Read the article content → ask the user step by step → generate the prompt.
Important: ask only one question at a time, and wait for the user to answer before asking the next.
Question flow (ask one by one, do not ask all at once)
Question 1: Choose a style
List the following 10 styles and have the user choose one:
- Dark Gradient Style — The person is centered, large text covers behind them, dark gradient background, high contrast and strong impact
- Solid Color Flat Style — Cutout feel for the person, solid color background, clean and fresh
- Product Key Visual Style — UI screenshots/products occupy the main body, the person makes a guiding gesture, suitable when there are material images
- Comparison Card Style — The person holds two comparison cards, near large and far small, pushing toward the camera, suitable for before-and-after comparison content
- Minimalist White Space Style — Large areas of white space, text is the main visual hammer, the person appears lightly
- Poster Collage Style — Multiple reference images are overlaid in the composition, layered with depth, suitable for content with rich materials
- Person Offset with White Space Style — The person is offset to one side occupying a small proportion, large space is left for the title, the layout is grand
- Back View Composition Style — The person faces away from the camera, creating immersion and room for imagination
- Partial Appearance Style — Only a hand/half face/profile is shown, the product or text is the absolute protagonist
- Front-Facing Eye Contact Style — The person looks directly at the camera, eye contact, text surrounds the face on all sides
Question 2: Image 1
Ask the user whether there is a face reference image (Image 1):
- Yes → have the user upload it, and write "reference the facial features of Image 1" in the prompt
- No → have the user describe the person (gender, general appearance), and describe directly in the prompt
Question 3: Person's expression
(Back View Composition Style skips this question)
Give the user the following options:
- Hand over mouth in surprise — Both hands over the mouth, eyes wide open, a sense of shock
- Open-mouthed shock — Mouth slightly open, eyes widened, strong surprise
- Happy laughter — Corners of the mouth raised, eyes curved, genuine joy
- Excited and elated — Beaming with joy, body leaning forward, excitement that cannot be hidden
- Confident and smug — Corners of the mouth slightly raised, gaze firm, a "it's just that simple" feeling
- Chin on hand in thought — One hand supporting the chin, gaze thoughtful
- Recommended product-seeding feel — Smiling and nodding, gaze toward the camera, "you definitely have to try this"
- Let the model decide
Question 4: Additional materials
Ask the user whether there are additional materials such as Image 2, Image 3 (UI screenshots, product images, other reference images):
- Yes → ask how many there are and what each one contains
- No → skip
Question 5: Background color tone
Give the user the following options:
- Light color scheme — white/off-white/light gray, clean and fresh
- Dark color scheme — dark gray/ink black, steady and powerful
- Warm tones — beige/warm brown/soft orange tones, warm feeling
- Cool tones — light blue-gray/cyan-gray tones, calm and professional
- High-saturation contrasting colors — the model selects a pair of high-contrast colors based on the topic, energetic feel
- Let the model decide
Note: The minimalist white-space style and the character-offset white-space style default to a light color scheme background; this question can be skipped or only ask about specific light color tones.
Question 6: Font style
Give the user the following options:
- Ultra-bold sans-serif — thick and powerful, first choice for practical/tech content
- Soft rounded font — rounded strokes, warm and friendly
- Handwritten doodle font — natural strokes, strong personality
- Minimal sans-serif — even strokes, international style, design feel
- Retro serif — serif, classical nostalgic feel
- Let the model decide
Question 7: Font color effect
Give the user the following options:
- Pure white — first choice for dark backgrounds, clean and clear
- Pure black — first choice for light backgrounds, steady and powerful
- Gradient color — the model matches it based on the background tone
- Outline effect — adds an outline to the font, increasing layering
- Let the model decide
Question 8: Cover title
Extract 1-3 candidate titles (4-8 characters) based on the article content, and let the user confirm or modify them.
Core requirements for generating the prompt
After receiving all answers, generate the prompt according to the corresponding style template. The prompt must achieve:
- Specific character pose: clearly state the body position, where the hands are, and the details of the action; do not use vague descriptions like "she made a movement"
- Detailed main elements: what the main visual element looks like, what content is on it, size proportions, dynamic or static
- Clear spatial relationships: what is in the foreground/midground/background, and which occludes which
- Background and font differ by style: do not apply the same set of colors and fonts to all styles
Style 1: Dark gradient style
Character centered, large text covering behind the character, dark gradient background, high contrast and strong impact.
Visual features
- Background: dark gradient (user-specified tone, or the model decides based on the topic)
- Font: ultra-bold, white or gradient color, overlaid behind the character
- Foreground: a small number of flat icons or 3D collage objects as accents
- Overall: high saturation, an aesthetic that is both neat and messy
Fixed
- Character: refer to the facial features in reference image 1, half-body only
- Elements do not exceed the safe area
Variables
- Character expression and action
- Main visual element and narrative behavior
- Central large text
- Foreground scattered small elements
- Specific background tone
- Font style and color
Prompt template
参考图1的[性别]五官特征,保持五官一致性,只要半身。
[具体表情描述],[人物身体位置和姿势],[右手/左手的具体动作],
[主视觉元素的叙事行为,描述清楚人物和元素之间发生了什么],
[主视觉元素的详细描述:形状/内容/大小/动态],
(如有图2/图3:[参考图编号]作为[具体呈现方式和位置])
巨大的中文大字"[封面标题]"覆盖在人物背后,[字体风格],[颜色效果],被人物和主体部分遮挡,形成视觉层次感
背景:[深色渐变色调],柔和过渡
所有元素集中在中间区域,上下留边距,不超出安全区
前景散点分布几个极小的[主题相关小元素],数量极少,只做点缀,部分遮挡文字边缘
悬浮元素轻微投影,视觉层次感,高饱和度,又整齐又凌乱的美感
Style 2: Solid color flat style
Character cutout feel, placed on a solid-color background, clean and fresh, with the main prop pushed toward the camera using near-large far-small perspective.
Visual Features
- Background: solid flat color (user-specified or model-determined), no gradient
- Font: partially obscured by the subject, creating a sense of layering
- Foreground: a small number of flat icons with slight drop shadows
- Overall: subject highly saturated, background low saturation, clear contrast
Fixed
- Character: facial features from reference image 1, half-body only
- Elements do not exceed the safe area
Variable
- Character's expression and action
- Main visual element
- Central large text
- Background solid color tone
- Font style and color
Prompt Template
参考图1的[性别]五官特征,保持五官一致性,只要半身。
[具体表情],[人物身体姿势],[手的动作:推出/举起/展示],
[主体道具]占据画面大面积,带明显动态模糊,强烈近大远小透视感,
[主体道具的详细描述:外观/内容/材质],
(如有图2:[道具/界面]显示参考图2内容,极简干净)
背景:[纯色色调],简单纯色扁平,不要任何渐变
画面[上方/中央]超大中文"[封面标题]",[字体风格],[颜色效果],
字体被主体部分遮挡,形成视觉层次感
前景散点几个极小的[主题相关图标],轻微投影,部分遮挡文字边缘
主体高饱和度,背景低饱和,对比清晰
Style Three: Product Key Visual Style
UI screenshots or product images occupy the main body of the image, with the character kept small as a guide, suitable for content with asset images.
Visual Features
- Product/screenshot: occupies 60-70% of the image, clear on a white background, details readable
- Character: small, beside the product making a guiding gesture or pointing with their gaze
- Background: light or neutral color, letting the product be the absolute protagonist
- Font: dark text over the edge of the product, or light text over dark areas
Fixed
- Character: reference image 1, smaller in scale, making a guiding pose
- Image 2 must be provided (product image/screenshot)
- Elements do not exceed the safe area
Variable
- Character position and gestures
- Presentation of the product/screenshot (floating/displaying/filling, etc.)
- Large text position
- Background color tone
- Font style
Prompt Template
参考图1的[性别]五官特征,保持五官一致性,只要半身,人物体量偏小。
人物置于画面[左/右]侧偏下,占比约25%,[具体引导手势:右手伸出指向/侧身朝向等],
[具体表情],眼神朝向主视觉方向,
参考图2作为画面主视觉,占据画面约65%,[具体呈现方式:悬浮在空中/人物身旁展开/充满背景等],
[图2内容的具体描述:界面的哪些部分清晰可见,白底还是深色底,内容如何排列],
(如有图3:参考图3作为[具体位置和呈现方式])
"[封面标题]",[字体风格],[颜色效果],压在产品[顶部/一侧]边缘,形成层次感
背景:[浅色/中性色调,让产品成为主角]
所有元素集中在中间区域,不超出安全区,轻微投影,视觉层次感
Style Four: Contrast Card Style
The character holds two contrast cards, with the foreground large and the background small, pushed toward the camera, suitable for before-and-after comparisons and good-versus-bad comparison content.
Visual Features
- Foreground card: large, bright, occupying the main body of the image, with motion blur
- Background card: small, dark, in the distant background, with perspective shrinking
- Character: centered, holding two cards in each hand, strong expression
- Background: dark gradient, setting off the card contrast
Fixed
- Character: reference image 1, centered
- Clear contrast relationship between the two cards, one bright and one dark, one large and one small
- Elements do not exceed the safe area
Variable
- Specific content and comparison dimensions of the two cards
- Character's expression
- Card style
- Large text
- Background color tone
- Font style
Prompt Template
参考图1的[性别]五官特征,保持五官一致性,只要半身。
[具体表情,要有强烈对比感:得意/震惊/推荐感],她/他居中,
左手捏着一张[小/暗淡/破旧感]的卡片,偏远景,
卡片上写着"[卡片B内容,代表before/坏的一面]",[卡片B视觉描述],
右手将一张[大/明亮/发光]的卡片猛地推向镜头前方,
卡片上写着"[卡片A内容,代表after/好的一面]",[卡片A视觉描述],
前景卡片占画面大面积,带明显动态模糊,强烈近大远小透视感,
两张卡片[具体对比关系描述],对比强烈,
巨大的中文大字"[封面标题]"覆盖在人物背后,[字体风格],[颜色效果],形成视觉层次感
背景:[色调,深色系衬托卡片对比]
所有元素集中在中间区域,不超出安全区
悬浮元素轻微投影,前景小元素部分遮挡文字边缘
Style Five: Minimalist White Space Style
Large areas of white space, text is the main visual hammer, the character appears lightly, overall restrained with a strong sense of breathing room.
Visual Features
- Background: light/white, large areas of white space
- Character: small in scale (20-25%), toward a corner
- Font: large and clean, occupying the main body of the white space area, the only visual hammer
- Overall: very few elements, no embellishment, restraint first
Fixed
- Character: reference image 1, small in scale, toward one side or the bottom
- White space area > 60%
- Elements do not exceed the safe area
Variable
- Character position (left/right/bottom)
- Blank background color (white/off-white/light gray)
- Large text position and arrangement
- Font style (lean thin or medium, not extra bold)
Prompt template
参考图1的[性别]五官特征,保持五官一致性,半身,体量偏小。
人物置于画面[左下/右下/一侧],占比约20%,[简单姿势],
[轻量表情,不要过度夸张],眼神/身体朝向留白区,
画面[对侧/上方]大面积[白色/米白/浅灰]留白,
超大中文"[封面标题]"占据留白区主体,[字体风格,偏细或中等粗细],[深色],
字间距宽松,呼吸感强,
[如有副标题:一行小字灰色,作为补充信息],
背景:整体[浅色色调],极简,无装饰
构图克制,留白是设计的一部分,不超出安全区
Style Six: Poster Collage Style
Multiple reference images layered and overlaid, with a clear foreground, midground, and background relationship, rich narrative and strong depth.
Visual characteristics
- Foreground: the largest asset image or person, with depth of field/shadow
- Midground: person or secondary asset image, with occlusion relative to the foreground
- Background: blurred background assets, rich but not overpowering the main visual
- Typography: pressed between the layers, white or with a stroke effect
- Background: dark, to set off each layer
Fixed
- Person: Reference Image 1
- It is recommended to provide at least one additional asset (Image 2)
- Elements do not exceed the safe area
Variable
- The position of each element in the foreground/midground/background
- Overlay method
- Large text
- Background color tone
- Font style
Prompt template
参考图1的[性别]五官特征,保持五官一致性,只要半身。
[具体表情],[人物姿势和身体位置],
前景:[前景元素详细描述,大小、位置、视觉细节],占画面约[比例],带轻微投影和浮动感,
中景:人物主体,与前景元素有遮挡关系,[人物被前景遮挡哪个部位],
后景:[后景元素描述,轻微模糊处理,内容简述],[数量]个,错落排列,作为背景层,
各元素之间有前后遮挡关系,制造层次和纵深,
巨大的中文大字"[封面标题]",[字体风格],[颜色效果,建议白色描边或纯白],
压在人物和前景元素之间,与多个层次形成遮挡关系,
背景:[深色沉稳色调,衬托各层次]
所有元素集中在中间区域,不超出安全区,悬浮元素有轻微投影,层次感强
Style Seven: Side-Placed Subject with Blank Space Style
The person is positioned to one side, leaving a large area of space on the opposite side for the title, creating a strong, imposing layout contrast.
Visual characteristics
- Person: offset to one side, occupying 25-35% of the width, gaze directed toward the blank area
- Blank area: occupies 60-70%, filled with title text, large and powerful
- Background: neutral or light color, not competing with the layout
- Typography: mainly dark, large size, can be vertical or horizontal
Fixed
- Person: Reference Image 1, offset left or right, occupying 25-35%
- Sufficient blank space on the opposite side
- Elements do not exceed the safe area
Variable
- Whether the person is on the left or right
- Blank background color
- Title arrangement (vertical/horizontal/combination of large and small text)
- Font style
Prompt template
参考图1的[性别]五官特征,保持五官一致性,只要半身。
人物置于画面[左/右]侧,占画面宽度约30%,[具体姿势],
[具体表情],眼神和身体自然朝向[对侧留白区],
画面[对侧]约65%是大面积留白,
超大中文"[封面标题]"占满留白区域,[字体风格],[颜色:深色],
[排列方式:竖排从上到下/横排分两行/主字大副字小等],
[如有补充信息:一行小字灰色在标题下方],
背景:[中性或浅色色调,整体柔和]
人物与文字之间保留足够的呼吸距离,不拥挤,不超出安全区
Style Eight: Back-View Composition Style
The person faces away from the camera, face not shown, creating a sense of immersion, suitable for inspirational, open, new-beginning content.
Visual characteristics
- Person: back view, only the back of the head, shoulders, and back visible, with a sense of motion
- Ahead: the goal or main visual element the person is facing
- Background: grand, expansive, with a sense of depth
- Typography: above the person, loose and expansive
Fixed
- Person: back view from Reference Image 1, face not shown
- Elements do not exceed the safe area
Variable
- Person's orientation and action (raising hands/looking up/walking, etc.)
- Main visual element ahead/above
- Large text
- Background color tone (usually darker, creating a sense of expansiveness)
- Font style
Prompt template
[性别]人物背对镜头,只见后脑、肩膀和背部,发丝清晰,不露脸,
[具体动作:仰头看向前方/双臂展开/缓步向前等],[身体姿势描述,有动势感],
人物前方[或上方]是[主视觉元素的详细描述:内容/大小/状态/动态],
[主视觉元素和人物之间的空间关系描述],
巨大的中文大字"[封面标题]"在人物[上方/左右两侧],[字体风格],[颜色效果],
字间距宽松,与背景有足够对比,
背景:[大气深色渐变,营造开阔纵深感]
画面有[叙事感描述:释放/出发/到达等],代入感强,不超出安全区
Style Nine: Partial Appearance Style
Only hands, half a face, or a side profile are shown; the product or text is the absolute protagonist, and the person is merely a witness.
Visual characteristics
- Product/screenshot: occupies the main body of the image, clear and sharp, details readable
- Person: partial appearance, occupying 15-25% at the edge, not overshadowing the main subject
- Background: light/neutral, making the product clear
- Typography: pressed on top of or beside the product, dark or white
Fixed
- Person: Reference Image 1, only partial appearance
- Elements do not extend beyond the safe area
Variables
- Exposed parts (hand/half face/side face)
- Local actions (click/touch/point/display)
- Detailed presentation of the product/subject
- Large text
- Background color tone
- Font style
Prompt template
画面主体是[产品/截图/卡片的详细描述:内容布局/颜色/关键信息],
占画面约[比例],白底或[底色],清晰锐利,
(如有图2:主体内容即参考图2,[图2的展示方式])
画面[右下角/左侧边缘/底部边缘]局部出镜:
参考图1的[手部特写/半张脸/侧脸轮廓],
[具体局部动作:食指轻触/拿起/指向某个位置],
[局部出镜的细节:手的姿势/脸的朝向/与主体的距离关系],
"[封面标题]",[字体风格],[颜色效果],压在主体[上方/左侧/下方]
背景:[浅色/中性色,让主体成为绝对焦点]
产品是焦点,人物只是见证者,不超出安全区
Style Ten: Direct Gaze Style
The person looks directly into the camera to establish eye contact; text frames the face from above and below, with direct and intense emotion.
Visual features
- Person: centered front-facing, looking directly into the camera, the face is the visual core
- Text: two lines above and below framing the face, not blocking the eyes or mouth
- Background: can be solid high-saturation color, or a dark gradient
- Overall: emotionally rich, direct, impactful
Fixed
- Person: refer to image 1, centered front-facing, looking directly into the camera
- Text does not block the eyes or mouth
- Elements do not extend beyond the safe area
Variables
- Person's expression (emotion must be clear and strong)
- Text arrangement above and below
- Background color tone (high-saturation solid color or dark gradient both work)
- Font style
Prompt template
参考图1的[性别]五官特征,保持五官一致性,半身,正面直视镜头。
[具体表情描述,情绪要强烈明确],[手/身体的配合姿势],与观众形成强烈眼神接触,
"[封面标题前半]"[字数]个字在人物脸部上方,[字体风格],[颜色效果],字号极大,
"[封面标题后半]"[字数]个字在人物脸部下方,[字体风格],[颜色效果],略小,
两行文字上下夹住脸部,形成自然框架,不遮挡眼睛和嘴巴,文字与脸部有呼吸距离,
(如有点缀:脸部两侧各一个极小[主题图标]飘浮,轻微投影)
背景:[色调,高饱和纯色或深色渐变,与人物形成高对比]
人物是绝对焦点,画面直接有力,情绪满溢,不超出安全区
General prompt-writing guidelines
- Concrete metaphors: For abstract concepts, prioritize everyday recognizable physical objects
- No glowing effects: Holographic projections, data streams, and particle effects easily make the image look cheap
- Positive description: Only write what should appear in the image
- Reference image naming: Directly write "image 1/image 2/image 3", do not restate all the details in the image
- Each style has its own background and font logic: Do not apply the same set of colors and fonts to all styles