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

Oil painting(oil brush stroke) - 油画笔触

30
用一些狂放笔触的油画炼的，都是风景画，初学炼丹感觉效果一般般感兴趣的可以试一下，触发词（bichu） 机翻：Refined with some wild brushstrokes of oil painting, are landscapes, beginner alchemy feel the effect of general interest can try, trigger words (bichu). have fun :Dartoil paintingstylestyles

LoRA · Other

[LORA] Chinese Doll Likeness

30
This LORA creates images resembling Chinese doll-like characters. It was trained using a dataset of SNH48 idols and Chinese actors, primarily focusing on portrait photos. Be aware that full body images may exhibit facial distortions.woman

LoRA · SD

Helltaker LoRA

30
Helltaker LoRA If you want to reproduce the original Helltaker-like character, enter the trigger word 'Helltaker'. Alternatively, I recommend that you apply weights between 0.5 and 0.6 without applying trigger words. Or If you apply a weight of 0.3~0.4, It will emphasize the outline.animehelltakerstyle

LoRA · Wan Video 2.2 I2V-A14B

Slap (and Self Slap) - Wan 2.2 I2V / T2V

5470
Now updated with non self slapping material. Its also a little better at targeted slaps (like breast slapping) but not formally trained on that yet. Whether there's a bug on your face, or you are tired and need to wake yourself up.conceptself slapslap

LoRA · Wan Video 2.2 T2V-A14B

WAN2.2/WAN2.1 PAWG

00
The WAN21 - PAWG LoRA is a Stable Diffusion extension tailored for generating images with enhanced lower-body proportions. Optimized for the WAN 2.1 framework in ComfyUI, it allows users to create photorealistic or stylized renders emphasizing curvaceous hips, toned thighs, and plump buttocks. It offers adjustable intensity and works best with detailed prompts for various SFW/NSFW applications.buttconceptpawgthick

LoRA · Qwen

Qwen Edit Figure Maker By Aldniki

00
Qwen Edit Figure Maker is an upgraded model trained on Qwen Image Edit that reliably produces realistic figures in the style of Nano Banana (Gemini). It allows scene changes via prompt modifications and offers excellent Chinese support.3d figurineanime2figurefiguremodel renderingnano bananananobananapvcstyle

LoRA · Other

[NSFW]Tentacles LoRA | 触手

20
Check belowing images about comparison between no-lora, version1, version2, and concepts, and v2 compability amoung models. Compared to V1, I added general images (images without tentacles) generated with AOM3 model and caption them with "style of AOM3" in V2 training dataset, I hope this may help preventing style influence.hentai

LoRA · Other

Mesugaki - LoRA

00
This LoRA is trained on mesugaki images to enhance image quality. It may sharpen eyes and add a smug or lewd expression to characters. Personal testing suggests it acts as a mild quality enhancer.animeconceptfemalegirlssmugwoman

LoRA · Other

Pantyhose pull

20
This LoRA model generates images of characters pulling down pantyhose or panties in various poses. It differentiates between "pantyhose pull," "panty pull," and requires specific poses like bending over for "undressing" scenarios.animecharacterpantyhose

LoRA · SD

concept Testicle sucking

20
As a beginner, I made this because I didn't find lora about it. I'm trying to improve it.(But it has failed until now XD) It is still not stable,especially in pov.Just generate more. Setting weight 1.0 is ok. In the example figure, I used some characters Loras. If there is any offense, please contact me to delete themactionanimelicking penispenistesticle

LoRA · SD

POV Facesitting

60
This model generates low-angle "imminent facesitting" shots, and other low-angle images, trained on a dataset of custom images. The optimal settings are around 0.7-0.8, and the "from below" tag may improve consistency, but nose distortion is a known issue that may be addressed in future versions.camera angleconceptposesexy

LoRA · Other

breasts_on_tray

30
animehentai

LoRA · SD

Anime Lineart / Manga-like (线稿/線画/マンガ風/漫画风) Style

200
animeart styledigital illustrationfemalegirlsillustrationlineartmangamonochromestylewoman

LoRA · SD

JapaneseDollLikeness (v1.5)

70
JapaneseDollLikeness (v1.5) generates images with a Japanese doll-like aesthetic, featuring a variety of faces. It's recommended to use specific settings like the DPM++ SDE Karras sampler and to experiment with different LORA weights. The model is openly accessible under the CreativeML OpenRAIL-M license, which includes usage restrictions against harmful or illegal outputs.asianjapanesephotorealisticstylewomen

LoRA · Wan Video 2.2 I2V-A14B

Meatspin - Wan 2.2 I2V

1470
The Meatspin Wan 2.2 I2V model is a LoRA (low-rank adaptation) designed for WAN 2.2 I2V to generate videos of characters with their penis spinning in a circular motion, primarily tested on shemale characters. Key features include increased training steps, improved versatility for different poses, and the ability to create longer videos, though it may exhibit a slight initial delay. Use trigger word m34tsp1n.meatspinpenisspinspinningstyle

LoRA · SD

Mikuya

60
Mikuya is a LORA model trained with Kohya GUI using 40 images and 4000 total steps. The creator seeks feedback on improving the model's realism for a future version 2.0.realisticwoman

LoRA · Wan Video 2.2 T2V-A14B

Wan2.2 - Korean Women

00
Wan2.2 is a LoRA model focused on generating images of Korean women, aiming to improve image quality and address deformation issues from previous versions. It requires the Wan 2.2 Self-forcing LoRAs (High/Low) and benefits from specific sampler and scheduler settings for optimal results. The model's dataset has been updated to refine skin textures and character appearance, striving for a balance between realism and a fictional representation.character

LoRA · Wan Video 2.2 I2V-A14B

WAN 2.2 I2V - POV Paizuri / Titfuck

1K0
This model generates images of POV titfucking, trained on clips where either the man or woman initiates the movement. It is specifically tagged for "gather" when the woman wraps her breasts around the penis, and "fingersTogether" when her fingers interlace. The model may be further refined in the future.action

LoRA · Other

Condom belt

30
This LoRA model generates images of condom belts, and is compatible with other LoRAs. The number of condoms can be adjusted via prompt strength, and it triggers when a strap or belt is detected.charactercondomcondom belt

LoRA · Other

Standing with spread legs pose

20
This LoRA is created to generate images of standing girls with spread legs. The best weights is 1.3-1.4. Trigger words: spread legs standinganimeposessexyspread legsstanding

LoRA · SD

concept Excessive pubic hair

20
Strengthen the distribution and density of pubic hair. Change the weight to control the level. Adding "armpit hair" to the negative prompt to avoid producing thick armpit hair.animeconceptfemalehairypubic hair

LoRA · Wan Video 14B t2v

Trans Woman/Futanari/Male Masturbation (Wan 2.2 and 2.1)

30
This model generates images of trans women, futanari, and male masturbation. It uses a trigger word "str0king_it" for activation and requires both HIGH and LOW models with specific settings. Prompt suggestions are provided for generating desired actions and visuals.concept

LoRA · SD

沁彩 Colorwater

70
沁彩 Colorwater is a LoRA model that applies a watercolor style to images. Different versions offer varying transparency and model compatibility, with v4 being the most recommended. It's still under development and users are encouraged to experiment with weights and CFG settings to achieve desired effects.animeart styleartstylechinesepaintingsstylestyleswatercolor

LoRA · Other

アニメ塗り Anime Style LoRA with Better Flat Color

30
This LoRA produces anime-style coloring without requiring complex prompts. It avoids logos and text artifacts due to clean training data. A Better Flat Color Simplifier is included for detail reduction.animeanime girlart styleflat colorlow-file sizestyle
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.