{
    "model_id": 543369,
    "model_name": "Mikoshi [Pony]",
    "model_description": "<p>\n <strong>\n  <em>\n   <u>\n    <span style=\"color:rgb(250, 176, 5)\">\n     Best settings:\n    </span>\n   </u>\n  </em>\n </strong>\n</p>\n<ol>\n <li>\n  <p>\n   <strong>\n    <span style=\"color:rgb(250, 82, 82)\">\n     Sampling Method:\n    </span>\n    Euler A or Euler A SGMUniform\n   </strong>\n   -\n   <strong>\n   </strong>\n   you can mess around with this\n  </p>\n </li>\n <li>\n  <p>\n   <strong>\n    <span style=\"color:rgb(250, 82, 82)\">\n     Sampling Steps:\n    </span>\n    20 ~ 60\n   </strong>\n   depending on your sampling method\n  </p>\n </li>\n <li>\n  <p>\n   <strong>\n    <span style=\"color:rgb(250, 82, 82)\">\n     Hires. fix:\n    </span>\n    Enabled\n   </strong>\n  </p>\n </li>\n <li>\n  <p>\n   <strong>\n    <span style=\"color:rgb(250, 82, 82)\">\n     Hires Steps:\n    </span>\n    15 ~ 30\n   </strong>\n  </p>\n </li>\n <li>\n  <p>\n   <strong>\n    <span style=\"color:rgb(250, 82, 82)\">\n     Upscaler:\n    </span>\n    4x_NMKD-YandereNeoXL_200k\n   </strong>\n   - You can test others.\n  </p>\n </li>\n <li>\n  <p>\n   <strong>\n    <span style=\"color:rgb(250, 82, 82)\">\n     Denoise Strength:\n    </span>\n    0.4\n   </strong>\n  </p>\n </li>\n <li>\n  <p>\n   <strong>\n    <span style=\"color:rgb(250, 82, 82)\">\n     Upscale by:\n    </span>\n    Preferably x2\n   </strong>\n  </p>\n </li>\n <li>\n  <p>\n   <strong>\n    <span style=\"color:rgb(250, 82, 82)\">\n     CFG Scale:\n    </span>\n    6 ~ 9.5\n   </strong>\n  </p>\n </li>\n <li>\n  <p>\n   <strong>\n    <span style=\"color:rgb(250, 82, 82)\">\n     Clip Skip:\n    </span>\n    <span style=\"color:rgb(134, 142, 150)\">\n    </span>\n    2\n   </strong>\n  </p>\n </li>\n</ol>\n<p>\n</p>\n<p>\n <strong>\n  <em>\n   <u>\n    <span style=\"color:rgb(250, 176, 5)\">\n     Important:\n    </span>\n   </u>\n  </em>\n </strong>\n</p>\n<p>\n <strong>\n  <span style=\"color:rgb(130, 201, 30)\">\n   Positive prompt\n  </span>\n </strong>\n must contain:\n <strong>\n  <span style=\"color:rgb(130, 201, 30)\">\n   score_9,score_8_up,score_7_up,\n  </span>\n </strong>\n</p>\n<p>\n <strong>\n  <span style=\"color:rgb(230, 73, 128)\">\n   Negative prompt\n  </span>\n </strong>\n can be empty or contain the default pony negatives.\n</p>\n<p>\n</p>\n<p>\n The promting format from Pony Diffusion V6 works well:\n</p>\n<pre><code>score_9, score_8_up, score_7_up, just describe what you want, tag1, tag2</code></pre>\n<p>\n If you haven't already, I highly reccomend reading through the\n <a href=\"https://civitai.com/models/257749/pony-diffusion-v6-xl\" rel=\"ugc\" target=\"_blank\">\n  Pony Diffusion V6 XL model description.\n </a>\n</p>\n<p>\n</p>\n<p>\n</p>\n<p>\n I use\n <a href=\"https://github.com/lllyasviel/stable-diffusion-webui-forge\" rel=\"ugc\" target=\"_blank\">\n  Forge\n </a>\n as my main UI because it has so many optimizations for running XL/Pony models on lower end GPU's despite me having an RTX 4090, generations are alot smoother. I'm not sure how this model performs on other UI's.\n</p>\n<p>\n</p>\n<p>\n I haven't done much testing. If there is anything amiss, please let me know. Feedback is always appreciated :).\n</p>\n<p>\n Post results and, most importantly, have fun!\n</p>\n",
    "model_url": "https://civitai.com/models/543369",
    "model_type": "Checkpoint",
    "model_tags": [
        "anime",
        "base model",
        "girls",
        "ponyxl",
        "ponyv6"
    ],
    "download_link": "https://civitai.com/api/download/models/822931",
    "preview_image_url": "https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/f1a7fa5b-d5b2-49fb-8ff5-3798dd04a0df/width=450/28526922.jpeg",
    "model_version_id": null,
    "model_version_download_link": null,
    "model_version_image_url": null
}