CivArchive
    Preview 47478542
    Preview 47455924
    Preview 47455995
    Preview 47480247
    Preview 47460047
    Preview 47471432
    Preview 47454738
    Preview 47478150
    Preview 47464158
    Preview 47469132
    Preview 47495135
    Preview 47495136
    Preview 47484765
    Preview 47491956

    Hassaku aims to be a anime model with a bright and distinct anime style.

    My Discord for everything related to anime models and art.
    If you'd like to support my work, you can do so at: SubscribeStar, every bit helps and is truly appreciated!

    ____________________________________________________________

    Supporters:

    Thanks to my supporters tamashiicolle, pttcot, SETI and Kodokuna

    ____________________________________________________________
    Using the model:

    • The model was trained primarily on images with minimal disruptive elements, such as floating text, logos, speech bubbles, or signatures. If any of these elements appear in an image, please include "signature" in your negative prompt.

    • Danbooru Metadata and franchise tags are not used in training, so please avoid using them in prompts. Metadata such as "highres" and franchise tags like "re:zero kara hajimeru isekai seikatsu" are not used in training.

    • A CFG scale between 3 and 7 is recommended. For best results, use CFG 6 with the Euler a sampler.

    • Refer to the example images, to understand, how the model should be used.

    • The model is designed to be simple and straightforward to prompt. Use the following tag order:

      1. Number of characters

      2. Character Tags

      3. All remaining tags
        Example: 1girl, rem \(re:zero\), standing, masterpiece, upper body

    • Here are some resolution options for SDXL:

      • 1536 x 640

      • 1344 x 768

      • 1216 x 832

      • 1152 x 896

      • 1024 x 1024

      • 896 x 1152

      • 832 x 1216 (most recommended)

      • 768 x 1344

      • 640 x 1536

    Lora use:

    • NoobAIs Loras working for the most part better in compare to Illustrious loras

    • Versions 2 and 3 is less compatible to most loras trained on Illustrious or Noob, because it got further trained away of both. Style A is still the most compatible for it.


    ______________________________________________________

    Version and License info:

    Description

    Model 1.2 without early access as christmas gift! examples are now with highress fix.
    V1.2 my attempt to reduce eye and image artifacts


    Whats better:
    - Eye and image artifacts reduced
    - More stable in image composition
    - Some concepts having a clearer image output

    Focusing on making backgrounds better next, without getting artifacts back, especially on the eyes.

    Edit: After people tested it, it seems artifacts got worse on specific promts, but got overall better. Working on that also. The new version is a bit more prompt following in compare to the last version, so its more important to use the right negative promt "bad quality" instead of "low quality".

    FAQ

    Comments (12)

    BMTZZDec 25, 2024· 10 reactions
    CivitAI

    This is gonna be a really stupid question, but how do you use highrez fix like you mentioned in the description? Never touched it cause i usually run everything through Adetailer

    anotherstabledif1714Dec 25, 2024· 4 reactions

    Do you use Forge or A1111? Because its pretty easy. I personally enable it, turn high rez steps on 10 and then increase the size by 1.5 or 2.0, dependent on what you want to do with them. So it works also as a resize, while fixing anatomical stuff. Hope it helps!

    JigglePhysicsDec 26, 2024· 2 reactions

    @anotherstabledif1714 i generally do half of the sampling steps in hires steps, so for example, 30 sampling steps then i do 15 hires steps.

    BMTZZDec 27, 2024

    @anotherstabledif1714 i use reforge, i know how to turn it on i was more asking if they does resize instantly with it or if they dont resize at all while still in txt2img and how they split their normal vs highrez steps

    BMTZZDec 27, 2024

    @JigglePhysics interesting ill try that

    Izen72Dec 27, 2024· 6 reactions
    CivitAI

    Amazing as always. Hopefully there will also be a V-Pred version

    Madafada1991Jan 5, 2025

    unlikely i think, noobai has stupid license where you are forced to put and show prompt online

    From https://civitai.com/models/833294/noobai-xl-nai-xl

    "III
    . Open Source Community

    To foster a thriving open-source community,users MUST comply with the following requirements

    Open source derivative models, merged models, LoRAs, and products based on the above models.

    Share work details such as synthesis formulas, prompts, and workflows.

    Follow the fair-ai-public-license to ensure derivative works remain open source."

    "MUST comply with the following requirements... ...share work details such as synthesis formulas, prompts, and workflows." Users of NoobAI-XL and derivatives are compelled by this license to share "...prompts, workflows". It is a requirement.

    noobai is also seemingly released under TWO licenses because there is the license spelled out on the CivitAI description, and then there is the addendum license here: https://civitai.com/models/license/1140829

    While
    the restrictions laid out between the two licenses are similar, they're not the same. It's confusing but the limitations placed are still extreme no matter which way you look at it.

    Ikena
    Author
    Jan 5, 2025

    Yup, thats exacly why i don't use noobai but illustrious as base model.

    Dead_Internet_TheoryJan 11, 2025· 1 reaction

    @Ikena Doesn't Illustrious use v-Pred too?

    Also, what is extreme about requiring derivatives be open-source? You're not worried people will "steal your hard work" after training an AI on artist's works, I hope?

    Madafada1991Jan 11, 2025· 1 reaction

    @Dead_Internet_Theory No, HassakuXL Illustrious is not trained under ANY NoobAI variant which is why those scat tags that are present in NoobAI is not there. Illustrious XL v0.1 CAME BEFORE ANY NOOBAI VARIANT EXISTS. The only model that predates Illustrious is KohakuXL Beta Rev 5. V-Pred is NoobAI variant, trained from IllustriousXL v0.1 and has NEW license of their own SO ILLUSTRIOUS XL DOES NOT COME FROM ANY NOOBAI model, Come on, can't you see the DATE? IllustriousXL v0.1 came out on September 25 2024 while the VERY FIRST VARIANT OF NOOB AI CAME IN OCTOBER 2024. People actually thought NoobAI is the original for Illustrious huh?

    HassakuXL is combined from WAI NSFW, Illustrious v0.1, and probably AnimagineXL3.0 at very low strength which explains why animaginexl prompt tags work. And of course, @Ikena's original setting of training too. Its completely different branching after IllustriousXL v0.1.

    Here's my take, if you dont like it, leave.

    @Madafada1991 I didn't mean it as "Doesn't Illustrious use NoobAI's v-Pred too?" but as in, I thought Illustrious also used the same v-prediction model, with no implication that it took it from NoobAI (I'm not sure who came up with v-Pred).

    josefstalin3d866Jan 4, 2025· 7 reactions
    CivitAI

    Hi guys - I'm looking for line and paint styling, loras, for this specific model. I appreciate anyone who helps.