从2022年到现在…🥲 断断续续折腾了整整3年！终于靠着Ai的力量，一个人完成了设计师最不可能完成的挑战：
👉 写出一个「定制化图层拼图程序」🙌
👉 搞定NFT的拼图和metadata！
🎉 计划在 10月1日 上架Opensea盲盒～
可喜可贺，打算2025年10月1日上线opensea盲盒，随意设置了一个毛毛雨价格（大概1美元不到），把自己这个心愿了了，拖得实在是难受了 :(
项目地址：
https://opensea.io/collection/wiseshot
💡这个NFT的起点
这个项目的初衷是和好友Aether在疫情期间一次无聊的闲聊，两人便开始尝试以「核酸」为主题创作一套有趣的头像NFT。不过2022年年末NFT市场实在是太过于萧条，基础traits（图层组件）创作完成之后，就一直搁置了大半年，我是UIUX，没有开发，拼图程序也成为我的心头痛。因为没办法手动拼图，手动处理几千个拼图几乎是要磨死人的节奏。
🛠️修罗场式折腾历程
其间我还作为UIUX设计师参与了小幽灵自己的NFT市场项目ManeStudio，正式实战接触并梳理了整个NFT的发行流程，有想过使用ManeStudio来上架自己的项目，无奈ManeStudio的后台其实暂时无法实现复杂的条件化图层拼接的，紧接着我一顿实验了市场上所有我能发现的NFT发行平台，几乎都遇到一样的问题。但是其中唯一非常nice的产品就是「Bueno」，能够非常人性化的实现条件化图层组合，还能一定程度优化耦合和冲突的情况，我很开心的玩了好一段时间，深度参与了下Bueno社区的活动，但是最后还是放弃使用Bueno来发行项目，因为收费实在是太贵了，需要大概1个eth左右。这个价格，牛市时候硬着头皮上还好，熊市的时候还是有点让人觉得难受的，尤其是发的是一个Art的作品，而不是为了割韭菜的话。
这样就又回到了原点。
还记得23年的时候还通过「良心DAO / lxdao」找到了HashDNA-Art-Engine的项目负责人，咨询能否实现条件化拼图，但是也许是自己对HashDNA使用不甚了解，实验了一堆感觉还是比较难把控最终的效果，就又放弃了。
24-25年Ai项目开始热闹了起来，我自己也开始折腾Ai项目起来。期间偶然在咖啡店打开「manus」开始尝试试试Vibe Coding写一个拼图程序，1个小时的折腾，弄出了一个demo版本的python程序，效果有点让我喜出望外，结果接下来好几天我在manus上付费迭代了好几个版本，但是期间发现manus其实非常不适合迭代代码，太慢了，这个是产品框架上的不适合，我马上换到了cursor付费版，用claude来继续干，这个时候的claude和gpt都很强了，我几乎在一周内完成了可行性比较强的版本。
然后一边试一边继续迭代，迭代到第11个版本的时候完成了～我用v11完成了全部项目的拼图和metadata的处理工作，最后使用cursor直接针对opensea优化了metadata的csv文件。还联系Aether童鞋沟通信息并完成了WiseShot的官网制作（可有可无，但是有的话有趣一些吧）。
最后在figma上完成最后的pr用的banner图之类，到opensea上连接钱包，按步骤进行资料编辑（没想到opensea目前的上架流程针对用户已经算是非常丝滑了），最后钱包确认信息上链，就等着10月1发布了。
写了个流水账，算是对自己这个项目的一个总结吧。
🎨经验分享（有缘人请收好🍀）
1️⃣ Figma能做NFT！
我的整个NFT拼图构成是全部使用FIgma来设计的，其中「组件」功能帮上了大忙，非常适合用于定位和灵活修改设计文件。图形项目非常复杂巨大的时候，打开也不会卡顿到让人心烦。另外，如果不放心线上保存项目的话，也可以自己备份.fig的离线设计文件。
2️⃣ Vibe Coding没那么难！
很多时候遇到需要程序解决的问题， 可以先自己用Ai软件试试看，说不定你觉得是「天塌了」的技术问题，其实没那么夸张。
3️⃣ 对艺术家：建议有条件的朋友都把自己的艺术作品上链作为作品证明
除开传统的作品版权证明，也可以考虑玩一下把作品图上架opensea，适当覆盖一下链上痕迹。毕竟未来链上是多媒体艺术持续开花发芽的地方。
4️⃣ 新人推荐去 https://nft101.lfgkids.com/ 看看
这个也是我前几年写的一个烂尾掉的网站，但是里面还有一些基础的内容可以对新人小白学习了解下（如果有条件有时间我回头再更新吧，写东西好难好累，hh）。
From 2022 until now… 🥲 three whole years of on-and-off tinkering! Finally, with the help of AI, I managed to pull off what once felt like the ultimate impossible challenge for a designer, all by myself:
👉 Wrote a customized layer-puzzle program 🙌
👉 Solved the NFT puzzle + metadata workflow!
🎉 Planning to drop it as an Opensea blind box on October 1st ~
So yeah, mission finally accomplished after dragging this forever. Gonna list it for a drizzle-price (like… less than $1), just to fulfill this long-stalled dream 😭
Project link:
https://opensea.io/collection/wiseshot
💡 Where it all began
This whole thing started during the pandemic with my friend Aether, just a random chat that turned into:
“Hey, what if we make an NFT collection themed around PCR tests?” 😂
We built out the basic traits (layer components) but by late 2022 the NFT market was ice cold, so everything sat on the shelf for months. I’m a UI/UX designer, not a dev, and the puzzle program became my nightmare. No way I could manually stitch thousands of combinations—it would’ve been soul-crushing.
🛠️ The trial-by-fire grind
In between, I worked as a UI/UX designer on a friend’s NFT marketplace project, ManeStudio, where I got real hands-on with the NFT publishing pipeline. I even considered using ManeStudio for my project, but the backend didn’t support complex conditional layer rules.
So I tested literally every NFT launch platform I could find. Almost all had the same issue. The only tool that felt amazing was Bueno—super intuitive conditional layer system, conflict management, really fun to use. I got involved in their community too. But… the cost was brutal: around 1 ETH just to publish. In a bull market maybe you’d grit your teeth, but in a bear market—especially for an art project, not a “pump-and-dump”—it just didn’t make sense.
So back to square one.
In 2023, I even contacted the maintainer of HashDNA-Art-Engine through lxdao, asking if it could handle conditional puzzles. But after messing around with it, I realized it was still too tricky to control the final output. Gave up again.
By 2024–25, AI tools exploded. I started experimenting too. One day in a café, I opened manus, tried “Vibe Coding,” and in about an hour had a Python demo spitting out puzzle combinations. I was shocked. For days I kept paying for iterations on manus, but eventually realized it wasn’t suited for serious coding (too slow).
So I switched to Cursor + Claude (with GPT in the mix). Within a week I had a strong working version. I kept iterating—by version 11, it was fully functional. That version processed the entire project’s puzzle + metadata. Cursor even let me optimize the CSV for Opensea directly.
Meanwhile, I synced with Aether and we put together a simple website for WiseShot (optional, but fun). Made a promo banner in Figma, connected my wallet to Opensea, filled in the details (the flow is surprisingly smooth now), signed the wallet tx, and… it’s ready for October 1st launch!
Kind of a diary-dump, but it’s my little closure on this project ✨
🎨 Lessons learned (take these if useful 🍀)
1️⃣ Figma works for NFTs!
All my NFT layers were built in Figma. Components = lifesaver. Huge complex files stayed smooth, no lag. You can also back up .fig offline files if you don’t trust the cloud.
2️⃣ Vibe Coding isn’t that scary!
Sometimes problems feel like “the sky is falling,” but AI tools can make coding experiments much more approachable. Try first—you might surprise yourself.
3️⃣ For artists: put your work on-chain.
Even if you already have traditional copyright proof, consider minting your art on Opensea. It leaves a trace on-chain, where multimedia art will keep flourishing.
4️⃣ For beginners: check https://nft101.lfgkids.com/
It’s a half-finished site I wrote years ago, but still has some basics for newbies. (Maybe someday I’ll update it… writing is hard, haha 😅).
