AI porn models for adult image and video generation
Browse the building blocks behind the factory. Filter checkpoints and LoRAs by image or video support, base model, and tags, then open any version directly in the unified generator.
Sex, Nudes, Other Fun Stuff (SNOFS)
SNOFS is a natural language trained model focused on generating images of sex, nudes, and related content. It excels at understanding full sentence prompts to create images with various sexual acts, positions, and scenarios. The model is flexible and can be used with other LoRAs for specific likenesses.
Instagramification
The Instagramification LoRA creates images of women resembling Instagram influencers. It enhances features like lips, skin tone, hair, and makeup, often adding large breasts, based on datasets from popular Instagram Baddies LoRAs and trained on Wan2.1-T2V-14B, working for both text-to-video (T2V) and image-to-video (I2V) applications.
[ZIT] Mystic XXX
The Mystic XXX model unlocks enhanced capabilities on Z Image Turbo. It provides specific improvements to the model's performance on the older ComfyUI version 0.3.75, with adjustments made for newer versions. V3 or bigger gives the most accurate results.
Wan NSFW Posing Nude
This model generates NSFW nude portraits of standing, posing figures, focusing on anatomical accuracy and physics in a photoshoot setting. It was trained on a dataset designed to minimize bias toward specific facial features or accessories, allowing users to control details via prompts. It works best with specific aspect ratios and benefits from the use of lightning LoRAs for efficient generation.
NSFW POV All In One SDXL (Realistic/Anime/WD14 - 74MB Version Available)
This NSFW model generates realistic and anime style POV images focusing on basic scenarios like blowjobs, butterfly sex, cumshots, doggy style, and others. It comes in two versions: a higher quality 1.78GB version and a smaller 74MB version. Watermarks may appear but can be avoided by adjusting LoRA weight or seed.
Xpenis
Xpenis is a model trained with high-resolution material (832x1216 or 1216x832) focusing on penis imagery. It requires an external VAE and works best with SDXL base or low CFG settings. Key terms include flaccid and erect penis variations.
Hand in Panties - Masturbation - Wan 2.2 - Official Release!
Wan 2.2 is a specialized model for generating images depicting masturbation, specifically a hand in panties. It requires both high and low models and uses specific trigger phrases. The workflow is provided in the training data file.
Big Clit - WAN, Pony, IL, SDXL, Flux
This model generates images featuring a prominent clitoris, specifically an improved version (v2) built on the Pony model. It aims to create more realistic and aesthetically pleasing results compared to the original, addressing issues like unintentional penis-like features. The trigger words are "big clit" and "big clitoris".
(CONCEPT) Fellatio gesture with finger circle / フェラ素振り(指で輪っか) (Illustrious)
エッッッッッッ🍄💦💦なジェスチャーを作りたかったので作りました。 This LoRA provides the fellatio gesture of making a circle with the fingers to seduce. Trigger word:fellatiogesturewithfingercircle, fellatio gesture, finger circle, oral invitation, hand on mouth, Strength:0.9
Sex Smash Cut [WAN 2.2]
Sex Smash Cut [WAN 2.2] is a pair of LoRAs trained on WAN 2.2 I2V 14B models to generate images of a man thrusting into a woman's vagina. They work best on wider aspect ratios and may occasionally produce body horror.
Doggy Style Sex - Official Wan 2.2 Release!
Wan2.2 T2V v4 is a model focused on generating images and videos of a man and woman in the doggy style sexual position, now with POV support. It uses specific triggers and prompts to generate these images/videos, including options for front and side views. The model requires both high and low components to function correctly.
WAN DR34ML4Y - All-In-One NSFW
WAN DR34ML4Y is an all-in-one NSFW LoRA model designed to handle various specific concepts like double blowjobs, missionary position, reverse cowgirl, and doggy style. It accurately generates images based on specified keywords and positions. It is recommended to use both models to produce ideal outputs.
Wan POV Blowjob
The Wan POV Blowjob model generates images and videos of oral sex from a first-person perspective. It is capable of depicting cumshots and can handle images where the penis is not initially present. The model supports multiple angles and requires specific prompts to trigger certain actions, like the appearance of cum.
Wan 2.2/2.1 POV Missionary
Wan 2.2/2.1 is a model designed for generating images and videos of missionary-position sexual activity. The model comes in T2V (text-to-video) and I2V (image-to-video) versions, with improvements in handling different resolutions and incorporating more realistic movements. Prompting tips are provided to control elements like movement speed and breast size.
WAN 2.2 Anime Cumshot Aesthetics – Precision Load I2V (Beta version)
WAN 2.2 Anime Cumshot Aesthetics – Precision Load is a LoRA designed to enhance the density, precision, impact, and aesthetic realism of cumshots in anime-style scenes. It focuses on thickness, directional accuracy, and artistic expression, delivering impactful visuals and can pair with live-action content. This beta version features improved motion and detail through high and low noise models, but may still produce some unnatural results.
Wan Cumshot (2.2 / 2.1)
Wan 2.2 uses highnoise and lownoise LoRAs, while Wan 2.1 is tailored for Image-to-Video (I2V) and Text-to-Video (T2V) applications. Wan 2.1 (I2V) allows for origin points outside the frame, while T2V allows for the generation of specific scenarios and actions, including the generation of intimate acts.
Wan POV Doggy Style (i2v)
This model is trained for image-to-video generation of doggy style scenes, specifically optimized for vertical videos and using the trigger word "POVdog." Version 1.1 provides consistent motion for the doggy style position, with emphasis on ass movement and bounce, although it may have image artifacts due to the low resolution source video data. The model reportedly works well in conjunction with aipinups69's Wan 2.1 LoRA at a strength of 0.5.
WAN2.1 Facefuck
WAN2.1 Facefuck is a LORA model focused on generating images of a specific sexual act. It was trained on low-resolution videos to exaggerate movement and allows control over insertion speed and depth. The model aims for further refinement in future versions to improve realism and add scene variations.
Side lying Sex - Wan I2V 14B
This LoRA model generates images and videos of side-lying sexual intercourse. It has two versions, one for text-to-video (t2v) and one for image-to-video (i2v), with the i2v version recommended for better performance. The i2v version can be triggered using "They are having side sex."
Slop Twerk WAN 2.2 I2V
Slop Twerk WAN 2.2 I2V is a LoRA trained on AI-generated videos of characters twerking, focusing on generating ass shaking and jiggling effects. It was trained using a modified Musubi Tuner GUI and the dataset includes manually tagged videos with descriptions of twerking actions. The LoRA is designed to be used with prompts related to twerking and ass shaking, with some breast jiggle also present.
NSFW/Female Genitals helper for Wan T2V/I2V
This Lora model enhances female genital generation, specifically addressing Wan's undertrained areas. It was trained on a dataset of images and videos to improve detail and motion, and the trigger words are "Innie vagina" or "Outie vagina." It's designed for nude poses in standing, sitting/lying, and bending positions.
creampie (nakadashi/cum inside) concept WAN2.1 I2V
This model, a creampie concept WAN2.1 I2V, generates images involving vaginal penetration. It focuses on accurately depicting scenes of cum dripping from a vagina, although penis depiction may require prompt adjustments. It aims to improve text-to-video generation by understanding female anatomy better.
(wan 2.2 experimental) WAN General NSFW model
Explore available lora versions and generation modes.
Wan Front View Doggy Style (i2v)
This model generates image-to-video content depicting doggy style scenes with the woman facing the camera. It focuses on realistic faces and intense action, using the trigger word "Pl0wView" for prompt generation. The model was trained using diffusion-pipe on a 480p i2v 14B model.
How checkpoints and LoRAs fit together
Start with a checkpoint
A checkpoint controls the base rendering system. Its supported modes determine whether it can create images, edit an input, or produce video.
Match the base model
A LoRA must target the same base-model family as the checkpoint. Similar names do not guarantee compatible weights.
Check the validated workflows
A catalog version can be documented before it is generator-ready. Only versions with validated broker and artifact mappings can spend credits.