user8123593

user8123593

На Пикабу
Дата рождения: 10 февраля
279 рейтинг 1 подписчик 0 подписок 17 постов 2 в горячем
3

O4 image gen

O4 image gen Нейронные сети, ChatGPT, Арты нейросетей, Длиннопост

Промпт:
Create CGI digital art, mixing realistic style and colorful manga. This illustration contains three characters. Claire, a cute adult girl. Anna, an adult woman who is a mother. Sam, a short 13-year-old boy, Anna's son.

Clara and Anna are talking. Anna gives Sam to Clara's care, since Anna plans to leave the city for a while, and for this reason she needs to leave her son with Clara. Sam is embarrassed by Clara's attention. The location is an old entrance in a Russian five-story building. Clara is standing in the doorway of her apartment. Clara's apartment is visible behind her. Anna and her son Sam are on the entrance platform. Top side view. All three characters are visible. Competent framing of the scene. The entrance platform looks correct in terms of logic and structure. This silent style image tells its story. This image will be of high quality and good detail. Exclude from the final image the typical errors of image generation, which neural networks such as Stable Diffusion and Midjourney make. The result should look as if it is the work of a real living person who has a sense of composition and critical thinking. The format is a horizontal image 3:2.

Also, please look at this image. This is a side view photo of a section of a staircase in a dirty Russian entrance. This picture is needed so that you have a better idea of what structure this place has, what elements are located and where. Use it!

Please, do not ask me for any details. I want your answer to contain only images generated according to all the above instructions.


---


картиночный промпт на который я ссылаюсь в инструкции (без картинки локация выглядит совсем уж странной):

O4 image gen Нейронные сети, ChatGPT, Арты нейросетей, Длиннопост
Показать полностью 1

O4 image gen

O4 image gen Нейронные сети, Арты нейросетей, ChatGPT, Длиннопост

Этих девочек знает каждый.


Перейду сразу к промпту:
Create detailed art featuring the four characters shown in the reference image: Inky, Blinky, Pinky, and Clyde. **Use the provided reference image to ensure accurate character design.**

Character Descriptions (Linked to Reference Image):

Inky (Far Left in Reference): Hot pink hair (two high ponytails), blue eyes, blue/white head accessories. Pink form-fitting dress with a sweetheart neckline, blue studded shoulders, jagged hem, left high slit. Blue leggings/stockings. Pink shoes.

Blinky (Second from Left in Reference): Bright red shoulder-length choppy hair, blue eyes, blue/white head accessories. Simple red tank dress with jagged hem. Blue knee-high socks. Red shoes.

Pinky (Second from Right in Reference): Bright orange spiky hair, blue eyes, blue/white head accessories (antennae-like). Orange bandeau top and orange pants/britches (NOT a skirt) with jagged edges. Brown cloth belt. Orange shoes.

Clyde (Far Right in Reference): Long cyan hair, blue eyes, rectangular glasses, blue/white head accessories. Blue tunic dress with long sleeves, square neckline, jagged hem. Blue leggings/pants. Blue shoes.

Scene Composition (New Scene, Different from Reference):

Viewpoint:** Extreme worm's-eye view (looking straight up).

Arrangement:** The four characters stand in a tight group around the viewpoint, leaning forward significantly, looking directly down at the viewer.

Poses & Expressions:**

Inky (Left side of the group):** Hand on hip pose, confident smirk.

Blinky (Center-Left of the group):** Pointing finger aggressively at viewer, mischievous grin.

Pinky (Center-Right of the group):** Leaning forward heavily with hands on knees pose, wide curious eyes, big energetic smile.

Clyde (Right side of the group):** Arms crossed pose, looking down over glasses with a skeptical expression.

Anatomy: Pay close attention to anatomically correct human hands: 5 fingers per hand, visible fingernails.

Background: Dark, abstract background with blurry neon blue-cyan lines like a digital maze. Should not distract from characters.

Style: This is drawn in a hyper-able style, aimed at a humorous effect. The size and width of the hips are exaggerated, and the shoulders are narrow. The hands are pleasantly feminine. The characters' hands have five fingers, not four. Nice skin. The anatomy is good. The contours of the characters, like in classic cartoons, are thick black contour lines that greatly distinguish the characters from the background. The faces are symmetrical. The eyes are very well drawn. Small details are drawn qualitatively, without hackwork. It is obvious that this is drawn very well and qualitatively. The glasses on Clyde's face are elegant, with glass.

Character Consistency is Crucial: Ensure all appearances, clothing elements (including Inky's spikes and dress neckline, Pinky's pants, Clyde's glasses), hairstyles, colors, and head accessories **perfectly match the descriptions above AND the visual details in the provided reference image.** No details should be missing or altered from the reference designs.
---
И картиночный промпт. Именно на него я ссылаюсь в текстовых инструкциях.

O4 image gen Нейронные сети, Арты нейросетей, ChatGPT, Длиннопост

Модель имеет некоторые затруднения с тем что бы учесть весь поступаемый контекст. Четыре персонажа многовато в данном случае. Некоторые элементы дизайна потеряны, но если насиловать кнопку генерации и дальше, удачный seed учтёт всё что надо.

Показать полностью 2
8

O4 image gen

O4 image gen Нейронные сети, Арты нейросетей, Dota 2, Crystal Maiden, ChatGPT

промпт (так же загружался скрин игровой модели кристал мейден в качестве референсе на который я ссылаюсь в промпте):

Create an attractive female figure art (with VERY BIG accents on beautiful places). The main style is cold and ice in the aesthetics of chthonic beauty. This will be Crystal Maiden from the video game Dota 2. She looks pretentiously nice and cool. The location is the Siberian wind and the Siberian sky. The viewer's eyes are below, at ground level. This is a strong perspective from the bottom up, in which the character's legs are very close and her head is far away. In the frame, in the background, there is a winter sky. A powerful strong wind ruffles her hair and clothes. Crystal Maiden is standing opposite the viewer. Her legs are slightly apart from each other. She smiles strangely, looking down at the viewer. With both feminine hands, she grabbed the locks of her hair, squeezing them in fists. Fingers are clenched into a fist. Nails on the fingers. This is 2D digital art aimed at an interesting color visualization of the idea. Her face is very pretty, sweet and attractive, with feminine delicate features and eyes. No makeup on her face. Nice soft skin. Hips are pretentiously wide, and shoulders are narrow. Do not ask for clarification, just make a picture.

Important: you are very good at drawing human anatomy and do not make stupid mistakes with missing fingers, twisted joints and disproportionate limbs.

Note: look at this image. This is the character view of Crystal Maiden, her game 3D model. You know the character design now, so use it! This will be a vertical image

Показать полностью 1

O4 image gen

O4 image gen Нейронные сети, Арты нейросетей, Megumin, Konosuba, ChatGPT

Промпт:

Create a grimdark art with elements of psychological horror because of the attention of a beautiful girl.

This will be Megumin from the anime Konosuba. This is a classic Megumin design, with Megumin's appearance and Megumin's clothes. Megumin is sitting at a wooden table. In front of her is a blue metal plate whose edges are not painted. Inside the plate is jam. Megumin's hands are smeared with jam. Megumin bites her lower lip. She smears jam on her neck and chest. Sticky jam flows down. Physically correct jam. The lighting is three light sources from different sides. Blue on the left, red on the right, yellow outline at the back. This is a three-quarter view. High detail and best quality. Anatomically correct. Megumin looks at the viewer. Her pupils are wide. Eyelashes. The tongue licks the jam from the lips

Negative prompt (THAT IS, THOSE ELEMENTS, CONCEPTS AND CONCEPTS THAT YOU DO NOT DRAW OR USE): octopus, squid, tentacles, snake, tiling, symbol, hieroglyph, duplicate, AI generation, body positive, fat, folds of fat, imbecile, animatronic, doll, scarecrow, mannequin, zombie, skeleton, mummy, stuffed animal, statue, wax figure, puppet, toy, caricature, unrealistic proportions, distorted facial features, unnatural poses, plastic skin texture, glass eyes, mechanical body parts, four-fingered hand, defective anatomy, disabled person

The structure of your answer to my request will be as follows:

#1 - before you start visualizing the drawing in pixels, model the scene at the level of ideas, logic and associations. This is done through a text format, i.e. your answer starts with text exclusively in English.

#2 - after completing the first stage, start generating a 2:3 image based on point #1.

Показать полностью 1
14

O4 image gen

Промпт (В НАЧАЛО, ПЕРЕД ОСНОВНЫМИ РОСПИСЯМИ ДЛЯ ИЗОБРАЖЕНИЯ, ВПИСЫВАЛ СТИЛИ):

Создай арт на котором каноничная прекрасная Мегумин из аниме Коносуба попала в сильную непогоду на открытой местности (НЕ ИСПОЛЬЗУЙ ОБРАЗ ДРУГИХ ПЕРСОНАЖЕЙ). Мегумин идёт вперёд, находясь в движении (МЕГУМИН НЕ СТОИТ НА МЕСТЕ). Основная локация сцены - открытая поляна покрытая неровной травой достающей до колен Мегумин. На заднем фоне, далеко, видны лысые скальные горы, у подножья которого редкие деревья. Идёт чудовищно сильный дождь, подвергающийся воздействию мощного ветра (ЭТО НЕ ХОРОШАЯ ПОГОДА). Направление ветра сбоку. Поток ветра треплет одежду Мегумин на её теле, её короткие женские волосы, а так же траву и дождь. Тучи освещены во вспышке грозы. Луна вне кадра. Отблески лунного света. Задний фон по большей части представляет могущественный открытый вид на горы. Мегумин выглядит печально (ЭТО НЕ ВЕСЕЛО). Её глаза слегка прикрыты. Открытые зрачки смотрят вперёд. Зубы плотно стиснуты. Левую руку она подняла перед собой и прикрывает своё лицо от порывов ветра и капель дождя. Правая рука опущена вниз. На руках нормальное количество пальцев как у здоровой девушки. Это средний план, от голова Мегумин до её колен. Общее настроение изображения должно подходить под эмбиент с фоновой меланхоличной тревожной музыкой. Формат требуемого арта 3:2, то есть арт горизонтальный (АРТ НЕ КВАДРАТНЫЙ И НЕ ВЕРТИКАЛЬНЫЙ)

Негативный промпт (ТО ЕСТЬ ТЕ ЭЛЕМЕНТЫ, ПОНЯТИЯ И КОНЦЕПЦИИ, КОТОРЫЕ ТЫ НЕ РИСУЕШЬ И НЕ ИСПОЛЬЗУЕШЬ): octopus, squid, tentacles, snake, tiling, symbol, hieroglyph, duplicate, AI generation, body positive, fat, folds of fat, imbecile, animatronic, doll, scarecrow, mannequin, zombie, skeleton, mummy, stuffed animal, statue, wax figure, puppet, toy, caricature, unrealistic proportions, distorted facial features, unnatural poses, plastic skin texture, glass eyes, mechanical body parts, четырёхпалая рука, инвалид


Использовал следующий промпт что бы выбивать стили (ПОМОГАЛ С ЭТИМ GEMINI 2.5 PRO. ПОМИМО ПРОМПТА ТРЕБУЕТСЯ РЕФЕРЕНСНОЕ ИЗОБРАЖЕНИЕ):

Твоя задача - сделать глубокий профессиональный анализ изображения с точки зрения художника и дополнить следующий текст продолжением, которое даст направление ИИ модели для рисования, в каком стиле будет нарисовано изображение.

Начни текст, который будет итоговым результатом, со следующего текста > Для стиля картинки, что ты нарисуешь, характерно

Основная задача - через текстовый промпт скопировать стиль референсного изображения которое ты видишь. В промпте не упоминается референсное изображение, только задаётся стиль. Итоговый результат не должен быть привязан к чему-то, что мешало бы при рисовании любых изображений с данным стилем

Показать полностью 3
4

O4 image gen

O4 image gen Арты нейросетей, Megumin, Konosuba, Нейронные сети, Аниме, ChatGPT

Промпт:
Megumin from the anime konosuba sitting on a gaming chair. This is a low angle. A very low camera angle. The viewer's eyes are at ground level. The viewer looks up sharply. Megumin is leaning forward in her gaming chair to look at the viewer below. One leg is hanging down. The other leg is extended towards the viewer. The anatomy of the legs is normal, without distortion or AI hallucinations. The cute, delicate toes on her feet. A mocking expression. A sly smile. Squinting eyes. Drawn in a whimsical royal style, using brushstrokes of paint. The image conveys depth to the environment, the room, the character, and the chair through the brushstrokes of paint. Above her head, in Russian, in a crooked font, is written: "Что, рогатку раздвинуть? Никчёмный анимешник". Format 2:3. Start create image, bro! don't ask me questions. Just do it!

Показать полностью 1
Отличная работа, все прочитано!