 2
  • Reply
  • [hidden]

    JohnCena said:

    Any eason why model is missing, given that it's your generated piece ? Also how the model knew to paint Rin or Emiya, let alone this fancy position ?

    Sorry, I'm far from pc and quite some time passed, can't answer from.my acc. I used a Lora for the position and then remade the character faces and hair with i2i

  • 0
  • Reply
  • [hidden]

    UnknownNo3 said:

    Thanks a lot for your reply! Never thought to find a model on 4chan before. (つ﹏⊂)

    Couple of months back 4chan was the place to get Lora's for the English-speaking Stable Diffusion users, until Civitai became a more viable platform for models.

  • 2
  • Reply
  • [hidden]

    UnknownNo3 said:

    I didnt find mayl's lora on civitai. So sweetmix can generate her without a character lora? Or did you use some private model?
    Sorry for may poor english if my expression looks offensive.
    I'm still a noob on using Stable Diffusion, dont know how to generate a specific character without a lora.

    Your English is fine. Sweetmix won't generate Mayl all by itself. If you want to make specific characters that aren't Reimu, Marisa, or Miku, you will need a Lora.

    I didn't find this Mayl Lora on Civitai. This one was a Lora made by anon on 4chan's /hdg/: https://boards.4chan.org/h/thread/7459769/7461770#p7461770

  • 2
  • Reply
  • [hidden]

    I didnt find mayl's lora on civitai. So sweetmix can generate her without a character lora? Or did you use some private model?
    Sorry for may poor english if my expression looks offensive.
    I'm still a noob on using Stable Diffusion, dont know how to generate a specific character without a lora.

  • 0
  • Reply
  • [hidden]

    Done using the Mistoon_Anime model, and based on a prompt from a DeepBooru of rule34.xxx/index.php?page=post&s=view&id=3822471. Upscaling and small edits were done alongside an inpaint for a more Kula-ey face. Still some odd stuff like the around the eyes and the leftmost hand.

  • 0
  • Reply
  • [hidden]

    The commentary for this one is interesting. It's a 7 syllable 4 line 绝句, or Chinese quatrain. I'd try to translate it but I'm afraid my qualifications on translating poetry are quite lacking. It's mostly about standing at the head of the river looking lovesick far in the distance, and as per any sort of Chinese poetry it's rife with literary allusions to stuff that I haven't read about.

  • 2
  • Reply
  • [hidden]

    antlers_anon said:

    @CrimsonKage
    This was made with the oldest version of anything (which for some reason is v3), but you should get better results with the newer one anyway (and a different sampler).

    Denoising is for highres fix, which is also working differently now (highres fix, not denoising). The size it has listed is actually after highresfix, newer versions of a1111 save the firstpass size instead. I don't know what the fistpass size was for this image, probably something like 420x700x but it's just my guess,

    Thanks for your reply, I had narrowed down that it was hires fix, unfortunately, hires fix has never worked for me. The anything v3 though, that's new info, I had made a merge with v4, so if I want to replicate this (as a starting point, mind you) that's useful information. If I could ask for more info, do you happen to remember your hires steps, upscaler or upscale by values?

    Seems that for whatever reason, my results only have one or two colored surfaces, not the lovely colorscape you ended up with at all. Will keep trying though.

  • 0
  • Reply
  • [hidden]

    @CrimsonKage
    This was made with the oldest version of anything (which for some reason is v3), but you should get better results with the newer one anyway (and a different sampler).

    Denoising is for highres fix, which is also working differently now (highres fix, not denoising). The size it has listed is actually after highresfix, newer versions of a1111 save the firstpass size instead. I don't know what the fistpass size was for this image, probably something like 420x700x but it's just my guess,

  • 0
  • Reply
  • [hidden]

    Perfect again.
    You never cease to amaze me.
    How do you get so good pictures for this Lora?
    I also tried, but I didn't find a checkpoint with which it would go well. Perhaps the problem is in prompt?

    If you write the name of the model used and prompt, I will be very grateful.

  • 1
  • Reply
  • [hidden]

    I'm not that skilled. :P

    Also, this isn't done locally since I don't have anything capable of running SD software, so I'd be totally depending on "lucking out" when it comes to inpainting using the PixAI tool.

    Updated

  • 0
  • Reply
  • 1 12 13 14 15 16 17 18 19 20