这类视觉作品以前经常在 X 上爆火。现在你只需要用 GPT Image 2，加上一段像下面这样的提示词，就能做出来：

These kinds of visuals used to go viral on X. Now you can make them with GPT Image 2 with a prompt like this: [A clean isometric 3D render that turns a single abstract statistic into a physical object you can feel the weight of — soft studio lighting, matte clay materials, gentle ambient occlusion shadows, a neutral concrete-grey backdrop. The point is scale shock: take a number nobody can intuit and rebuild it as one tangible mass beside a familiar object for comparison (all the gold ever mined as a single cube next to a school bus; every plastic bottle used in one minute stacked into a tower beside the Eiffel Tower; the world's drinking water as one sphere next to the dry planet; a year of human attention given to one app rendered as a mountain beside a person). One precise, oversized hand-set numeral anchors the composition, with a short caption in clean grotesque type stating the metric and its source year. Subtle measurement bars and a tiny human silhouette provide scale. No clutter — generous negative space, one hero object, one comparison. The spirit: a science-museum exhibit model that reframes a statistic as architecture, so a figure you'd scroll past becomes something your body understands.]

这类视觉作品以前经常在 X 上爆火。现在你只需要用 GPT Image 2，加上一段像下面这样的提示词，就能做出来： [一张干净的等距 3D 渲染图，把一个抽象的统计数据变成一个能让你感受到其重量的实体物件 —— 柔和的影棚灯光，哑光黏土材质，轻柔的环境光遮蔽阴影，中性的混凝土灰色背景。核心目的是制造尺度冲击：挑一个没人能直觉理解的数字，把它重塑为一个实实在在的物体，放在一个熟悉的物体旁边作对比（人类有史以来开采过的所有黄金，凝成一个方块，旁边放一辆校车；一分钟内消耗的每一个塑料瓶堆成一座塔，旁边是埃菲尔铁塔；全世界所有的饮用水汇聚成一个球体，旁边是干涸的地球；人类在某款 App 上一年投入的注意力，化作一座山，旁边站着一个人）。画面中一个精确的、超大号的手工排版数字作为视觉锚点，配一段简洁的 grotesque（无衬线）字体短标题，说明该指标及其数据来源年份。克制的测量标尺和一个微小的人形剪影提供尺度参照。不要杂乱 —— 留足负空间，一个主角物体，一个对比参照。其精神内核是：像一座科学博物馆的展品模型，把统计数据重新架构为建筑，让一个你本来会一划而过的数字，变成你的身体能够理解的东西。]