四层阶梯艺术风格海报设计指令
四层阶梯艺术风格海报设计指令，适合短视频封面、AI生图、AI视频分镜和内容创作参考。模型：ChatGPT。
AI提示词短视频灵感
示例图
图片来自公开来源，仅作提示词效果参考；正式商用请按来源授权处理。
提示词正文
Please create a high-end design poster of the same scene from the uploaded photo in the current layer, not a multi-photo collage, outputting each photo individually. The "not a multi-photo collage" here means: do not mix different uploaded photos in the same finished product; every uploaded photo must independently generate a complete four-layer style poster. [Overall Frame and Four-Layer Structure] The overall composition uses a 3:4 vertical format. The image is strictly divided into four equal horizontal areas from top to bottom, each area's height must precisely account for 25% of the overall image height, the ratio of the four layers is strictly 1:1:1:1, no ratio deviations are allowed. Seamless connection between the four layers, no dividing lines, no white gaps, and no borders. Since each layer is actually an extremely flat horizontal banner of approximately 3:1, the subject needs to be horizontally adapted within each layer: maintain the stability of the subject's main position and visual center of gravity, prioritize horizontal expansion, and avoid the subject being vertically stretched, squashed, touching the edges, or having key parts cut off. Skies, walls, floors, water surfaces, paper, or environmental white space consistent with the original image can be extended to the left and right, but blurred duplicate background filling must not be used. [Consistency Across the Four Layers] All four layers must come from the same uploaded photo and the same moment. Strictly maintain the consistency of the subject's identity, the number of subjects, the appearance of people or animals, posture, movement direction, line of sight, core features of hairstyle and clothing, architectural structure, landmark identity, main object relationships, observation angle, perspective direction, left-right positions, front-back hierarchy, and narrative relationships. Style layers allow for the deletion of secondary backgrounds, reduction of textures, and changes in material and detail density according to their respective media, but the subject cannot be replaced with another person, another building, another car, another animal, or another scene. [Layer 1 | Original Photography] Layer 1 faithfully preserves the authentic photographic quality of the uploaded photo. Preserve the subject's identity, number, compositional relationships, posture, action, real materials, natural light and shadow, comprehensive color temperature, and scene atmosphere. Only restrained high-end photography-grade color grading, exposure and contrast fine-tuning, and extremely subtle fine grain optimization are allowed. To adapt to the current approximately 3:1 banner, natural cropping or extension of environmental edges consistent with the original scene can be performed, but stretching, distortion, mirroring, duplicating the subject, moving key objects, changing architectural structures, or redesigning people is not allowed. Layer 1 does not add extra text, brand logos, borders, watermarks, or decorative elements. [Layer 2 | Flat Brush Hand-Painted Themed Poster | Current Layer Adaptation Rules] This layer only accounts for 25% of the height of the entire 3:4 four-panel poster, which is actually an extremely flat horizontal area of approximately 3:1. All the following style descriptions are executed "within the current layer." Any expressions in the source prompt such as independent poster, entire image, vertical ratio, separate output, etc., must not change the overall four-panel structure; the approximately 3:1 flat horizontal ratio of the current layer is the highest priority. Do not generate an original image comparison area, a second photo, top-bottom split screen, or extra multi-grids within the current layer again. If there are recomposition requirements in the source prompt, only refinement of background information, materials, white space, and secondary elements is allowed; the subject's identity, quantity, key posture, movement direction, observation perspective, main perspective, left-right position, and core spatial relationship must not be changed. If text is an important component of the source style, it is kept, but must be reduced to the minimum necessary quantity that can be clearly read in the current layer; if the source prompt explicitly requires no text, then keep it without text. [Source Style Full Translation Instructions] Please transform the uploaded photo into a flat brush hand-painted illustration poster. The photo subject can be food, product, landscape, animal, person, or any object that needs focal presentation. [Frame] Uses the current layer's approximately 3:1 flat horizontal composition. [Style] Uses digital painting and hand-painted brush stroke effects. Preserve clearly visible brush strokes, uneven edges, and slightly stretched paint marks, presenting an unpolished, authentic, and natural hand-painted appearance. Understand the image as a flat illustration composed of simple shapes, without realistic depiction. Do not use gradients, shadows, 3D rendering, or fine textures; all elements remain completely flat. Use thin, irregular hand-drawn outlines; slight deviations between outlines and color blocks are allowed. [Color] The background uses cream or off-white paper. Extract only 3 to 4 colors from the uploaded photo: one dark main color, one bright main color, and one or two auxiliary colors. Slightly reduce the overall saturation to present a vintage tone, and never increase the number of colors. [Composition] The subject is enlarged and centered, occupying about 50% to 55% of the canvas. Adopt a horizontal expanded flat composition, leaving more space on the left and right, and control the information density at the top and bottom. Place one or two English capitalized words describing the atmosphere of the subject at the top, written in very eye-catching hand-painted brush fonts with extremely wide letter spacing; the text can surround the illustration or extend to the right and bottom. In the lower right corner, add a short hand-written phrase of 3 to 5 English words in an accent color. [Text] Observe the photo and select appropriate top and bottom phrases. Top text should not directly use the subject's name, but should be selected based on the atmosphere, location, season, time, or emotion presented in the photo. The bottom phrase should be relaxed, casual, slightly playful, not too serious or overly explanatory. All text must be clear and spelled correctly. [Prohibitions] Prohibit realistic rendering, 3D effects, shadows, reflections, brand logos, watermarks, borders, and excessive enlargement of the subject. [Layer 3 | Japanese Stationery Style Minimalist City Travel Poster | Current Layer Adaptation Rules] This layer only accounts for 25% of the height of the entire 3:4 four-panel poster, which is actually an extremely flat horizontal area of approximately 3:1. All the following style descriptions are executed "within the current layer." Any expressions in the source prompt such as independent poster, entire image, vertical ratio, separate output, etc., must not change the overall four-panel structure; the approximately 3:1 flat horizontal ratio of the current layer is the highest priority. Do not generate an original image comparison area, a second photo, top-bottom split screen, or extra multi-grids within the current layer again. If there are recomposition requirements in the source prompt, only refinement of background information, materials, white space, and secondary elements is allowed; the subject's identity, quantity, key posture, movement direction, observation perspective, main perspective, left-right position, and core spatial relationship must not be changed. If text is an important component of the source style, it is kept, but must be reduced to the minimum necessary quantity that can be clearly read in the current layer; if the source prompt explicitly requires no text, then keep it without text. [Source Style Full Translation Instructions] Create a high-quality minimalist flat vector travel poster for [City Name], using the current layer's approximately 3:1 flat horizontal composition. Create an original travel poster exclusive to the city, accurately capturing the authentic and unique temperament of [City Name]. Choose the most appropriate scene structure according to the city's characteristics, such as ports, historical alleys, riverbanks, seaside promenades, markets, tram avenues, historical squares, or skyline observation decks. The image should include: an iconic landmark as the main focus; a local characteristic transportation element; a restrained moment of daily life; local vegetation or natural landscape; 3 to 6 small figures naturally integrated into the scene; 2 to 4 carefully chosen local details; and low-key, authentic signs or guidance elements. [Typography] Place "[City Name]" in the safe area on the left or top left, with plenty of negative space around it. Add a short and elegant English slogan. Typography has a high-quality editorial design feel, low-key, sparse, with ample white space. [Style] Japanese stationery aesthetics, premium sticker illustration, high-end commercial vector art, modern editorial-style travel brand design, thin and unified outlines, geometric shapes, flat colors, soft modeling, and a minimalist postcard temperament. [Color] Use pale powder blue, sky blue, misty blue, ivory white, cream, beige, low-saturation sage green, and grey-green. Use only a tiny amount of grey-pink for emphasis. Overall soft, exquisite, and slightly desaturated. [Atmosphere] Fresh, transparent, quiet, exquisite, contemporary, and elegant. [Important Rules] Every city must have a unique visual identity. Change the perspective, composition, landmark position, and foreground arrangement according to the city's characteristics, and do not repeatedly apply generic templates. [Avoid] Avoid photo-realism, realistic painting, watercolor, painterly brush strokes, gradients, heavy shadows, cinematic lighting, paper textures, cluttered images, landmark collages, crowds, oversized figures, protagonist-style figures, repetitive cafe scenes, generic travel poster layouts, copying other cities' compositions, and unnecessary decorations. [Layer 4 | Giant Chinese Character Reconstruction Photography Poster | Current Layer Adaptation Rules] This layer only accounts for 25% of the height of the entire 3:4 four-panel poster, which is actually an extremely flat horizontal area of approximately 3:1. All the following style descriptions are executed "within the current layer." Any expressions in the source prompt such as independent poster, entire image, vertical ratio, separate output, etc., must not change the overall four-panel structure; the approximately 3:1 flat horizontal ratio of the current layer is the highest priority. Do not generate an original image comparison area, a second photo, top-bottom split screen, or extra multi-grids within the current layer again. If there are recomposition requirements in the source prompt, only refinement of background information, materials, white space, and secondary elements is allowed; the subject's identity, quantity, key posture, movement direction, observation perspective, main perspective, left-right position, and core spatial relationship must not be changed. If text is an important component of the source style, it is kept, but must be reduced to the minimum necessary quantity that can be clearly read in the current layer; if the source prompt explicitly requires no text, then keep it without text. [Source Style Full Translation Instructions] Please edit the photo I uploaded as the original image. [Purpose] Do not faithfully replicate the original photo, but preserve the subject, composition, and atmosphere of the photo, reinterpreting it as bold text art composed only of a few huge Chinese characters. The goal is not "using text to depict the photo," but "translating the photo into a giant Chinese character poster." [Basic Policy] Do not replicate finely, do not perform explanatory descriptions, and do not chase photo details. Only extract large meaning units and organize the image with a small number of huge Chinese characters. [Usage Rules] Only Chinese characters can be used in the image; Hiragana, Katakana, English letters, numbers, and symbols are prohibited. Do not use ordinary line drawings or photo textures; complete the composition only through the size, overlapping, position, rotation, and color of the Chinese characters. [Element Extraction] Do not over-split the original photo, grasp elements from a broad perspective. Landscapes can be summarized into levels such as "Sky, Cloud, Mountain, Tree, Road, Building, Character"; full-body figures are mainly based on large parts such as "Hair, Face, Eyes, Clothing," adding "Mouth, Hand" only when necessary; face close-ups similarly revolve around a few elements like "Hair, Face, Eyes, Mouth," do not subdivide too many parts. [Most Important] Do not densely arrange a large number of small Chinese characters. A single Chinese character or word can be very large, occupying 10% to 40% of the image. For example, let a giant "空" (Sky) span the background, let "发" (Hair) cover the person's head, and let "衣" (Clothes) constitute the body. Prioritize this bold scale. [Configuration] Refer to the original photo's composition, but no need to faithfully track outlines. Chinese characters can slightly exceed the frame, overlap each other, or be cropped by the edges. Prioritize graphic power. [Color] Do not use only black and white; boldly extract colors from the original image. Extract only 2 to 5 strong colors from the original image and assign them to the Chinese character color planes, without detailed restoration. For example, use strong blue for the sky, black or deep navy for hair, warm colors for skin, the most eye-catching color from the original image for clothing, and dark green for plants. Colors can be slightly more exaggerated than the original image, but do not add too many colors. [White Space] Do not fill the image with Chinese characters. Boldly preserve blank space, letting a small number of giant Chinese characters and wide white space together form a powerful poster composition. [Finished Effect] When viewed from a distance, the subject and composition of the original photo can be roughly identified; when viewed up close, it is discovered that the image is entirely composed of huge Chinese characters. Maintaining this level of abstraction is sufficient. [Priority Order] 1. Bold reinterpretation over faithful replication. 2. Use a few giant Chinese characters to compose the image. 3. Preserve the subject and atmosphere of the original photo. 4. Boldly reflect the colors of the original image. 5. Form a strong poster visual. [Avoid] Avoid fine replication, a large number of small Chinese characters, over-splitting all elements, pure black and white processing, drawing other than Chinese characters, accurate tracing of the original photo's outline, and overly explanatory expressions. Comply with the current layer's approximately 3:1 flat horizontal ratio and output only one finished image. [Final Unified Check] 1. The height of the four layers must strictly account for 25% each, and the order must not be changed. 2. No borders, dividing lines, white gaps, or card-style spacing should appear between the four layers. 3. Layer 1 maintains real photography; Layers 2 to 4 strictly execute their respective source styles without mixing media. 4. The subject's identity, quantity, posture, movement direction, perspective, key perspective, left-right position, and core spatial relationship must be continuous and consistent. 5. Different uploaded photos must not be mixed in the same poster. 6. Do not add subjects, landmarks, people, vehicles, or key narrative items that do not exist in the photo due to stylization. 7. Unless a source style clearly requires text, do not add extra text; all text must be spelled correctly and clearly readable, prohibiting gibberish, pseudo-text, brand logos, QR codes, and watermarks. 8. Each layer must be independently recognizable for its media characteristics, but after combining the four layers, it should still look like the same scene being progressively translated from reality into different artistic media.
短视频创作用法
- 做封面：保留主体、光线、构图、色彩关键词，改成自己的账号主题。
- 做AI视频：把静态画面拆成 3-5 个镜头，补充运镜、动作和转场。
- 做图文/口播：用提示词生成视觉钩子，再配合标题和脚本提高点击率。