Porn Factory AILogin

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.

1487 models available

LoRA · SD

Open towel

212
Open towel is a LoRA trained to generate characters holding an open towel. It can also be used to create an 'open' effect with other clothing. The trigger word is "open towel".animeclothesclothingconceptualnudetowel

LoRA · IL

Mutual Masturbation (Standing) POV - Illustrious

012
Illustrious is a LoRA model for generating images of mutual masturbation, specifically in a standing pose. It uses trigger words like "hetero, 1boy, arms, pov, standing, penis handjob, pussy, fingering another" and is most effective at full strength. While designed for a standing pose, complex finger poses may still be distorted.concept

LoRA · SDXL

DMD2 Speed LoRA [SDXL, Pony, Illustrious]

111
This is a LoRA implementation of the Distribution Matching Distillation (DMD2) method for fast image synthesis with SDXL. It enables image generation with fewer steps while maintaining quality, especially when using LCM sampler and a low step count. Recommended settings include 8 steps, LoRA weight of 0.7, and a CFG between 1.4-1.8; it can be merged into a checkpoint for direct usage.dmd2hyperlcmloraspeedtoolturbo

LoRA · SDXL

Da Vinci Ink sketch

011
This model creates images of biomechanical figures and animals in the style of Leonardo Da Vinci's ink sketches. It generates anatomical and medical studies with a steampunk or cyborg twist, often incorporating annotations and a sepia tone. The model responds well to keywords like "biomech," "anatomical study," and "sepia" to influence the image's style.biomechanicalcyborgda vincidrawinginksketchstyle

LoRA · SD

Public restroom

211
This LoRA model generates images of male public restrooms. It was trained to produce public restrooms rather than private bathrooms, using both SFW and NSFW data. The trigger word is "restroom," and optional tags include objects typically found in restrooms.backgroundbackgroundsbathroomindoorsrestroomurinal

LoRA · SDXL

publicmix_zoya

011
Mix numerous popular female faces to create suitable realistic faces. The sources are all from AI materials. Name ZOYA to help create a sense of intimacy with real people, ZOYA is not a real person.animecharacterfemalegirlsphotorealisticsexywoman

LoRA · SD

Style Squeeze [LoRA]

011
Style Squeeze is a LoRA model that enhances image generation by improving color, composition, and clothing styles. It aims to inject "coolness" into generated images. It is not ideal for generating text or text-like shapes.helperstylestylishtransform

LoRA · IL

Removing Condom | Goofy Ai

011
This model facilitates the removal of condoms in generated images. Use it with a LoRA strength of 0.7-1 and ADetailer for faces. Upscaling with 4x-ultra sharp is recommended.animeconceptcondomcondom on penisnsfwremoving condomsexwoman

LoRA · SD

Skirt tug / dress tug / clothes tug

311
This LoRA generates images of characters tugging at their clothing, such as skirts, dresses, shirts, and sweaters. It can be triggered using terms like "clothes tug," "skirt tug," or "dress tug," and allows for variations like "clothes pull" or incorporating environmental factors like wind.animeclothes tugclothingdress tugposeskirt tug

LoRA · Flux.1 D

Realistic Amplifier for UltraReal Fine-Tune

011
Not much to say about this LoRA - at strength 0.9-1.0, it gives an effect similar to old Nokia phone photos. At strength 0.5-0.7, it simply amplifies the realism of my checkpoint. I didn’t plan to upload this for a while (thought no one would need it), but after getting some PMs asking about it, I decided to share. I use this exclusively with my own checkpoint. For flux.dev it just serves as a minor realism enhancement.realisticstyle

LoRA · IL

Sos Adult style | Goofy Ai

111
The Sos Adult style | Goofy Ai model generates images in an adult style. Key features include the need for Adetailer for faces and the suggestion to use img2img upscaling with 4x-ultra sharp. It is recommended to use the Lora at a strength of 0.7-1.animesos adultstylestyleswoman

LoRA · IL

[BSS] - Korean girls (Illustrious)

111
This model, BSS - Korean girls (Illustrious), is designed to generate imaginative and creative images. It encourages diverse styles beyond the provided preview images. The model is copyrighted and unauthorized use is prohibited.style

LoRA · IL

Cunnilingus on Back - Concept

011
This model generates images featuring cunnilingus performed on a person's back. Key features include depictions of nudity, sexual acts, and associated details such as blush, sweat, and specific body parts.concept

LoRA · IL

Female Masturbation Infront Of Laptop - Concept

111
This model generates images depicting a female masturbating in front of a laptop. Key features include depictions of blush, sweat, and various details of the scene like clothing, body parts, and objects present. The trigger word is female-masturbation-infront-of-laptop-illustriousxl-lora-nochekaiser.concept

LoRA · SD

Murky's - Dangling Spitroast LoRA

311
This LoRA model generates images featuring a specific sexual pose referred to as "DanglingSpitroastMS." It requires the use of trigger words and base sexual prompts for image generation.pornposes

LoRA · IL

Nuru Massage (Illustrious)

110
This LoRA model generates images of women covered in large amounts of sticky lubricant, typical of nuru massages. It enhances the 'nuru', 'slime', and 'lotion' effects in generated images, with a recommended strength of 1 for optimal results. Using a higher weight than 1 can negatively affect the image quality.animeconceptgirlshentailoralotionlubemassagemessynuruoiledsexslimyslipperystickywoman

LoRA · IL

[Modifier] Face Humping on back fellatio

010
This model generates images of fellatio with a focus on a "face humping" variation where one person is lying down. The activation tags are "oral" and "lying," with an optional modifier, "sitting on breasts," still in development. This is an experimental release intended for continuous refinement.conceptfacefuckingfellatiomidimidison back

LoRA · SD

Violent Explosion Style [LoRA 1.5+SDXL]

010
Trigger: ral-vlntxplzn ☕ Buy me a coffee: https://ko-fi.com/ralfingerai (https://ko-fi.com/ralfingerai) 🍺 Join my discord: https://discord.gg/g5Pb8qNUuP (https://discord.gg/g5Pb8qNUuP)explodingexplosionexplosionsexplosivestylestyles

LoRA · SDXL

Lunatic Portrait

010
editorialfashionphotogrphyportraitretrostyle

LoRA · Wan Video 2.2 I2V-A14B

Doggy anal cut

310
My very first LORA ! Initially made mostly for learning, model is inspired by other LORA that directly transition from an image to a sexual scene. In this case an anal scene viewed at a side angle or directly from behind. It can do a few different poses: Man can be behind (doggy) or "straddling" over her with his legs spreads. I am not sure if straddling is the right term, see examples. The woman can have her head on a chair/sofa/bed or directly on the floor. It currently isn't perfect for likeness and the woman with her head on the floor and man straddling over her will typically have trouble conserve likeness. "Head on chair" + "Man behind her" will conserve likeness better . It can conserve some elements of clothing especially on the head or upper body (bra, tank top, glasses etc.). It can sometimes work for panties down the thighs or skirts etc. but it is more hit and miss. Additionally, it is currently lacking in diversity and as a tendency to produce very similar large ass and similar genitals, I invite you to try controlling that through more prompting and extra LORAs to change the shape. Some stuff for me to think about a version 2.0 I guess. Suggested config 0.8 High 1 Low strengths. Here is a prompt with wild card you can use to test it out which summarizes the above: The scene immediately transitions to a scene where the same woman is on all four {on a red chair|head against a bed|head on the floor} while a {black| } man is {roughly| } fucking her ass. She is {completely naked with her jeans and panties down her thighs|completely naked|naked from the waist down with her tanktop|naked with her black lingerie panties down her thighs}. The man is {behind|straddling above} her. {Back|Side} view. This was trained on real videos only (no anime) and I haven't really tested it for anime. All samples were generated with WanVideo nodes, lightx2v 10 steps (5/5) cfg 1 shift 6 dpm++_sde. Interpolation GIMM-VFI and the workflow can be found in the files. Use responsibly and for entertainment only.

LoRA · NAI

Contortion Suspended Hanging - Concept (Illustrious)

110
This model, Contortion Suspended Hanging - Concept (Illustrious), generates images of figures in suspended, contorted poses with bound limbs. It utilizes LoRAs to achieve specific aesthetic and detail enhancements based on a defined prompt and negative prompt. The model is designed to work with UnholyDesireMixSinisterAesthetic_V5_Illustrious base model.arms behind backbdsmbound legscapturedconceptcontortiondefeatedfoldedhanginglegs upoverpoweredpublic displayshibarisuspendedtied up

LoRA · Wan Video 2.2 T2V-A14B

POV Titty Fuck -T2V & I2V

010
This model generates images and videos based on text prompts, specifically focused on the "tit_fuck" theme. It supports both text-to-video (T2V) and image-to-video (I2V) generation related to this theme.concept

LoRA · IL

girl restrained by large monster (illustrious, pony)

010
lora to generate a girl restrained by a large monster with large hands and penis. strength ~0.8.concepthentailarge monstersize difference

LoRA · SD

Murky's - Thigh Sex LoRA

310
This LoRA model enhances the "Thigh Sex" concept in images. It was trained with 400 images and works on various poses, with prompts like "thigh sex," "grinding," and related terms recommended. Use a weight of 0.8 to 1 depending on the base model used.animeposessexthighs
Load more models

Showing 24 of 1487

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.