CivArchive
    Middlefinger + Ahegao + Drooling - high noise
    NSFW

    It felt like a waste to train a lora just for showing the middle finger so I added ahegao with drooling as a second concept. You can use either concept of both at the same time. The concept work quite well except for the crossing eyes part.

    The Triggers are:

    • Shows her middle finger, flipping off the viewer

    • Sticks out her tongue and rolls her eyes

    • she/he is drooling

    Description

    FAQ

    Comments (19)

    BarelyAISep 25, 2025· 5 reactions
    CivitAI

    Great stuff. Hope you keep making WAN 2.2 Loras 👍🏻

    ffjggrtbjibv
    Author
    Sep 25, 2025· 3 reactions

    I have 4 more loras ready to publish. I plan to upload them in the coming days

    BarelyAISep 25, 2025

    @ffjggrtbjibv 🙏🏻😁

    SheebusOct 7, 2025

    @ffjggrtbjibv still plan on posting them?

    fluxxesSep 25, 2025· 1 reaction
    CivitAI

    awesome

    RealHUGESep 26, 2025· 3 reactions
    CivitAI

    Thank you for this! You read my mind! It's perfect!

    Thank you also for your other work, which I've used a lot: Tgirls LoRa for SD1.5 and SDXL. For a long time, they were the only ones capable of generate women with realistic penises. The CockXL model is also excellent. You've left your mark on the history of T-Lovers.

    AlbanianDoorknobSep 29, 2025

    *men

    rjoxSep 26, 2025· 2 reactions
    CivitAI

    impressive results. likeness stays in tact really well

    shinizayaSep 26, 2025· 1 reaction
    CivitAI

    Great lora, thank you!
    Fyi the high noise lora is categorized as SD1.5

    ffjggrtbjibv
    Author
    Sep 26, 2025· 1 reaction

    Thanks for the heads up, it's fixed now

    JellaiSep 27, 2025
    CivitAI

    Seems to be the best middle finger action lora for Wan, and that's surprising, because it also contains two other concepts. Really good work.

    danielamocanukty333Nov 5, 2025
    CivitAI

    can someone explain high/low noise wan to me? If I use both, they always give broken images

    tadirel864375Nov 9, 2025

    Load High Model -> High Loras -> KSampler high pass; Low Model -> Low Loras -> KSampler low pass, chained from the KSampler high pass -> Output (Like video decode or whatever); Low loras strength should be ~half of the high ones

    tadirel864375Nov 9, 2025
    CivitAI

    How to fix the face completely changing in I2V when using this lora? Always completely changes the original face in the image making it a lot more plain - am I doing something wrong? Tried High:0.7/Low:0.5 and around there but nothing seems to help...

    artyclawNov 27, 2025

    Same for me with smoothMix. Are you using it as well? Worked fine with regular wan 2.2.

    ffjggrtbjibv
    Author
    Nov 29, 2025

    I never had this problem, something must be wrong with your settings, or are you using a merged checkpoint? These sometimes degrade quality allot especially if you use the low noise model. Also you should use the low noise model of this Lora at close to 1, 0.5 is too low

    JalmirJan 13, 2026

    I'm also getting face change, and it's VERY obvious, like a wrap within the first 2-3 frames, using Remix model. I'll try another model to see it that helps!

    pernillestruve7195Dec 14, 2025· 2 reactions
    CivitAI

    bad with crossing eyes. Doesn't roll eyes at most of the results :(

    TommyVVFeb 6, 2026
    CivitAI

    Rolling eyes doesn't work, except for this which is practically perfect.

    LORA
    Wan Video 2.2 I2V-A14B

    Details

    Downloads
    5,882
    Platform
    CivitAI
    Platform Status
    Available
    Created
    9/25/2025
    Updated
    6/17/2026
    Deleted
    -

    Files

    flipOffAhegao_high_noise.safetensors

    Mirrors

    Available On (1 platform)

    Same model published on other platforms. May have additional downloads or version variants.