CivArchive
    C666D Anal and Vaginal Fisting Couple Moss - C666D Fisting Moss Image
    NSFW
    Preview 149133701
    Preview 149133696
    Preview 149133708
    Preview 149133709
    Preview 149133716
    Preview 149133718
    Preview 149133721
    Preview 149133722
    Preview 149133725
    Preview 149133726
    Preview 149133730
    Preview 149133732
    Preview 149133734

    C666D Anal and Vaginal Fisting, with support for multiple people and poses.

    This model can do females, males (but was not trained on any male anal penetration content), and works with other lora's, including characters and styles, but also other concepts like bodytypes, make-up, outfits and more.

    The model is also useful for multi-character pose control, and clothing control on each character if you follow the same prompt structure I used, meaning without nsfw too.

    So check the examples, try to understand how it works and you'll be able to do nearly anything with it that involves multiple characters. ❤️

    Trigger / Starting phrase: 🔥🔥🔥

    1. a pornographic erotica photo of: ( -> followed by prompt).

    2. anal fisting with a arm inside the ass, and anus stretched around it.

    Helpful notice: 💌💌💌

    1. When using TURBO models removing the word "anus" is recommended, with SFT it works normally.

    2. When a female is lying on her back, adding the word "pussy", helps prevent vaginal fisting. 😼

    Prompt Building / Scene Descriptions: 🎥🎞️🖼️
    It is required to describe each character in a single paragraph, one by one, including the objects they interact with. Always describe the scene from left to right, ( <- to -> ), followed by the background, and scene details such as lighting, style,

    Character Descriptions in Paragraphs: ♂️♀️🎬

    Start by decribing the 1.side (left/right), 2.haircolor, 3.character, 4.gender, 5.direction they face, 6.pose, 7.the action, 8.facial expression, 9.outfit, and 10.the angle, 11.furniture or object. and try to keep it within a single paragraph / sentence.

    Use comma's (",") to seperate everything in a single paragraph, and close with a dot (".").
    This order is how I trained and captioned the dataset, and seems to works best.

    Framing:

    Everything that's captured within the scene is refered to as "the frame". So describe the scene as a movie director.

    1. Who does the fisting: The "male/female" leans forward into the frame with arm bent at elbow, or arm stretched, and reaches out.

    2. Who is fisted: The "male/female" is (add pose) in the middle of the frame.

    Poses:

    1. Standing.

    2. Sitting.

    3. Kneeling.

    4. Squatting.

    5. Lying.

    6. Bent over.

    7. Legs spread up in a yoga pose.

    Angles:

    1. from above with a high angle.

    2. from below with a low angle.

    3. from side with a sideview profile.

    4. for frontal don't describe the angle.

    What or who they look at:

    1. When a single character looks at the other character use: they face the other (female/male), within their own paragraph.

    2. When a single character looks at the viewer use: (and is looking at me directly), within their own paragraph.

    3. When both characters look at the viewer use: (they are both looking directly at me), before the first or after the last character description.

    4. With any other object add the object's location: (s/he is looking at the object inbetween them, below/above/beside him/her).

    Short Example Prompt:

    a pornographic erotica photo of:
    a couple of women, anal fisting with a arm inside the ass, and anus stretched around it.
    On the left a blonde woman kneels on the floor facing to the right, is leaning into the frame as she bents over forward, her arm bent at the elbow, while reaching out to the other female with a grin on her face, wearing a flirty dress.
    On the right a brunette woman is lying on her back with her legs spread apart on a bed, her barefoot feet up beside her head, and her arms back behind her thighs, pussy exposed, her chin lifted and as she moans with pleasure.
    They are inside of a modern bedroom, with playful vibrant colors and cinematic lighting, the light shines into the room from the window and hits them from the left.

    Testing and Usage: 🤖🤖🤖

    1. the model works best inbetween a 1:1 1MP resolution, to a 2:3 1.4MP resolution, at higher resolutions the poses and actions stop working.

    2. Extreme angles might bug, extreme low/high,

    3. The yoga pose doesn't work from behind.

    4. Sometimes it can create a "pussy" on the male's testicles, so put this in negative.

    5. I've been able to control a maximum of 8 characters, by describing the main actions of the first 4, and having the rest behave as a group, so behind him/her a group is performing action "X".
      It can influence bodytypes, and sometimes makes female's skinnier if their bodytype is not defined, using a prompt, bodytype or character lora fixes this.

    6. With tattoos you can get certain patterns, which Is something I should fix in V2, but doesn't bother me that much.

    7. The triggers aren't really necessary, but help guiding the model when the poses aren't described very well. But when you don't add the bent/stretched arm, a disembodied fist-shaped object might appear inside of the anus.

    Afternote:
    For the examples I used Moss-SFT_int8 basemodel, CLIP: qwen3.5_9b_heretic_v2_int8_hq from Artius, VAE: flux2-vae, euler-euler_a-ERSDE, simple-sgm, 30-50 steps, and CFG: 3-3.5, so experiment, Moss is pretty flexible.

    All captions have been done manually, and took me over 16 hours. (Yes I spent a entire weekend on this, and I'm absolutely questioning all my life choices right now). But it's also a proof of concept to show the spacial and reasoning capabilities of MOSS.

    I also added danbooru style tokens, which can help if prompting like a normal sentence is too complicated, but honestly, this lora takes away most of the hassle of having to prompt as a 18th century shakesperean poët just to see a action, it's blunt, simple, and efficiënt when used properly.

    No artificial data. No anime, 2D, or 3D, painting, etc.etc. Does not create a real person's face. Intended for photographic quality. No solo/1person images in dataset.


    Help me power up: ⚡👍⚡

    Hit the like button on the model page, and on my images if you enjoy what I create! And posting a image to the modelpage grants me ⚡50 blue buzz for each user that uploads to a model. Also saving my images to your gallery helps me out a little bit! So I would appreciate it alot if you could spare a few seconds! ⏰ I'm almost out of buzz AGAIN, because I keep spending it all on training, and every bit helps so I can keep going! Donations are welcome too! ❤️❤️❤️

    Description

    v1. manually captioned, photo quality. Works best using MossBase-SFT. 1/1.4MP.
    Resolutions: 1:1, 2:3.

    LORA
    MOSS Image

    Details

    Downloads
    19
    Platform
    CivitAI
    Platform Status
    Available
    Created
    11/15/2026
    Updated
    11/18/2026
    Deleted
    -
    Trigger Words:
    a pornographic erotica photo of:
    anal fisting with a arm inside the ass, and anus stretched around it.
    The "male/female" leans forward into the frame with arm bent at elbow, or arm stretched, and reaches out.

    Files

    c666d-couple-anal-fisting-moss-v1-moss-image-base-epoch-10.safetensors