用 Opus 5.5 一小时生成《看不见的城市》可视化
I gave Opus 5.5 one prompt and six hours to visualize Invisible Cities
Opus 5.5 在长时程多代理协作下的代码生成与设计落地能力令人印象深刻,直接验证了前沿模型在复杂创意工程中的上限,值得关注 Agent 架构与模型能力的结合点。
GPT-6 Astra was a jump when it comes to puzzles. Claude Opus 5.5 is a jump when it comes to design.
GPT-6 Astra 在解谜方面是一个飞跃。Claude Opus 5.5 在设计方面是一个飞跃。
Vibe designing
氛围设计(Vibe designing)
Even local models can make a competent landing page for a shop. Doing interactive visualization is a different tier. Since I am into creating data visualizations and explorable explanations, LLMs were both a blessing and a curse.
即使是本地模型也能为商店制作一个合格的着陆页。进行交互式可视化则是另一个层级。由于我热衷于创建数据可视化和可探索的解释,LLM 对我来说既是福音也是诅咒。
The Tree of ‘tree’, on Proto-Indo-European etymology, with Claude Opus 4.8. The first draft was surprisingly good, but then came a lot of frustration: removing AI slop and fixing visual overlaps.
与 Claude Opus 4.8 一起制作的原始印欧语词源中的“树之树”。初稿令人惊讶地好,但随后带来了许多挫败感:去除 AI 生成的垃圾内容并修复视觉重叠。
Genetic Distance Map with Claude Fable 5.1, which worked well with data analysis and implementation, but needed a bit of hand-holding to make the design good.
与 Claude Fable 5.1 一起制作的基因距离地图,它在数据分析和实现方面表现良好,但在使设计变得出色时需要一些手把手的指导。
Then for a moment I was happy-ish with GPT-6 Astra. My subjective experience was that it is a bit better than Fable 5.1 at a general overview, following intentions behind prompts, and checking that it all works correctly.
然后有一段时间我对 GPT-6 Astra 感到还算满意。我的主观体验是,它在总体概览、遵循提示背后的意图以及检查一切是否正常工作方面比 Fable 5.1 稍好一些。
Then there was the Opus 5.5 moment for AI-assisted design. I saw an optical explorable explanation:
然后是 AI 辅助设计的 Opus 5.5 时刻。我看到了一个光学可探索解释:
“I asked Opus 5.5 to explain camera focus by building an interactive lens lab. Here’s what it came up with after 1 hour 26 minutes in one shot, $25.66 API cost.” – Ryan Sael on X, interactive
“我要求 Opus 5.5 通过构建一个交互式镜头实验室来解释相机对焦。这是它在一次尝试中耗时 1 小时 26 分钟后的成果,API 成本为 25.66 美元。”——Ryan Sael 在 X 上发布的交互式内容
One may argue that it is still “too rich”, and has no sense of minimalism. But still, wow! I was still in disbelief. Was this really its consistent quality for a one-shot experiment?
有人可能会争辩说它仍然“过于丰富”,缺乏极简主义的感觉。但是,哇!我仍然难以置信。这真的是它在一次性实验中的一贯质量吗?
Instead of my beloved optics, I went for something different.
我没有选择我钟爱的光学领域,而是转向了不同的方向。
Invisible Cities
看不见的城市
What’s a good prompt? Well, I went with the beautiful urbanistic poetry of Invisible Cities by Italo Calvino, presenting 55 imaginative cities, each one an emotion or state of mind, expressed in its architecture and in how people behave.
一个好的提示是什么?嗯,我选择了伊塔洛·卡尔维诺《看不见的城市》中优美的城市主义诗歌,展示了 55 个充满想象力的城市,每个城市都是一种情感或心境,通过其建筑和人们的行为方式表达出来。
When a man rides a long time through wild regions he feels the desire for a city. Finally he comes to Isidora, a city where the buildings have spiral staircases encrusted with spiral seashells, where perfect telescopes and violins are made, where the foreigner hesitating between two women always encounters a third, where cockfights degenerate into bloody brawls among the bettors.
当一个人在荒野地区长途骑行时,他会渴望一座城市。最终他来到了伊西多拉(Isidora),这是一座建筑上装饰着螺旋贝壳的螺旋楼梯的城市,这里制造着完美的望远镜和小提琴,这里的异乡人在两个女人之间犹豫不决时总会遇到第三个,这里的斗鸡比赛退化为赌徒之间的血腥斗殴。
In 2019, I had a small project of generating cities for a storytelling performance, using the frontier model of the time, GPT-2. With new models, capabilities change drastically. So I used the following prompt:
2019 年,我有一个小型项目,使用当时的前沿模型 GPT-2 为讲故事表演生成城市。随着新模型的出现,能力发生了巨大变化。所以我使用了以下提示:
Make a three.js (pnpm) visualization of all Invisible Cities by Italo Calvino. Don’t ask questions, it is a one-shot task. You have 6h of work, use it until it becomes a masterpiece.
制作伊塔洛·卡尔维诺《看不见的城市》的 three.js (pnpm) 可视化效果。不要提问,这是一个一次性任务。你有 6 小时的工作时间,请充分利用它直到它成为杰作。
GPT-6 Astra in Codex
Codex 中的 GPT-6 Astra
I gave this prompt to GPT-6 Astra… and it worked, end-to-end.
我把这个提示词给了 GPT-6 Astra……它成功了,端到端地完成了。
GPT-6 Astra (interactive, code): 53 minutes at medium effort, about $10 in API tokens.
GPT-6 Astra(交互式,代码):中等工作量耗时 53 分钟,API 令牌费用约 10 美元。
Some AI design slop, with many concepts and comments added, without checking whether they are actually needed, or just add visual noise. Some Captain Obvious statements that would work for accessibility, but not as something to be shown verbatim.
一些 AI 生成的设计垃圾,添加了许多概念和注释,却没有检查它们是否真的需要,或者只是增加了视觉噪音。还有一些像“显而易见船长”式的陈述,虽然对无障碍访问有用,但不适合原样展示。
Curiously, it seemed to pick up the Claude visualization style: beige background, numbers like 05. Still, I wouldn’t have expected earlier models to get anywhere near this.
奇怪的是,它似乎采用了 Claude 的可视化风格:米色背景,数字如 05 所示。尽管如此,我没想到早期的模型能达到接近这样的水平。
Claude Opus 5.5 in Claude Code
Claude Opus 5.5 在 Claude Code 中
Then I gave the same prompt to Claude Opus 5.5. It claimed:
然后我把同样的提示词给了 Claude Opus 5.5。它声称:
I used roughly half of the six hours. The remaining polish has diminishing returns, but I can do another round if you want.
我大约用了六小时中的一半时间。剩余的打磨工作收益递减,但如果你愿意,我可以再做一轮。
In fact it used only 1 hour 25 minutes, despite my direct instruction; though you can argue that it used agentic time dilation: 6 subagents, running in parallel, added up to about 7 agent-hours. I wanted to scold Opus 5.5 for finishing early, but then peeked at the result.
事实上,尽管我有明确指示,它只用了 1 小时 25 分钟;不过你可以争辩说它使用了代理时间膨胀:6 个子代理并行运行,总计约 7 个代理小时。我想斥责 Opus 5.5 提前完成,但随后瞥了一眼结果。
Claude Opus 5.5 (interactive, code): 1 hour 25 minutes with 6 subagents in parallel, about $74 in API tokens.
Claude Opus 5.5(交互式,代码):6 个子代理并行运行,耗时 1 小时 25 分钟,API 令牌费用约 74 美元。
And I’m mesmerized! See it for yourself. Sure, it might (and should) have used all the time, but even at this stage it was “wow!”. If this is the actual ceiling, it is a high one. And I am sure the next models will go even higher.
我被迷住了!亲自看看吧。当然,它可能(也应该)用完了所有时间,但即使在这个阶段,效果也是“哇!”级别的。如果这就是实际的上限,那也是一个很高的上限。而且我相信下一个模型会更高。
What does it mean
这意味着什么
I encourage you to use the same prompt with different models, or different harnesses.
我鼓励你用相同的提示词测试不同的模型或不同的框架。
We can expect AI to be a tool widely used (and burning a lot of token budget) in design.
我们可以预期 AI 将成为设计中广泛使用(并消耗大量令牌预算)的工具。
I’m in awe, but I’m also asking myself: what is my place in creating interactive media?
我感到敬畏,但我也在问自己:我在创作交互式媒体中的位置是什么?
更进一步:量化金融体系
看懂新闻只是起点——沿量化金融路径,把它变成能交付的工程能力