all / g / s / fur / art / ai / 3dcg / f / rp / req / p2p / dis / lit / kaki chat
[Settings] [Home] [Catalog] [Search] [Thread list] [Stats] [Reports] [Watcher] [PMs] [Admin]
[Return] [Bottom]

Posting mode: Reply
BBCode
(for deletion)
  • Allowed file types are: gif, jpg, jpeg, png, webp, webm, mp4
  • Maximum file size allowed is 25000 KB.
  • Images greater than 255 * 255 pixels will be thumbnailed.
  • 159 unique users in the last 30 minutes (including lurkers)
    First
    [1]
    		Last
    File: {6+girls}, {{{{{{{{{{{{guro}}}}}}}}}}}}(…).png
    (2.98 MB, 1536x1024) ▶
    3124059
    No.980 [Report] AS
    ▶
    Does anybody know how to find AI models with some decent guro?
    I'm using NovelAI now, but apparently the AI has absolutely no idea how organs work, so in most cases instead of properly gutting the characters, it just draws a pile of guts or spread it all over the place like it's 'Saya no Uta'.
    Tags like guro, disembowelment, bisection barely works, even if you forcefully shove it using A LOT of brackets.
    >>
    No.981 [Report]
    ▶
    File: 00011-1271464179.png
    (1.1 MB, 1344x896) ▶
    1152429

    Given that most online AI resources are subject to censorship restrictions, Stable diffusion may be your only way to get accurate R18G pictures.

    If your computer is good, use SDXL, if not, use regular SD. Novices should use webUI first. After they become proficient, they can start using comfyUI, which is very convenient for creating pictures or videos.
    >>
    No.982 [Report]
    ▶
    File: 00045-3014081505.png
    (908 KB, 1344x896) ▶
    929843
    Compared with AI-generated images on the Internet such as NOVEL AI, midjourney, etc., the advantage of SD is that you can train customized ckpt and LORA to produce ideal results. For example, training AI to understand abstract concepts such as "beheading", "cutting off limbs", "head turning into penis", and "BEFORE AFTER", and letting AI understand these concepts and produce similar compositions in similar patterns.
    >>
    No.983 [Report]
    ▶
    File: 00041-2618329108.png
    (1010 KB, 1344x896) ▶
    1034304
    Generally speaking, SD runs normal pictures. If the user is not experienced enough, the results will most likely be worse than NOVEL AI and midjourney. But SD is valuable because of its strong plasticity. If you spend a certain amount of effort to understand its principles and train LORA and ckpt well enough, then SD can realize most interesting tricks that the human brain can come up with.
    >>
    No.984 [Report]
    ▶
    File: 00072-3435992630.png
    (1.27 MB, 1344x896) ▶
    1328898
    https://mega.nz/file/eIMFzLjZ#6gojy6u0oS_ODgJUx-sh4N4gg34daG-PyMf2j1zuW6k

    SD1.7 novice pack by aaaki,try to learn it, and wish you enjoy.
    >>
    No.985 [Report]
    ▶
    well I made an ayaswan model.
    A v2 came out today.
    You can find it (and others from the web) in the rentry repository linked in this thread

    >>39852
    >>
    No.1576 [Report]
    ▶
    Can you do more guts?
    [Top]
    Delete post: []
    First
    [1]
    		Last