Монтаж | ИИ с ALSI

Решила поделиться с вами одним из уроков моего курса по монтажу и нейросетям 👀 🌸Тема: идеи и насмотренность 🌸Урок находится здесь Если хотите прокачать свои навыки в монтаже, научиться создавать необычные ролики и внедрять нейросети в свои видео — буду ждать вас на обучении! Новый набор в группу уже активно идёт 😍 Стартуем 18.09! Будем много практиковаться, монтировать, работать с нейросетями и создавать работы, которые действительно выделяются 🤩 Подробную информацию о программе и обучении можете посмотреть на сайте — здесь А по всем вопросам пишите @aliya_sng 🤍

Инструкция: 1. Переходим в VeoSeeBot 2. Нейросеть Kling 3.0 Motion Control Вставляем свое отснятое видео и такой промпт: Transform the entire uploaded video into the exact 2D comic-book illustration style of the provided reference image. Keep the original video completely unchanged in terms of motion, timing, performance, composition, and camera. Preserve every original movement, gesture, pose, head turn, hand movement, body movement, facial expression, eye movement, blinking, and lip movement exactly as in the source video. Only change the visual appearance and rendering style. Convert the woman, her clothing, skin, face, hands, hair/hijab, accessories, objects, room, background, lighting, shadows, and every visible element of every frame into the same cohesive 2D illustrated comic-book style as the reference image. Use: clean hand-drawn outlines detailed comic-style linework subtle ink texture and cross-hatching soft cel shading slightly textured 2D painted surfaces cinematic contrast and lighting rich deep shadows vivid but controlled colors polished modern graphic-novel aesthetic The result must look like the original video was entirely drawn and animated in this 2D comic style, not like a filter placed over realistic footage. Maintain strong temporal consistency between frames. The illustration style, facial features, clothing details, proportions, outlines, colors, textures, and lighting must remain stable throughout the entire video with no flickering, morphing, warping, or style changes. Do not change the woman’s identity or facial structure. Do not beautify or redesign her face. Preserve her exact facial features and proportions while translating them into the illustrated style. Do not add new movements or objects. Do not remove anything. Do not change the background, framing, perspective, camera angle, camera movement, or crop. The only transformation should be from realistic live-action footage into the exact 2D comic-book visual style of the reference image. 3. В монтажной программе поменяем маску на сгенерированном видео и по точкам кадр за кадром двигаем за движением пальцев рук

Инструкция: 1. Переходим в VeoSeeBot 2. Нейросеть Seedance 2.5 3. В раздел с референсом вставляем отснятое видео и в раздел с промптом такой запрос Do not change camera angle, camera stays stationary. when the woman in the video jumps up with her arm up, her finger tips stick onto the ceiling and she dangles from the ceiling from her fingers like spiderman, with subtle swaying of her legs and body as she stays suspended off the ground.

Инструкция: 1. Переходим в VeoSeeBot 2. Нейросеть Seedance 2.5 3. В раздел с референсами вставляем свое отснятое видео, где вы заходите в комнату и как бы тяните за собой животное. Также в раздел с промптом вставляем такой запрос (у меня с ламой, можете заменить на свое животное) Edit the existing video. Keep the original video, girl, camera angle, location, lighting, and movements unchanged. In the existing video, the girl is already holding an invisible leash and walking into the house. Add a small realistic llama attached to the leash, walking naturally behind her as she enters the house. The llama should follow the girl naturally and stay connected to the leash. After the girl leaves the frame, the llama continues walking toward the camera. At the end, the llama comes extremely close to the camera, looks directly into the lens, and starts sniffing it. Finish with an extreme close-up of the llama’s face and nose right in front of the camera, as if the llama is curiously sniffing the camera lens. Make the llama look completely realistic and naturally integrated into the original footage. Do not change anything else in the original video

Инструкция: 1. Переходим в VeoSeeBot 2. Нейросеть Seedance 2.5 3. В раздел с референсами вставляем 4 фото: человек просто стоит, человек стоит и рука в позе спайдер мена, человек держит предмет и крупный план самого предмета. И в раздел с промптом вставляем такой запрос: Use [IMAGE 1] only as the visual reference for the item/object. The object should match the item shown in [IMAGE 1] accurately, including its design, proportions, colors, materials, shape, surface details, and overall appearance. Do not use [IMAGE 1] as the scene composition or background. The object must be recreated and integrated naturally into the main scene. Use [IMAGE 2], [IMAGE 3], and [IMAGE 4] as the primary scene and action references. Preserve the same room, background ighting mood, subject identity, clothing, and overall environment. The animation should happen in my scene, not in the object reterence image Create one continuous cinematic shot. The scene starts with the model standing in a normal neutral position in the room. Then the object drops down into frame from above camera, entering from the top of the frame. It should appear large, close to the lens, and in the foreground, as if falling dire front of the camera. Make this moment feel dynamic and intentional, not static. The camera should shift focus onto the falling obje so that the object is tack sharp and in focus, while the background and the model become softly blurred with shallow depth of field and bokeh. As ine objectalls, it slows down in midair and enters a dramatic slow-motion moment. It should feel like it is floatina brieflv in front of the camera. The object must look realistic, professionally lit, sharp, and naturally part of the environment, with believable reflections, shadows, and perspective matching the room. At that moment, the model performs the Spider-Man web-shooting hand gesture. A visible, realistic Spider-Man-style web shoots out from the hand in slow motion. The web should shoot forward quickly, attach clearly onto the object, and wrap onto it in a one dad me a ve mis de visible ana strandlike, sinlar to a opider vian wee ical, not fabric-like, cloth-like, or thick like ribbon. Avoid a soft fabric app stretch isnould reelike a strona sviere wee ume wima clean solder-Man aesmerc, nounke ciom or smaune siranos snould loc slightly glossy and tight, with natural tension as they shoot forward, latch onto the object, and retract. Keep the web elegant, sharp, and cinematic while still feeling grounded and bellevable. Then the web retracts and pulls the floating object all the way back toward the model's hand. The motion should feel fast smooth, and satisfying, with the object traveling from the foreground back scene. Important: - [IMAGE 1] is only for the object design reference [IMAGE 2), [IMAGE 3], and [IMAGE 4] define the main scene, subject, and action • Do not anenor too neavily to ine object reterence traming - The object must actually fall in from the top of frame and exist inside the main environment Keep the result as one continuous shot - Ivo jumo culs, no teleporne, no sudden appearance drune oulec • No fake fade transitions - The object motion, focus pulls, and web interaction should feel cinematic, smooth, and physically believable Style: cinematic, sharp, professional lighting, realistic integration, shallow depth of field, dramatic slow motion, clean VFX, strong foreground-background separation. AUdIO: No music at all. No soundtrack, no background music, and no chimes. Use only professional sound effects and natural ambient audio. The audio should include subtle room ambience, clean slow-motion sound design, and a satisfying Spider-Man-style web sound design should feel polished, cinematic, and professional, but strictly SFX only with no music or melodic elements

🔠🔠🔠🔠🔠🔠🔠 Инструкция: 1. Переходим в VeoSeeBot 2. Нейросеть Nano Banana Pro 3. Вставляем фото своего помещения и пишем промпт. Можно попросить chat gpt вам помочь. Вот мой промпт для подвешенной машины: Add a brand-new modern luxury white car attached upside down directly to the ceiling. The car is completely inverted, with its roof facing the floor and all four wheels pointing directly upward toward the ceiling. CRITICAL: all four wheels are firmly and physically attached to the ceiling. Each wheel is secured directly to the ceiling with strong metal mounting brackets, bolts and heavy chains. The car must NOT appear to float or hang loosely. The four wheels must be the clear structural connection points between the car and the ceiling. The car should be pristine, brand-new, glossy white, with a sleek modern luxury design, clean bodywork, polished details, realistic reflections and premium materials. Add several heavy metal chains hanging down from each of the four wheel attachment points, plus a few additional chains attached to the underside of the vehicle. The chains should hang naturally under gravity and have realistic weight and thickness. Make the connection between the wheels and ceiling extremely convincing: visible metal brackets, bolts, mounting plates, contact shadows and subtle ceiling deformation around each attachment point. There must be absolutely no visible gap between the four wheels, their mounting hardware and the ceiling. Keep the original room, ceiling, walls, lighting, perspective, camera angle and composition completely unchanged. Photorealistic, cinematic, highly detailed, realistic physics, natural shadows and seamless integration 4. Дальше переходим в CapCut. Вставляем сгенерированное фото с новым потолком и вырезаем человека, наложив его поверх ИИ фона

Посмотрите, что уже творят ученики всего на 4-й неделе обучения 🤯 Этот поток просто разрывает. По баллам — один из самых сильных: все идут в хорошем темпе, никто не сдается. Уже не могу дождаться финала, когда будем подводить итоги, определять победителя и вручать главный приз — новый MacBook 💻🔥 *Запись на следующий сентябрьский поток уже открыта, кто хочет быть с нами, пишите @aliya_sng Подробнее про курс смотрите здесь

Инструкция: 1. Переходим в VeoSeeBot 2. Нейросеть Seedance 2.0 3. Вставляем свое видео, где вы делаете вид, что растягиваете лицо (можно и в разговорном формате) И также такой промпт: Turn the person's pretend face-stretching gesture into a real, exaggerated rubbery cartoon effect — smooth and clean, with no glitching, flickering, or warping artifacts. The moment their hands touch their cheeks, the face skin begins stretching outward, following the hand movement exactly and in sync from the video. As the hands pull outward, ALL facial features stretch elastically along with the skin — the eyes widen and stretch sideways, the mouth stretches wide, and the cheeks pull far out to the sides like elastic rubber, tracking the hand motion the whole time. At the end, when the hands let go and release the cheeks, the stretched face springs and shrinks smoothly back to its normal, natural shape, all features returning to normal proportions. Keep the exact same face, identity, skin tone, lighting, background, and camera framing throughout. Keep the mouth movements matched to the original speech

Делюсь с вами сайтом, где можно использовать множество готовых анимаций Теперь не нужно это делать часами вручную, просто вставляешь свои материалы в уже готовые шаблоны и применяешь в работе 😏🔥 Сайт здесь С вас «🤍», если было полезно

Инструкция: 1. Переходим в VeoSeeBot 2. Нейросеть GPT image 2 Вставляем свое реальное фото в полный рост, референс образа и пишем такой промпт «Замени одежду на девушке на первом фото на образ из второго фото» Так делаем 5-6 разных вариантов, при этом фон везде один и тот же. Также нужно отдельно сгенерировать фото каждой сумки на белом фоне. 3. Переходим в Seedance 2.0 Вставляем образ1, сумка1, образ2, сумка2, образ3, сумка3 И пишем промпт, который я оставила в комментариях (обязательно подстройте под себя через чат гпт) Итого у вас есть первая часть 15 сек видео. Точно также генерируете вторую часть, вставляя оставшиеся сумки и образы. Можно брать разные предметы: кепка, очки, часы и тд