ToriiGate-v0.3
A multimodal LLM finetuned with anime images to be used for captioning of anime images. No limits, no censorship, advanced knowledge, SOTA with something complex and NSFW yet WIP
Might be very useful for dataset preparation, especially for FLUX/SD3/... models.
New version comes with improved knowledges, accuracy, now can caption comic/manga and handles multiple characters better.
Basically it became more stable and in bad conditions tends to provide shorter and more general description instead of hallucinating like other models. However mistakes are still happening (like with every vlm or generative ai in general).
Code examples and tricks
You will need any torch >=2.1.2 and dev transformers pip install git+https://github.com/huggingface/transformers
to run it.
Example script for single image, for batch processing.
Quantization
To reduce VRAM requirements you can load model in 4 or 8 bit.
Example:
Prompt format
For best results stick to one of trained user prompt:
- Describe the picture in structured json-like format.
- Give a long and detailed description of the picture.
- Describe the picture briefly.
First one provides most informative results and should be used when combined with further NLP. Second is just a typical large sloppy rubbish output for most vlms. Third gives only meaningful and convenient description and can be used directly for training, but combination of first with extra LLM processing will be better.
Do not forget to add system instruction like in provided templates and stick to format!
Grounding with booru tags
To achieve best accuracy and improve understanding of complex scenes you should use booru tags
even from wd3tagger ensemble is fine
It should be standard booru-like appearance, but natural text captions from clip or ever json/xml/yaml captions from api will work.
Examples
Examples below are shown as is.
They are picked mainly based on their unusualness, complexity or interesting appearance, not because of good captioning results. So some mistake or missing object are present, so you can estimate overall performance.
Zero shot captions (without providing extra tags or info about image) are shown in blue fields
Grounded captions (with booru tags added in prompt) are shown in green fields
SFW
{
"comic_format": "Comic of 2 frames",
"1st_frame": "A fox girl with white hair, fox ears, and green eyes is enthusiastically talking about glasses. She is wearing a white sailor-style outfit with a blue neckerchief and has a playful expression. Her hands are making a fox shadow puppet gesture. The background is yellow with text discussing the versatility of glasses and their appeal.",
"2nd_frame": "The fox girl is now facing a cashier with orange hair and purple eyes. The cashier, wearing a uniform with a blue and white design, looks slightly confused. The setting appears to be a store, with a cash register visible.",
"character_1": "The fox girl has white hair, fox ears, and green eyes. She is wearing a white sailor-style outfit with a blue neckerchief. Her expression is playful and energetic.",
"character_2": "The cashier has orange hair and purple eyes. She is wearing a uniform with a blue and white design. Her expression is slightly confused.",
"texts": "The first frame contains a large text bubble with the fox girl's enthusiastic speech about glasses. The second frame has a smaller text bubble where the cashier responds, 'Ma'am, this is a KFP.'",
"meaning": "The comic humorously portrays the fox girl's love for glasses, leading to a misunderstanding at a store where she is mistaken for a KFC employee."
}
{
"character": "The image features a girl with purple hair styled in low twintails, adorned with a purple ribbon. She has purple eyes and a playful expression, with a slight blush on her cheeks. She is performing a handstand, looking directly at the viewer with an open mouth and a smile. Her outfit includes a purple dress with a strapless design, a hooded jacket that is unzipped, revealing her cleavage. She is wearing purple thighhighs with ribbed legwear, and white panties are visible due to her upside-down position.",
"background": "The background is a simple, light-colored design with abstract shapes, adding a playful and dynamic feel to the image.",
"texts": "There is a signature at the bottom left of the image, dated 2017.2.18.",
"atmosphere": "The atmosphere is lively and playful, with a sense of energy and whimsy due to the character's dynamic pose and expression."
}
{
"character": "The image features a girl viewed from the side, with a focus on her lower body. She is wearing a high-waisted black skirt and a blue shirt. Her legs are mechanical, with visible robot joints and prosthetic elements. She is wearing a single black thigh-high stocking on one leg, and her footwear is futuristic and high-heeled.",
"background": "The background is a simple orange color with a green border framing the image.",
"texts": null,
"atmosphere": "The atmosphere is futuristic and slightly surreal, with a focus on the contrast between the organic and mechanical elements of the character."
}
{
"character": "The image features a girl with short white hair and green eyes, performing a handstand on a large ball. She is wearing a white leotard with red accents, black thigh-highs, and red footwear. Her outfit is complemented by black fingerless gloves. Her expression is cheerful, with a slight blush on her cheeks, and she is looking directly at the viewer. Her pose is dynamic, with one leg extended upward and the other bent, showcasing her flexibility.",
"background": "The background is simple and white, emphasizing the character and her pose.",
"texts": "There is a small text in Japanese on the right side of the image.",
"atmosphere": "The atmosphere is playful and energetic, with a focus on the character's acrobatic skill and cheerful demeanor."
}
{
"character": "The image features a girl with long, flowing purple hair and striking blue eyes. She is lying on her back, wearing a white dress that is slightly disheveled, revealing her bare shoulders and a hint of her navel. Her expression is calm, with a slight blush on her cheeks. She holds a small, glowing orb in her hand, and a pearl necklace rests around her neck.",
"background": "The setting is a cozy library with bookshelves filled with books. The girl is lying in an open chest, surrounded by scattered books and a few other objects, creating a sense of curiosity and adventure.",
"texts": null,
"atmosphere": "The atmosphere is serene and slightly mysterious, with a sense of discovery and tranquility. The soft lighting and the girl's relaxed pose contribute to a peaceful, introspective mood."
}
{
"character": "None",
"background": "The image features a dynamic and vibrant scene of multiple fighter jets in flight against a clear blue sky. The jets are depicted in various angles and positions, creating a sense of motion and action. The jets include the F-22 Raptor, F-15 Eagle, F-16 Fighting Falcon, and Su-30, among others. The aircraft are detailed with realistic features, including their canopies, wings, and tail fins. The sky is a gradient of blue, transitioning from a deeper blue at the top to a lighter blue near the bottom.",
"texts": "Congratulations on ACE COMBAT 7 release!!",
"atmosphere": "The overall atmosphere of the image is energetic and celebratory, capturing the excitement and thrill of aerial combat. The presence of multiple jets in flight suggests a large-scale military operation or a high-stakes battle scenario. The image conveys a sense of power, speed, and precision, typical of an ace combat setting."
}
{
"character": "The image features a girl with black hair styled in twintails, adorned with a maid headdress. She has brown eyes and a gentle expression. Her outfit is a classic maid uniform with a black dress and a white apron, complemented by a choker around her neck. The character has visible robot joints, indicating she is an android. She is sitting on a mechanical platform with caterpillar tracks, with her hands together in front of her.",
"background": "The background is simple and white, emphasizing the character and the mechanical platform.",
"texts": "The platform has the text 'MU-01' displayed prominently.",
"atmosphere": "The atmosphere is a blend of futuristic and classic, with a focus on the android maid's serene and composed demeanor."
}
{
"comic_format": "Comic of 4 frames",
"1st_frame": "A chibi character with long, gradient hair and pointy ears is holding a blanket with a cute design. She has a happy expression with closed eyes and a small fang showing. The blanket has the word 'blanket' written on it.",
"2nd_frame": "The character is now lying on the blanket, stretching her arms out in front of her. She looks content and relaxed.",
"3rd_frame": "The character is rolling up into a ball, with a motion line indicating movement. The word 'ROLL' is prominently displayed in bold letters.",
"4th_frame": "The character is completely rolled up in the blanket, with a satisfied expression. The text 'INA INA ROLL' is written above her, and a thought bubble says 'It's warm'.",
"character": "The character has long, gradient hair that transitions from purple to orange. She has pointy ears and a small fang, giving her a cute and playful appearance. She is wearing a crown-like accessory.",
"texts": "The text includes 'blanket', 'ROLL', 'INA INA ROLL', and a thought bubble saying 'It's warm'.",
"meaning": "The comic depicts a playful and cozy scene where the character enjoys wrapping herself in a blanket, rolling up, and feeling warm and content."
}
{
"character": "The image features a dark-skinned female character with long black hair and purple eyes. She has animal ears and a black tail, adding a unique, animalistic charm to her appearance. Her hair is adorned with a blue hairband, and she has blunt bangs framing her face. She is wearing a black jacket with long sleeves, which is open to reveal a white top that struggles to contain her large breasts, causing a button to pop off. Her midriff is exposed, showcasing her navel and a plump belly. She is also wearing a blue skirt that highlights her thick thighs. The character is holding a pink doughnut with heart decorations in her right hand, and there is a blush on her cheeks, suggesting a playful or embarrassed expression. Her mouth is open, revealing a small fang, adding to her animalistic features.",
"background": "The background shows an indoor setting with a computer monitor, keyboard, and mouse on a desk. There is a box of doughnuts on the desk, with some doughnuts scattered around, indicating a casual or relaxed environment. The lighting is soft, casting shadows on the wall behind the character.",
"texts": "None",
"atmosphere": "The overall atmosphere of the image is playful and slightly provocative, with a focus on the character's unique features and casual, relaxed setting. The combination of her animalistic traits and the casual environment creates a whimsical and engaging scene."
}
{
"character_1": "The first girl on the left has long, multicolored hair with a mix of grey and red hues. She has wolf ears and a fluffy tail, which is visible behind her. She is wearing a black kimono with a floral pattern and a red obi. She has yellow eyes and a playful expression, holding a sparkler in her hand.",
"character_2": "The second girl in the middle has long, blonde hair and fox ears. She is wearing a white kimono with a floral pattern and a pink obi. She has red eyes and a gentle smile, holding a sparkler in her hand.",
"character_3": "The third girl on the right has short, brown hair with a braid and rabbit ears. She is wearing a white kimono with a floral pattern and a yellow obi. She has purple eyes and a cheerful expression, holding a sparkler in her hand.",
"character_4": "The fourth girl has short, pink hair with a braid and rabbit ears. She is wearing a white kimono with a floral pattern and a red obi. She has green eyes and a happy expression, holding a sparkler in her hand.",
"background": "The background shows a traditional Japanese setting with wooden floors and sliding doors. There are some plants and a table with drinks and snacks. The scene is lit by the warm glow of the sparklers.",
"texts": "None",
"atmosphere": "The overall atmosphere is festive and joyful, with the girls enjoying the sparklers and each other's company in a traditional Japanese setting. The warm lighting and traditional attire add to the celebratory mood."
}
{
"comic_format": "Comic of 2 frames",
"1st_frame": "A girl with white hair and yellow eyes, wearing a red hoodie with the hood down, is holding a mug. She has animal ears and whiskers, resembling a rabbit. Her expression is unamused. The hoodie has text on it: 'Sing', 'Taken', and 'Naaa'.",
"2nd_frame": "The girl is looking at a computer monitor. The screen shows a Twitter post with a meme format, featuring a character named Nanachi. The post includes various emojis and text about Nanachi's 'oversight'.",
"character": "The character is a girl with white hair, yellow eyes, and rabbit-like features. She wears a red hoodie over a blue shirt. Her expression is serious and unamused.",
"texts": "The hoodie has text: 'Sing', 'Taken', and 'Naaa'. The Twitter post includes a meme about Nanachi, with various emojis and text.",
"meaning": "The comic humorously depicts the girl's reaction to a meme about Nanachi, highlighting her unamused demeanor."
}
{
"comic_format": "Comic of 2 frames",
"1st_frame": "A character with long red hair and green eyes, wearing a white dress with a blue trim and a golden headpiece, is smiling warmly. She has a blush on her cheeks and is looking directly at the viewer. The background is a blurred, greenish scene with some flowers. A speech bubble next to her reads, 'Keep your spirits up! Try to smile more everyday, okay? Just like I'm doing now.'",
"2nd_frame": "A character with white hair and red eyes, wearing a helmet with horns, is shown in a close-up. The background is black, emphasizing the character's intense expression. The character is smiling with a sinister grin, and the text 'Like this?' is written in a speech bubble. The word 'SMILE...' is written below, indicating a dark or unsettling tone.",
"character_1": "The first character has long red hair, green eyes, and a cheerful expression. She wears a white dress with a blue trim and a golden headpiece. Her demeanor is encouraging and friendly.",
"character_2": "The second character has white hair, red eyes, and a sinister smile. They wear a helmet with horns, and their expression is intense and unsettling.",
"texts": "1st frame: 'Keep your spirits up! Try to smile more everyday, okay? Just like I'm doing now.' 2nd frame: 'Like this?' 'SMILE...'",
"meaning": "The comic contrasts a cheerful, encouraging message with a dark, unsettling interpretation of that message. The first frame suggests a positive and supportive tone, while the second frame subverts it with a sinister and unnerving twist."
}
{
"comic_format": "Comic of 4 frames",
"1st_frame": "A pink-haired girl with fox ears and a maid headdress is licking her paw. She has a content expression with closed eyes and a slight blush. The background is filled with heart shapes, indicating affection or happiness.",
"2nd_frame": "The same girl is now licking her other paw, still with a blissful expression. The background is a simple white, emphasizing her actions.",
"3rd_frame": "The girl looks surprised and slightly annoyed, with a speech bubble saying, 'WHAT THE HELL ARE YOU DOING?!' Her expression is one of confusion and mild disgust.",
"4th_frame": "The girl is now sweating and looking flustered, with a speech bubble saying, 'I'M BATHING.' Her expression is a mix of embarrassment and defensiveness.",
"character_1": "The character is a fox girl with pink hair, fox ears, and a tail. She wears a maid outfit with a headdress and apron. Her expressions range from content to surprised to flustered.",
"texts": "Speech bubbles include 'WHAT THE HELL ARE YOU DOING?!' and 'I'M BATHING.'",
"meaning": "The comic humorously depicts the fox girl's innocent actions of licking her paws, which are misinterpreted by another character. The final frame adds a layer of embarrassment and defensiveness."
}
{
"comic_format": "Comic of 2 frames",
"1st_frame": "A girl with black and red hair, wearing a maid outfit with a headdress, is speaking. She has red eyes and a mole under her eye. She is holding her hand near her face, looking slightly annoyed. The background is a blurred indoor setting. She says, 'Even if the prey is being snatched up by other predators, sharks won't let their prey go.'",
"2nd_frame": "The girl is now shown from the back, with an anger vein mark indicating her frustration. Next to her, a boy with blue hair and a black shirt under a blue jacket is holding a disposable cup with a straw. He looks nervous, with sweatdrops and notice lines around him. He says, 'O-oh... Your eyes are somehow focused on me, will I be eaten?' The girl responds with 'Huh? Of course not.'",
"character_1": "The girl has black and red hair, red eyes, and a mole under her eye. She wears a maid outfit with a headdress and appears slightly annoyed.",
"character_2": "The boy has blue hair and is wearing a black shirt under a blue jacket. He holds a disposable cup and looks nervous.",
"texts": "1st frame: 'Even if the prey is being snatched up by other predators, sharks won't let their prey go.' 2nd frame: 'O-oh... Your eyes are somehow focused on me, will I be eaten?' 'Huh? Of course not.'",
"meaning": "The comic humorously depicts a conversation between a shark girl and a boy, with the girl making a shark-related comment and the boy nervously interpreting it as a threat."
}
{
"comic_format": "Comic of 4 frames",
"1st_frame": "Saki, wearing a white helmet with rabbit ears, is excitedly showing off a tall tower made of matches. She is smiling and has a confident expression. Miyu, with a helmet and a slightly worried look, comments on the tower's impressiveness. Moe, with a neutral expression, observes the scene.",
"2nd_frame": "Moe, with a blank face and wide eyes, stands at a crossroads, contemplating the fate of the tower. Saki and Miyu look on with curiosity, as indicated by the question marks above their heads.",
"3rd_frame": "Moe is overwhelmed with the decision of whether to strike down the tower or let it burn. Her face is flushed with excitement and anxiety, and she is shouting 'PANT!' repeatedly. The background is filled with dramatic sound effects, emphasizing her internal struggle.",
"4th_frame": "Moe, with a determined expression, decides to end the tower with a dramatic 'BOOM!' Saki and Miyu are shocked, with Saki shouting 'Why?!' in disbelief.",
"character_1": "Saki, with short hair, a white helmet with rabbit ears, and a confident demeanor. She is enthusiastic and proud of her creation.",
"character_2": "Miyu, with long hair, a helmet, and a slightly worried expression. She is curious about the tower's fate.",
"character_3": "Moe, with a blank face and wide eyes, who is overwhelmed by the decision of what to do with the tower.",
"texts": "Speech bubbles include Saki's excitement, Miyu's curiosity, Moe's internal struggle, and the dramatic 'BOOM!' and 'Why?!' in the final frame.",
"meaning": "The comic humorously depicts Moe's internal struggle and eventual decision to destroy the tower, highlighting her indecisiveness and the dramatic impact of her choice."
}
{
"character": "The image features a fox girl with long, flowing white hair and striking red eyes. She has fox ears adorned with black bows and a floral ornament, adding to her charming appearance. Her expression is soft, with a slight blush on her cheeks and an open mouth, giving her a gentle and inviting look. She is wearing a sleeveless white kimono with a floral pattern, complemented by a dark blue obi sash. Her outfit is accessorized with black elbow-length gloves and a single white thigh-high stocking. Notably, she has multiple fluffy fox tails, adding to her mystical allure. A pair of pink panties is playfully placed on her head, adding a whimsical touch to her ensemble.",
"background": "The background is a clear blue sky with a few scattered clouds, creating a serene and peaceful atmosphere.",
"texts": null,
"atmosphere": "The overall atmosphere of the image is whimsical and enchanting, with a blend of traditional Japanese elements and a playful, fantasy-inspired twist."
}
NSFW
{
"comic_format": "Comic of 2 frames",
"1st_frame": "A girl with red hair styled in twintails is standing with a ball and chain restraint attached to her ankle. She has a serious expression, with clenched hands and a sweatdrop on her face. The text above her reads, 'Others think horny has chained your life.'",
"2nd_frame": "The same girl is now smiling broadly with a sparkle effect around her. Her gigantic testicles, labeled 'HORNY,' are prominently displayed. The text above her reads, 'But actually horny is part of your life.'",
"character": "The character is a girl with long red hair in twintails, wearing a red and brown outfit with an ascot. She has a confident and playful demeanor.",
"texts": "The text in the first frame reads, 'Others think horny has chained your life.' The text in the second frame reads, 'But actually horny is part of your life.'",
"meaning": "The comic humorously contrasts the perception of being restrained by one's desires with the reality of embracing them."
}"]
{
"character": "The image features a close-up of a female character's lower body. She is wearing black shorts that are unbuttoned and partially pulled down, revealing her navel and a hint of her pubic area, which is censored with a mosaic effect. Her skin is smooth and fair, and her stomach is toned, showing a defined navel. The character's pose is relaxed, with her shorts hanging loosely around her hips.",
"background": "The background is a dark, gradient blue, creating a contrast with the character's skin. Several ghost-like figures with simple, expressive faces are scattered around the image, adding a playful and whimsical element.",
"texts": "None",
"atmosphere": "The atmosphere is playful and slightly mischievous, with a touch of whimsy due to the ghostly figures. The dark background and the character's relaxed pose contribute to a casual and intimate mood."
}
{
"character": "The image features a female character with short grey hair, tied with a black and orange headband. She is completely nude, performing a handstand with her legs spread wide. Her body is glistening with sweat, highlighting her curves and the details of her anatomy. Her breasts are medium-sized with visible nipples, and her navel is clearly defined. Her anus and pussy are also visible, with the pussy being censored by a mosaic effect. Her expression is not visible due to her upside-down position.",
"background": "The background is blurry, with a mix of warm colors, suggesting an outdoor setting, possibly a beach or a track field.",
"texts": "None",
"atmosphere": "The atmosphere of the image is provocative and explicit, focusing on the physicality and flexibility of the character. The use of sweat and the detailed depiction of her body contribute to a sensual and intense mood."
}
{
"character": "The image features a large, erect penis with a detailed and realistic depiction of the shaft, head, and testicles. The penis is adorned with a police uniform, complete with a blue shirt, black necktie, and a police hat. The shirt has long sleeves and is buttoned up, with two pockets on the sleeves. The necktie is neatly tied and hangs down the front of the shaft. The hat is a classic police cap with a badge on the front. The skin tone is a light pink, and the veins on the shaft are visible, adding to the realism of the depiction.",
"background": "The background is simple and white, with no additional elements or distractions.",
"texts": "None",
"atmosphere": "The overall atmosphere of the image is explicit and provocative, focusing on the sexual nature of the subject with a humorous twist due to the police uniform attire."
}
{
"character_1": "The first character has pink hair and is positioned on top of the second character. She is engaged in cunnilingus, with her head between the other girl's legs. Her eyes are closed, and she has a blush on her face, indicating arousal. Her arms are up, and she appears to be sweating, adding to the intensity of the moment.",
"character_2": "The second character has black hair and is lying on her back on a bed. Her legs are spread, and she is receiving oral pleasure from the first character. Her eyes are also closed, and she has a blush on her face. Her medium-sized breasts are visible, with nipples erect, and there is a hint of pubic hair visible. She appears to be sweating as well.",
"background": "The background is a bed with dark sheets, creating a private and intimate setting. The lighting is dim, with a warm glow that highlights the characters' bodies.",
"texts": "None",
"atmosphere": "The atmosphere is intimate and erotic, with a focus on the sexual interaction between the two characters. The sweat and blushes add to the intensity and passion of the moment."
}
{
"character": "The image features a blonde-haired girl with green eyes, standing in a doorway. She has a flushed expression, with a blush on her cheeks and a slightly open mouth, revealing her teeth and a bit of saliva. Her hair is styled in a side ponytail, and she is wearing a white shirt that is slipping off her shoulders, exposing her collarbone and cleavage. She is also wearing black socks. The girl appears to be in a state of undress, with her shirt pulled down, revealing her bare chest and navel. There is a visible trail of cum on her body, particularly on her chest and stomach.",
"background": "The background shows a doorway with a blurred interior, suggesting an indoor setting. The door frame and part of the door are visible, with a hint of a room behind her.",
"texts": "There is a speech bubble with the text: 'Oh! I'm sorry soul, were we too loud? I'll try to keep it down for you. Don't want to make you too excited with all the noise we're making in your room~'",
"atmosphere": "The atmosphere of the image is intimate and suggestive, with a focus on the girl's expression and state of undress. The setting and dialogue contribute to a sense of privacy and closeness, with a playful and teasing tone."
}
{
"character": "The image features a girl with short hair and purple eyes, positioned on all fours. She has animal ears and a cat-like tail, which is inserted into her anus. Her expression is playful, looking back at the viewer. She is nude, with her bare feet and soles visible, and her body is positioned in a way that emphasizes her buttocks and genitalia.",
"background": "The background is simple and white, with a shadow cast by the character.",
"texts": "None",
"atmosphere": "The atmosphere is provocative and playful, with a focus on the character's pose and expression."
}
{
"character_1": "A chibi-style girl with long brown hair, blushing, and wearing a surprised expression. She is completely nude, positioned on all fours, and engaged in a sexual act from behind with a faceless male character.",
"character_2": "A faceless male character with blonde hair, wearing sunglasses and a black shirt. He is also engaged in a sexual act with the girl from behind.",
"character_3": "A chibi-style fat man with glasses, wearing only white briefs. He has a surprised expression and is standing in the middle of the scene.",
"character_4": "A chibi-style boy with black hair, holding a bouquet of flowers. He is standing to the right of the scene, looking towards the other characters.",
"background": "The scene is set in a two-story structure with ladders connecting the floors. The background is minimalistic with a gradient sky.",
"texts": "There are Japanese characters in red around the characters, indicating sounds or expressions. At the bottom of the image, the text 'YOU LOSE...' is written in bold white letters.",
"atmosphere": "The atmosphere is humorous and chaotic, with a parody-like tone. The exaggerated expressions and chibi art style contribute to a comedic and absurd scene."
}
{
"character": "The image features a character with pink hair and animal-like features, including squirrel ears and a fluffy tail. The character is completely nude, with small breasts and an erect penis. They are kneeling and engaging in self-pleasure, using a dildo for anal insertion. The character's expression is one of pleasure. Sweat is visible on the body, and there is a significant amount of cum on the penis and surrounding area. Their pink eyes are focused, and they wear glasses with an 'X' hair ornament.",
"background": "The background is simple and white, emphasizing the character.",
"texts": "The image contains the artist's name, 'SERAZIEL,' in the top left corner.",
"atmosphere": "The atmosphere is explicit and focused on the character's sexual pleasure, with a sense of intimacy and self-indulgence."
}
{
"character": "The image features a girl with blue hair styled in twintails, adorned with red hair bobbles. She is wearing a hat and blue boots. Her eyes are closed, and she has a blush on her cheeks. The character is nude, with small breasts and visible nipples. She is squatting with her legs spread, and her hand is positioned between her legs, engaging in masturbation. There is a key hanging around her neck, and she appears to be peeing, with water visible around her.",
"background": "The background shows a waterfall, with water cascading down into a pool. The setting appears to be outdoors, with a natural, rocky environment.",
"texts": null,
"atmosphere": "The atmosphere is intimate and explicit, with a focus on the character's private moment in a serene, natural setting."
}
{
"character_1": "The central character is a girl with short, purple hair and purple eyes. She has animal ears and is wearing a revealing Halloween costume with fur trim. Her outfit includes elbow gloves and thigh-highs, both trimmed with fur. She has large breasts, which are exposed and covered in cum. Her nipples are visible and erect. She is straddling one of the male characters in a cowgirl position, with her back arched. Her mouth is open, and she is performing fellatio on another male character, who is grabbing her head. Cum is visible in her mouth and on her body. She is also giving a handjob to another male character, whose penis is in her hand. There is cum on her breasts and body, and she appears to be in a state of arousal and exhaustion.",
"character_2": "There are three male characters in the image. One is standing behind the girl, grabbing her head while his penis is in her mouth. Another male character is in front of her, with his penis in her hand. The third male character is underneath her, with his penis inside her vagina. All three male characters have visible penises, and cum is present on their bodies and the girl's body.",
"background": "The background is dark and nondescript, focusing all attention on the characters and their actions.",
"texts": "None",
"atmosphere": "The atmosphere is highly sexual and explicit, with a focus on group sex and the girl's involvement in multiple sexual acts. The image conveys a sense of intense arousal and physical exertion."
}
{
"character_1": "The main character in the image is a blonde-haired girl with blue eyes, wearing a blue dress with a white apron and white thigh-highs. She has a hair ribbon and a hairband, and her dress is lifted, exposing her breasts and genitalia. She is bound with green vines, which are also restraining her legs. Her expression shows distress and struggle, with her mouth open and cheeks blushing. Her breasts are being sucked by a purple flower with a face, and her vagina is being penetrated by a pink flower with a face. There are also green tentacles wrapped around her body, one of which is penetrating her anus.",
"character_2": "There are two flower characters with faces. One is purple with a face and is sucking on the girl's breast, while the other is pink with a face and is penetrating her vagina. Both flowers have green vines attached to them.",
"background": "The background is a nature scene with various flowers and mushrooms. The setting appears to be a forest or a garden with a green, lush atmosphere.",
"texts": "There is a watermark at the bottom left corner of the image with the text 'Tlarakanovich patreon.com/tarakakanovich'.",
"atmosphere": "The overall atmosphere of the image is dark and explicit, with a strong focus on the girl's distress and the sexual nature of the scene. The use of flowers and vines as sexual objects adds a surreal and fantastical element to the image."
}
{
"character_1": "The first girl is lying on her back on a bed. She has long black hair tied with a ribbon, and her blue eyes are closed. She is blushing and has a slightly open mouth, indicating a state of arousal. She is bottomless, exposing her small breasts and navel. Her legs are spread apart, and her feet are bare, with her toes slightly curled. Her body is glistening with sweat, adding to the overall erotic atmosphere.",
"character_2": "The second girl is positioned between the first girl's legs, performing cunnilingus. She has brown hair styled in a single bun with a ribbon, and her brown eyes are closed. She is wearing a yellow top with long sleeves and a dark skirt. Her mouth is open, and her tongue is out as she licks the first girl's pussy, which is visibly wet with pussy juice. Her hands are gripping the first girl's thighs, spreading them further apart.",
"background": "The background shows a bedroom setting with a bed covered in a brown sheet. There are pillows on the bed, and a window is visible in the background, letting in some light. A camera on a tripod is positioned to the side, suggesting that the scene is being recorded.",
"texts": "None",
"atmosphere": "The overall atmosphere of the image is highly erotic and intimate, with a strong focus on the sexual interaction between the two girls. The presence of the camera adds an element of voyeurism, enhancing the intensity of the scene."
}
{
"character_1": "The image features a blonde-haired fox girl with large fox ears and a fluffy fox tail. She has long hair and brown eyes, and her expression is playful with a smile and her tongue out. She is wearing a black micro bikini that barely covers her large breasts, with one nipple slipping out. Her body is glistening with sweat, and she is lying on her back on a bed with her legs spread. Her fox tail is prominently visible, adding to her animalistic features.",
"character_2": "A male character is partially visible, with a focus on his erection under his clothes. He is interacting with the fox girl, using anal beads for object insertion.",
"background": "The background consists of a bed with pillows, creating a soft and intimate setting.",
"texts": null,
"atmosphere": "The atmosphere is sensual and intimate, with a focus on the erotic interaction between the characters. The setting and expressions contribute to a playful and provocative mood."
}
{
"character_1": "The image features a girl with long white hair and red eyes, who appears to be albino. She is completely nude, with small breasts and visible nipples. Her expression shows a mix of emotions, including blush and tears, indicating a complex emotional state. She is engaged in a sexual act, performing fellatio on one of the boys while being penetrated vaginally by another.",
"character_2": "One of the boys is depicted with a muscular build, holding the girl's hair while engaging in the sexual act. His face is not visible, focusing the attention on the interaction.",
"character_3": "The second boy is also muscular and is involved in the sexual act, with his penis visible and engaged in vaginal penetration. His face is not shown, maintaining the focus on the physical interaction.",
"background": "The background is a deep red, possibly a bed or a similar setting, adding to the intimate and intense atmosphere of the scene.",
"texts": null,
"atmosphere": "The atmosphere is intense and explicit, focusing on the physical and emotional interactions between the characters. The use of color and lighting enhances the intimate and provocative nature of the scene."
}
{
"character_1": "The first character is a nude female with long brown hair and wolf ears. She has medium-sized breasts and a visible navel. Her legs are spread wide apart, and she is standing on one leg while the other leg is lifted. She has a wolf tail that is positioned behind her. Her expression is one of pleasure, with parted lips and an open mouth. She is engaged in sexual activity with the second character, with a visible penis penetrating her vagina. There is cum overflowing from her vagina, indicating recent ejaculation.",
"character_2": "The second character is a nude female with long brown hair and wolf ears. She has small breasts and a visible navel. She is standing behind the first character, engaging in sexual activity with her. Her penis is knotted and penetrating the first character's vagina. There is cum visible on her penis, indicating recent ejaculation. She is wearing a small pouch around her neck.",
"background": "The background features a tree with a thick trunk on the left side of the image. The scene is set outdoors, with a grassy field and distant mountains visible. The sky is clear with a few clouds.",
"texts": "None",
"atmosphere": "The atmosphere of the image is explicit and sexual, with a focus on the intense physical interaction between the two characters. The setting is natural and outdoor, contrasting with the intense and intimate act taking place."
}
{
"comic_format": "Comic of 2 frames",
"1st_frame": "Three female characters are depicted in a tavern setting. The Amazon warrior, with her muscular build and feather hair ornament, is holding a mug and looking surprised. The Elf, with her pointy ears and twin braids, is blushing and sweating, holding a staff. The Sorceress, wearing a witch hat and revealing outfit, is smiling mischievously. Speech bubbles show them discussing 'if we had sex?' and 'hat ha ha ha'.",
"2nd_frame": "The scene shifts to a more explicit and intimate setting. The characters are engaged in a threesome, with the Sorceress and Elf having sex with the Amazon. The Amazon is shown with a large, dark penis, ejaculating inside the Elf. The Sorceress is grabbing the Amazon's ass, and the Elf is moaning with pleasure. The atmosphere is intense and erotic.",
"character_1": "Amazon warrior: Muscular, wearing armor, with feather hair ornament, surprised and then engaged in sexual activity.",
"character_2": "Elf: Pointy ears, twin braids, wearing a revealing outfit, blushing and sweating, involved in sexual activity.",
"character_3": "Sorceress: Wearing a witch hat and revealing outfit, smiling mischievously, actively participating in the sexual encounter.",
"texts": "Speech bubbles include 'wouldn't it be funny', 'if we had sex?', 'hat ha ha ha', 'he he boo ba', 'fuuuck', 'we should like this more often...'.",
"meaning": "The comic humorously explores the idea of the characters engaging in a sexual encounter, with a mix of surprise and enjoyment. The atmosphere is playful and erotic, with a focus on the characters' interactions and expressions."
}
{
"character_1": "The image features a female character with blonde hair, identified as Princess Peach from the Mario series. She is positioned on all fours, with her back arched and her large, hanging breasts prominently displayed. Her nipples are visible, and there is a sheen of sweat on her skin, indicating exertion. Her face is not visible in the image, but her body language suggests she is in a state of struggle or intense sensation.",
"character_2": "The second character is Bowser, a large, anthropomorphic turtle-like creature. His large, brown shell is positioned over Princess Peach's back, and his huge penis is penetrating her anus. His testicles are visible, and there is a trail of pussy juice indicating arousal. The penis is depicted as large and veiny, with a heart symbol near it, emphasizing the intensity of the scene.",
"background": "The background is dark and nondescript, focusing the viewer's attention on the characters and their interaction.",
"texts": "There are no speech bubbles or text in the image, but there are small heart symbols near the penis, indicating pleasure or affection.",
"atmosphere": "The atmosphere of the image is highly sexual and intense, with a focus on the physical interaction between the characters. The sweat and pussy juice add to the raw, primal nature of the scene."
}
{
"character_1": "The image features a blonde-haired girl with animal ears, specifically wolf ears, and a collar around her neck. She has a flushed face, indicating arousal, and her eyes are half-closed with a look of pleasure. Her mouth is open, and she appears to be moaning. She is nude, with visible medium-sized breasts and nipples. Her body is covered in sweat, emphasizing the intensity of the moment. Her hand is placed on her own stomach, and there is a visible bulge in her lower abdomen, indicating internal ejaculation.",
"character_2": "The second character is partially visible, wearing a dark shirt with the text 'I \u2665 Internal Cum' on the sleeve. This character is engaging in a sexual act with the first character, as evidenced by the visible penis penetrating the girl. The penis is depicted with a condom, and there is a cross-section showing the internal cumshot.",
"background": "The background is a simple, muted color, which helps to keep the focus on the characters and their interaction.",
"texts": "The shirt has the text 'I \u2665 Internal Cum' on the sleeve. There are also sound effects in Japanese, indicating the sounds of the sexual act.",
"atmosphere": "The atmosphere of the image is highly sexual and explicit, focusing on the intense physical pleasure and the act of internal ejaculation. The expressions and body language of the characters convey a sense of intense arousal and satisfaction."
}
{
"character_1": "The first character is a muscular male with brown hair and a beard. He has blue eyes and is completely nude, displaying his well-defined abs and pectorals. His expression is intense, with a blush on his cheeks and a smirk. He is ejaculating forcefully, with cum spraying from his penis. His body is glistening with sweat, and he has arm hair and long sideburns.",
"character_2": "The second character is also a muscular male with red hair and a beard. He has red eyes and is completely nude, showcasing his large pectorals and defined muscles. He is positioned behind the first character, holding him by the thigh and shoulder. His expression is one of pleasure, with a smile and a slight blush. He is also ejaculating, with cum visible around the area of penetration.",
"background": "The background is simple, consisting of a tiled floor, suggesting an indoor setting, possibly a bathroom or shower area.",
"texts": "None",
"atmosphere": "The atmosphere is intense and erotic, with a focus on the physical and emotional connection between the two characters. The scene is explicit, emphasizing the sexual activity and the physicality of the characters."
}
{
"character": "The image features a female character with black and red two-tone hair, styled with a spiked headband and a maid headdress. She has red eyes and a mole under her eye. Her expression is a mix of a frown and a blush, with her mouth open and tongue out. She is completely nude, with large breasts and visible nipples, and there is a depiction of lactation. Her arms are raised behind her head, and she is squatting with her legs spread, revealing her genitalia. A shark tail with a scar is visible, adding to her distinct appearance.",
"background": "The background is a 3D setting with a soft, warm light, possibly indicating a sunset. There are bottles on the floor, suggesting a relaxed or casual environment.",
"texts": "There is a speech bubble with a cross-section illustration of fertilization, showing sperm cells and an ovum.",
"atmosphere": "The atmosphere is intimate and sensual, with a focus on the character's expression and pose, enhanced by the warm lighting and relaxed setting."
}
More spicy?
Disturbing R18G pictures below this line!!!
You've been warned
Basically a proof of understanding some thighs, even with zero-shot. Not good as it should be, but already usable.
{
"character_1": "The first character is a girl with short black hair and blue eyes. She has a mechanical appearance, with visible metal skin and exposed robotic joints. Her face is partially disfigured, with blood and exposed bone visible. She has a hole in her head, through which another character is emerging.",
"character_2": "The second character is a smaller girl with long white hair and blue eyes. She is partially inside the head of the first character, with her body and limbs exposed. She is wearing white panties and appears to be in a fetal position. Her face is also partially disfigured, with blood and exposed bone visible.",
"background": "The background is a solid red color, adding to the surreal and horror theme of the image.",
"texts": "None",
"atmosphere": "The overall atmosphere of the image is surreal and horror-themed, with a strong sense of disorientation and unease. The combination of blood, exposed bone, and mechanical elements creates a disturbing and unsettling scene."
}
{
"character": "The image features a girl with long pink hair styled in a braid and adorned with a hair ornament. She has blue eyes and is depicted in a provocative pose, upside-down. Her outfit includes a skirt and thigh-highs, which are torn, revealing her skin. She is wearing elbow-length gloves. Her breasts are exposed, and there is visible lactation. The girl's expression is one of surprise or pleasure, with her mouth open. Her body is being interacted with by tentacles, which are engaging in various explicit activities, including anal and clitoral stimulation.",
"background": "The setting appears to be an alley, with a rough, concrete surface visible beneath the character.",
"texts": null,
"atmosphere": "The atmosphere is intense and explicit, focusing on themes of fantasy and non-consensual interaction with a monster-like creature."
}
{
"character_1": "The first character is a blonde-haired girl with red eyes, wearing a white corset and white thigh-highs. She has a mob cap with a red ribbon and is adorned with a choker. Her expression is one of pleasure, and she is engaged in an intimate act with an orc. Her body is depicted with a noticeable bulge, indicating inflation.",
"character_2": "The second character is a blue-haired girl with red eyes, also wearing a white corset and white thigh-highs. She has a similar mob cap with a red ribbon and a choker. Her expression mirrors that of the first girl, and she is similarly engaged with an orc. Her body also shows a bulge, indicating inflation.",
"character_3": "The orc on the left is large and muscular, with a menacing appearance. It has a fanged mouth and is holding the blonde-haired girl. Its expression is aggressive, and it is involved in an intimate act with her.",
"character_4": "The orc on the right is similarly large and muscular, with a fanged mouth. It is holding the blue-haired girl and is engaged in an intimate act with her.",
"background": "The background features a canopy bed with red curtains. The setting is dimly lit, with a spotlight shining on the characters, creating a dramatic atmosphere.",
"texts": null,
"atmosphere": "The atmosphere is intense and provocative, with a focus on the intimate and explicit interactions between the characters and the orcs. The setting and lighting contribute to a dramatic and charged mood."
}
{
"character": "The image features a realistic depiction of a young woman with black hair. Her head is severed, and the severed head is shown in the air, with blood splattering from the neck. The woman's eyes are wide open, and her expression is one of shock or surprise. Her upper body is exposed, revealing her breasts, with visible nipples. The blood splatter from her neck is vivid and dynamic, adding to the graphic nature of the scene.",
"background": "The background is minimalistic and white, which contrasts sharply with the dark tones of the woman's hair and the bright pink of the blood.",
"texts": "None",
"atmosphere": "The atmosphere of the image is intense and unsettling, with a strong emphasis on the graphic and disturbing nature of the scene. The use of monochrome with selective color highlights the blood, adding a sense of shock and violence."
}
{
"character": "The image features a girl with medium-length brown hair, styled with an ahoge. She is lying on her back, with her legs raised and feet prominently displayed. Her expression is one of distress, with closed eyes and an open mouth, showing upper teeth. She is wearing a black short-sleeved shirt, and her lower body is exposed, revealing her bare skin. Her body shows signs of physical abuse, with visible red marks and bruises on her buttocks and legs. There is a cane visible, suggesting recent use, and her body is trembling, with sweat visible on her skin.",
"background": "The background is plain white, focusing attention on the character.",
"texts": null,
"atmosphere": "The atmosphere is intense and unsettling, highlighting themes of abuse and vulnerability."
}
{
"character_1": "The central character is a girl with blonde hair styled in a bob cut, adorned with a red bow. She is dressed in a black vest over a white shirt, a red necktie, and a black skirt. Her arms are outstretched and nailed to a large white cross, with blood dripping from her hands and wrists. Her face is smeared with blood, and she has a distressed expression with her mouth open, showing her teeth. Her legs are also nailed to the cross, and blood is visible on her legs and feet. She is barefoot, with her toes clenched in pain.",
"character_2": "In the background, there are several other characters, depicted in a monochrome style. One of them is holding a sign that says 'MONSTER'. Another character is holding a sign that says 'GO TO HELL'. The characters in the background appear to be a mob, surrounding the central character.",
"background": "The background is mostly white, with the large white cross as the central element. The cross is blood-stained, particularly around the central character's hands and feet. The mob in the background is depicted in a monochrome style, with their faces showing anger and hatred.",
"texts": "The signs held by the mob members in the background have the following texts: 'MONSTER' and 'GO TO HELL'.",
"atmosphere": "The overall atmosphere of the image is dark and intense, with a strong sense of violence and suffering. The central character's expression and the blood on her body convey a sense of pain and distress. The mob in the background adds to the feeling of hostility and hatred, creating a tense and dramatic scene."
}
{
"character": "The image features a girl with short black hair and a blush on her cheeks. She is depicted in an anatomical illustration style, with her body shown in a detailed, cross-sectional view. Her skin is a deep red, and her muscles and internal organs are visible, including her heart and lungs. She is nude, with her arms at her sides, and her expression is calm and slightly shy.",
"background": "The background is a simple, dark gradient, which helps to emphasize the character in the foreground.",
"texts": null,
"atmosphere": "The atmosphere is surreal and somewhat unsettling due to the anatomical depiction, creating a blend of beauty and complexity."
}
{
"character_1": "A girl with brown hair, wearing a school uniform consisting of a white shirt and a green skirt. She is on all fours, with her face close to the ground, and appears to be drooling. Her expression is one of pleasure, and she is sweating. Her skirt is lifted, revealing her bare bottom and the act of bestiality with a pig.",
"character_2": "Another girl with grey hair, also in a school uniform, is holding a phone and looking at the first girl. She has a smile on her face and is also sweating. Her expression suggests she is enjoying the scene.",
"character_3": "A pig is positioned behind the first girl, engaging in sexual activity with her. The pig's body is muscular, and it is holding a bucket in its mouth.",
"background": "The setting appears to be a barn with wooden flooring and walls. There are other pigs in the background, and the atmosphere is steamy and humid.",
"texts": "None",
"atmosphere": "The atmosphere is intense and explicit, with a focus on the sexual act between the girl and the pig. The scene is graphic and depicts a non-consensual act."
}
{
"character_1": "The image features a girl with medium-length hair, wearing a school uniform consisting of a pinafore dress and a long-sleeved shirt. She has a distinctive scar on her cheek and is holding a chainsaw above her head. Her expression is intense, with ringed eyes and parted lips, and she is looking directly at the viewer. The girl's outfit includes a bowtie and knee-high socks, and she is wearing shoes. Her pose is dynamic, suggesting action or combat.",
"character_2": "A severed head is depicted in the image, held by the girl. The head has sharp teeth and a cross scar on the cheek, adding to the intense and graphic nature of the scene.",
"background": "The background is monochrome and greyscale, with hatching texture adding depth. There are splatters and blood effects, enhancing the dramatic and violent atmosphere.",
"texts": "The signature 'OSSEN LIU' is present in the bottom left corner.",
"atmosphere": "The overall atmosphere is intense and chaotic, with a strong sense of violence and action. The monochrome palette and dynamic composition contribute to the dramatic and unsettling mood of the image."
}
{
"character_1": "The first character is a girl with long black hair and yellow eyes. She is wearing a school uniform consisting of a white collared shirt, a black vest, and a black skirt. A red ribbon is tied around her neck. Her expression is intense, with a wide smile and a scar on her cheek. She has an outstretched arm, and her hair appears to be floating.",
"character_2": "The second character is a decapitated head, seemingly floating in the air. It has a shocked or surprised expression with wide eyes and an open mouth. The head is depicted in a more abstract and blurry manner compared to the first character.",
"background": "The background is simple and grey, with a focus on the characters. There is a sense of depth created by the use of a blurred effect.",
"texts": "None",
"atmosphere": "The atmosphere is intense and somewhat unsettling, with a focus on the dramatic and violent interaction between the characters. The use of blood and the decapitation adds to the graphic and disturbing nature of the image."
}