CivArchive
    Anime3.5L - V1
    Preview 39214124Preview 39214125Preview 39214132Preview 39214127Preview 39214131Preview 39214123Preview 39214128

    Here is a Form you can request me Lora there (for Free too)

    https://forms.gle/h9iuLCCnaEs6W9ZB8

    If you need help training your Lora or want to help me. Join my discord server.

    https://discord.gg/8ZvaEtJUy2

    This is a Beta Version of the Anime style Lora trained on SD3.5 Large.

    I'm gonna be training for 3.5 Medium as well. Sd3.5 Anime Model

    You can use Any generation service you want, I believe most of you can't run it locally so you could use online generators.

    You could try generating images in ShakkerAI it is cheaper than in civit ai you can check yourselves https://www.shakker.ai/modelinfo/e0be750a21324c62b69e535947c30aae/Anime3-5L-Lora-BetaVersion

    use the following settings

    1 image

    DPM++ 2M

    15 steps,

    4-7 guidance

    costs around 7 tokens per images you get 200 token per day for free in Shakker, this is not Promotion, I just want you guys to be able to use the Lora.
    I trained this Lora using CivitAI trainer.

    pos - anime_style, anime_girl, 1girl, <your prompts>

    neg - bad quality images, nsfw, porn, vagina, nipples, bare ass, nude, naked, completely nude

    Note:

    if you use captions it might generate realistic images. Don't use NSFW keyword.

    My Findings:

    While generating images I found that SD3.5 has been poisoned by the AI images in its training set. It generated some images like this when I used NSFW keyword

    Description

    v1

    LORA
    SD 3.5 Large

    Details

    Downloads
    373
    Platform
    CivitAI
    Platform Status
    Available
    Created
    11/9/2024
    Updated
    9/30/2025
    Deleted
    -
    Trigger Words:
    anime_style

    Files

    Anime3.5L.safetensors

    Mirrors

    Huggingface (1 mirrors)
    CivitAI (1 mirrors)