CivArchive
    Wan2.2 - Anal (Side View) T2V/I2V - High Noise v1.0
    NSFW

    Features

    • Side view anal penetration

    • Trained on deep and quickly penetration

    • Compatible with both T2V & I2V Wan2.2 Models

    Tips

    • Strength between 0.5 to 1 is good. Lower strength on Low Noise model 0.5 to 0.65 if the face is changing or you experience artifacting.

    • T2V Low Noise Model may need to be below 1. The model is crispy with 1 strength.

    • HIGHLY recommend reviewing and understanding how to correctly split steps for Wan2.2 models: https://www.reddit.com/r/StableDiffusion/comments/1mkv9c6/wan22_schedulers_steps_shift_and_noise/?tl=fr

    • Workflow that correctly splits High / Low noise steps: https://github.com/stduhpf/ComfyUI-WanMoeKSampler

    • Suggested prompts. These prompts were used during training for the image/video text captions 'a man is penetrating a woman's anus with his penis', 'a large penis is penetrating her anus', 'man's penis is deeply penetrating a woman's anus', 'a man is inserting his penis into a woman's anus'

    My Wan2.2 Cover Videos

    Training

    Definitely one of the most challenging training sets. Constantly fighting the base model to learn. It was like herding a bunch of third generation escaped house cats over the wild rolling American plains while blindfolded, barefoot and lit on fire.

    Trained on Wan2.2 T2V base model. 90 high quality images and 40 videos that were handpicked and farmed raised to perfection using musubi-tuner. This LoRA works with both I2V and T2V. A good portion include 'normal' penetration anal the rest are 'deep' anal. Not trained on multiple penisesses

    Limitations

    • Struggles when el penis is not already visible in the scene

    • Struggles when el penis not already in the anus, e.g. this may work sometimes -> 'man inserts penis into womans anus'

    • Struggles when the man pulls out la penis. All shaft no penis tip.

    Description

    FAQ

    Comments (15)

    honryindianSep 7, 2025· 4 reactions
    CivitAI

    The model name says I2V/T2V, but does it really work with T2V? Working really good with I2V, but with T2V only blurry mess :)

    Kierkegaard420
    Author
    Sep 7, 2025

    @honryindian It should work with T2V. It was specifically trained on the T2V base model. The T2V base model works with both I2V and T2V base models. Make sure you are doing the steps correctly. You may need to adjust the strength by lowering it.

    Kierkegaard420
    Author
    Sep 7, 2025· 2 reactions

    @honryindian Yeah, just ran a couple tests. The T2V model is a bit crispy if you have the Low Noise LoRA strength at 1. Lower the Low Noise LoRA strength to below 1. I just randomly tested 0.8 and its much better then 1.

    Thank you letting me know.

    ClocksmithSep 7, 2025

    @Kierkegaard420 A T2V video example with metadata would be much appreciated.

    Kierkegaard420
    Author
    Sep 7, 2025

    @Clocksmith I dont normally use T2V so its not a polished workflow and it is missing the MoE KSampler that you really should be using. I talk about the MoE KSampler on the model page but here is the metadata that I just tested, https://civitai.com/posts/21950973.

    I just through this together in about 5-minutes for this comment so its not great. Again, its not going to look like the cover video because those are I2V. I have the I2V workflows posted on the model page.

    SD_AI_2025Sep 9, 2025

    You wrote: "It was specifically trained on the T2V base model."
    But Civitai states in the description:
    "Base Model Wan Video 2.2 I2V-A14B".
    I always trusted that... Damn :/

    Kierkegaard420
    Author
    Sep 10, 2025

    @SD_AI_2025 Yes, because Wan2.2 T2V trained LoRA models can be used for I2V and T2V. Its setup as I2V because you can run it on the I2V base model. It works for both base models.

    ClocksmithSep 7, 2025· 6 reactions
    CivitAI

    Can you please upload a video example with metadata? I just spent an hour trying to follow the advice in your Tips section. I got the suggested workflow but it didn't have the lightning lora (or any loras) so the gen speed was unacceptable. I tried rigging it in but the output was garbage. It would be a lot faster if you just provided a workflow example of your own lora functioning through metadata.

    Kierkegaard420
    Author
    Sep 7, 2025· 1 reaction

    @Clocksmith The workflows are posted on the model page. I use multiple workflows.

    zoom83Sep 9, 2025

    @Kierkegaard420 for me it looks also blurry in default Workflow. Tried moe k-sampler but looks wired. Workflow Sharing would be appreciated.

    TheLastRemainSep 11, 2025· 1 reaction
    CivitAI

    Funny thing is this lora is the first one that allowed to generate an action clip for a 'pussy' missionary position where the woman is lying diagonal, seen from above, slightly behind the man. I noticed however that the details of the man's penis is lost in the movement, (high noise @ 1.0, low noise @ 0.65)

    Could it be because of the 1.0 value in the high noise, or something related to some seeds ?

    Kierkegaard420
    Author
    Sep 12, 2025

    I think I need to review the dataset to handle El Penis, better. Add more videos / images with The PENIS

    TheLastRemainSep 13, 2025

    @Kierkegaard420, I will try some other settings, for now I changed the view to a more bottom/top POV image instead of a diagonal view which makes the details penis stand out and looks overall much better.

    The results are still not on par with my intent, as deep and fast penetrating strokes are not coming out the way I want, but I think that is probably fixed by tinkering on the prompt(s) as I found out on other clips.

    MMOFanSep 14, 2025· 1 reaction
    CivitAI

    This is great! Thanks for making it.

    jasonFly75Jan 14, 2026
    CivitAI

    Good lora but, the quality of the image, the saturation, and the contrast is weird, kinda grey and a little bit blurry, but good job on the animation and anatomy, if some one have a solution for the "grey" quality, would be great.

    LORA
    Wan Video 2.2 I2V-A14B

    Details

    Downloads
    8,865
    Platform
    CivitAI
    Platform Status
    Available
    Created
    9/7/2025
    Updated
    5/14/2026
    Deleted
    -

    Files

    Wan2.2_Anal-v1-HighNoise-I2V_T2V.safetensors

    Mirrors