CivArchive
    Preview 47976392
    Preview 47976497
    Preview 47976451
    Preview 47976500
    Preview 47976528
    Preview 47976588
    Preview 47980917
    Preview 47976791
    Preview 48332919
    Preview 48332922
    Preview 53246189

    This Lora replicates the style of the artist Soranamae.
    https://www.pixiv.net/en/users/3000666

    It tends to generate NSFW content.
    Lora responds to the word "soranamae_style".
    The dataset contained cropped images, so you can add phrases like "head out of frame, white space" to the negative prompt, but it's not mandatory

    V2
    In this version, there is less noise, and the backgrounds are slightly improved.
    I recommend generating at a higher resolution (the dataset had a resolution of 1280x1280) and using ADetailer.

    Flux version
    To activate the style, simply use the tag soranamae_style
    Please note that Flux D cannot create NSFW content, so it's recommended to use it in combination with my Lora along with something additional, such as this: https://civarchive.com/models/935673/nipplediffusion-flux
    To work with Lora and fine-tuned models, you need to adjust the scaling factor, which is the number written after the ":". This number varies for different models but is likely to be in the range of 1 to 3.5. I added an array in the second post for this model to make it clearer.

    For example, in <lora:soranamae-style-flux-test2:1>, the scaling is set to 1, meaning it has no amplification. However, in <lora:soranamae-style-flux-test2:3>, the effect of Lora is increased to 300%.

    Description

    Flux version

    To activate the style, simply use the tag soranamae_style
    Please note that Flux D cannot create NSFW content, so it's recommended to use it in combination with my LoRA along with something additional, such as this: https://civitai.com/models/935673/nipplediffusion-flux
    To work with Lora and fine-tuned models, you need to adjust the scaling factor, which is the number written after the ":". This number varies for different models but is likely to be in the range of 1 to 3.5. I added an array in the second post for this model to make it clearer.

    For example, in <lora:soranamae-style-flux-test2:1>, the scaling is set to 1, meaning it has no amplification. However, in <lora:soranamae-style-flux-test2:3>, the effect of Lora is increased to 300%.

    FAQ

    LORA
    Flux.1 D

    Details

    Downloads
    76
    Platform
    CivitAI
    Platform Status
    Available
    Created
    12/27/2024
    Updated
    5/15/2026
    Deleted
    -
    Trigger Words:
    soranamae_style

    Files

    soranamae-style-flux-V1.safetensors

    Mirrors