TutorialsTake It Home: Eight Templates and Five Red Lines
Eight prompt templates ordered by your working week, a seven-row visual brief sheet, a tool-selection table organised by scenario, and five red lines to check before publishing.
The eight templates are not ordered by importance but by the real rhythm of your working week: Monday promo → Wednesday poster → Thursday cover → Friday deck, plus four for everyday use.
Every one is given in full. Fill the bracketed slots and it works. You do not need to memorise them — save them.
The eight templates
Template 1 · WeChat Moments / community image (Monday)
竖版9:16朋友圈宣传图。主体：【活动/产品名】；场景：【在哪里、什么氛围】；
风格：【喜庆/简约/国潮】风，主色【颜色】；构图：主标题上方，卖点中间，二维码留位下方；
文字：主标题"【8字以内】"，卖点小字"【一句话】"。文字清晰，层级不超过3层。9:16 is the full phone screen. Always reserve the QR space inside the prompt — pasting it on afterwards usually covers the subject.
Template 2 · Event poster key visual (Wednesday)
帮我设计一张【活动名】宣传海报。
主体：【活动名】宣传海报
场景：【场地、氛围、人物】
风格：【扁平插画/写实/国潮】，主色【颜色】
构图：竖版3:4，标题在上三分之一，信息区在下
文字：主标题"【8字内】"，副标题"【一句话】"，小字"【时间 · 地点】"
约束：画面干净，文字清晰，层级不超过5层
开始前，先问我三个问题，确认信息后再开始This is the structure from run A with four slots swapped. It usually sets the visual baseline for the whole set — it fixes the palette and the rest follow.
Template 3 · Article cover (Thursday)
横版16:9公众号封面。主体：【文章主题】；场景：【画面元素】；
风格：【调性】，【主色】渐变背景；构图：左侧留文字区，视觉重心在右；
文字：大字标题"【6-10字】"，小字"【副题】"。
约束：文字大而清晰，手机小图也能看清标题，文字区留白。Keep the headline to 6-10 characters. Beyond that it turns into a clump on a phone thumbnail and the emphasis disappears.
Template 4 · Slide / deck visual (Friday)
PPT主题"【一句话】"，汇报对象【领导/客户】。
请先问我三个问题（PPT风格、汇报对象、画面比例），
然后输出3条不同风格的提示词（隐喻式/场景式/极简式）。
要求：16:9横版，不含真实数据与人名，风格与商务PPT协调。Ask for three different styles so you have something to choose from. Metaphor suits concepts, scene suits users, minimal suits data — pick whichever is most on topic.
Template 5 · Knowledge card (everyday)
竖版3:4知识卡片。顶部小字"【栏目名】"，中间大字金句"【20字以内】"，
底部小字"【署名/公众号名】"。风格：米白底＋【主题色】边框，极简，突出文字，四周留白充足。The essence is minimal plus large type. These are the easiest to turn into a series: same prompt, swap only the middle quote, and ten cards become one consistent set.
Template 6 · Product image redraw (reference image)
参考这张产品图，保持产品外观不变，把背景换成【场景：如节日货架/居家餐桌】，
光线改为【自然光/暖光】，风格统一为【品牌色调】，比例【1:1主图 / 3:4详情】，
产品形状、颜色、标识保持不变。This one is image-to-image, so upload the reference first. The closing clause matters: without "keep the shape, colour and logo unchanged", AI may happily redraw the product into a different bottle.
Template 7 · One-thing editing (anytime)
这张图保留整体构图和配色，只修改：【如"主标题改为XX""把背景换成夜晚""人物换成3人"】。
其他元素不变，一次只改一处。The universal phrasing. One change at a time is the only reason it works — three changes at once and AI redraws the whole image.
Template 8 · Let AI write the prompt (universal first step)
我要用即梦AI生成一张【海报/封面/卡片】，用途是【什么场合】，
画面上需要写这些文字：【列出】。
请按"主体＋场景＋风格＋构图＋文字"五件套帮我写提示词，写完先问我三个问题再定稿。Use this when you cannot write a prompt at all. It fills the five slots for you — but always read the result, especially the text slot, which AI loves to "improve" beyond recognition.
The visual brief sheet (blank, printable)
Fill seven rows, join them into one sentence, and you have a working prompt.
| Element | What I fill in | Note |
|---|---|---|
| Purpose (poster / cover / deck / Moments) | Purpose first — it determines ratio and style | |
| Subject (what is drawn) | One sentence, as specific as possible | |
| Scene (what environment) | Where, what light, what atmosphere | |
| Style (tone ＋ main colour) | Flat / realistic / Chinese retro; no more than two colours | |
| Ratio (3:4 / 16:9 / 9:16 / 1:1) | Skip this and the image is usually unusable | |
| Text on image (headline / subtitle / details) | The most forgotten and most error-prone slot | |
| What must not appear | Real faces, other logos, internal data |
Joined into one sentence:
Event poster. Subject: community reading event poster; Scene: community library, warm yellow light; Style: flat illustration, red ＋ beige; Composition: portrait 3:4, title in the upper third; Text: headline "读书分享会"
Choosing a tool: by scenario, not by ranking
| Tool | Best suited to | Characteristics |
|---|---|---|
| Jimeng AI (demo tool) | Assets with text: posters, article covers, knowledge cards | Renders Chinese headlines reliably; simple, works on phone and desktop, free tier is enough for everyday use |
| Kling AI | Realistic imagery, portraits, texture-heavy work | Detailed rendering and lighting; also does image-to-video |
| Tongyi Wanxiang | E-commerce product shots, frequent style variation | Rich style library; background swaps and scene redraws are smooth |
| Wenxin Yige | Chinese classical and traditional aesthetics | Strong on Chinese style and traditional patterns |
| Doubao | Iterating in conversation, absolute beginners | Conversational editing — describe the change, no prompt memorisation |
How to choose: text-heavy posters go to Jimeng; realistic imagery is finer on Kling; e-commerce product shots are easier on Tongyi Wanxiang; Chinese classical subjects suit Wenxin Yige; and if you cannot write prompts at all, start by chatting in Doubao. Each has its strengths; none replaces the others.
We demo one tool in class so that nobody gets stuck choosing. The principle does not change: swap the tool and the five-part formula and the templates still work.
The five red lines
Images ship fast and fail fast. Spend 30 seconds on these five before publishing — it carries your name and your responsibility.
| Red line | How to hold it |
|---|---|
| AI makes things up with confidence | Dates, times, prices and phone numbers must be checked by you — read every word on the image before publishing |
| Never feed confidential or sensitive information | Internal data, unpublished plans, other people's portraits — what you send cannot be taken back |
| The output is not the final draft | It carries your name; a typo or wrong date is on you |
| No misuse of portraits or brand logos | No real celebrity faces, no other companies' logos — copyright and portrait rights apply |
| Label AI-generated images | Mark them as AI-generated when publishing publicly |
Key points from this lesson
| Point | In one line |
|---|---|
| Templates follow the work week | Monday promo → Wednesday poster → Thursday cover → Friday deck |
| Master one template first | Run the one you need next week until it is boring, then copy the pattern |
| Seven rows on the brief sheet | Joined into one sentence it becomes the prompt — far faster than improvising |
| Preserve the product in redraws | Template 6's closing clause cannot be deleted |
| Choose tools by scenario | Text to Jimeng, realism to Kling, e-commerce to Tongyi, classical to Wenxin |
| Run the five red lines | Especially reading every word on the image before it goes out |
Hands-on: One Story, Three Images
Using a community reading event as the through-line, three prompts in sequence — portrait poster, landscape cover and slide visual, all produced live, each with the full copy-ready text.
Assignment: From Reproducing Three Images to Building Your Own Workflow
Three progressive parts in one assignment — reproduce the three in-class images with the course brief, run one real image you need this week through the full process, then turn it into three role-specific templates. All three required.