Turning a Recommendation into a Creative Brief — AI Image Prompt Guide
The source is a recommendation rather than a complete prompt. Clarify the subject, setting and style before turning it into an actionable creative brief.
English translation of the public case notes.
Information-quality case rather than a reusable generation prompt. The original post provides only Japanese recommendation copy and a short link; no complete public generation instructions are available. To reproduce a result, obtain the original prompt, intended image and any reference images. Separate promotional language from executable instructions. Subject, setting, style, composition, material, lighting and exclusions still need to be specified. Do not infer a prompt from the short link or invent unpublished model settings, author intentions or visual details. Practical controls: • Save the complete prompt and result image together when archiving a post, in case the short link expires. • Record recommendation text, actual prompt, reference images and editing requirements in separate fields. • Fill in the missing subject, scene, style, light and exclusions before testing. This case does not supply those missing instructions.
Original case notes (Chinese)
🧩 GPT-Image-2｜只有推荐语，不是提示词 类型：文生图 / 图像编辑。 真正难点在这里：这例更像一次信息甄别：原帖只给出日文推荐语和短链接，没有公开可复用的生成指令。适合提醒收藏者保存完整提示词、参考图与编辑目标，避免把宣传文误当作可执行提示。 控制逻辑拆开看： 类型：文生图 输入：目前只有日文推荐语与短链接；若要复现，需要补充完整提示词、目标画面和参考图（如有） 目标：把社交帖信息转成可执行的生成需求，而不是直接用推荐文生成图片 关键词：日文推荐语 / 短链接 / 信息缺失 / 提示词补全 / 可复现性 重点：先确认真正的原始指令，再区分“宣传文案”和“生成提示”；画面主体、风格、构图、材质与负面约束都不能缺席 约束：避免把链接内容臆测成提示词；不要补写未公开的模型参数、作者意图或画面细节 💡 复刻时抓住三点 • 保存原帖时同时截取完整提示词与结果图，防止短链接失效 • 将推荐语、提示词、参考图、编辑要求分栏记录，便于复用 • 复现前先补齐主体、场景、风格、光线和禁止项，再进行测试 #GPTImage2 #OpenAI #AI绘画 #提示词拆解 #日文短帖 #信息补全