Video yükleniyor...
Video Yüklenemedi
Zero game dev experience to building this in one week The creator downloaded Unity for the first time 7 days ago after discovering it supports MCP, then started building with AI Claude Code handles the development, Meshy generates 3D models, ElevenLabs creates sound effects, and Nano Banana 2 helps... show more
35,832 görüntüleme • 2 ay önce •via X (Twitter)
35 Yorum

别再胡乱鼓吹了,对于一个领游戏开发经验的人,能在一周之内想好自己要做什么都不容易,还能实现视野外怪物泛红提示,如此准确的火焰特效和附着身体的效果?还能明白用 shader 去处理一些光效?还能实现良好的音效配置和对齐?忽悠下外行,然后出售自己的产品,这才是你的目的吧。你这样的发言,真的是对游戏开发者最大的不尊重。AI 很厉害,但绝对无法无中生有地创作出你都没想好的东西。

this game or any game built with ai has no value. has no spirit. no one will buy these kind of shit

Want to build your own project with AI? QuadcodeAI is an all-in-one AI platform where you can access frontier coding, image, video, and reasoning models in one place Use different models for coding, assets, visuals, and other parts of your workflow without constantly switching platforms Use code WLCM0730 to get 1000 free credits

that's one way to generate slop fast.

He still has zero game dev experience

A part of that curve is definitely collapsing, but these news are always about weeks of vibecoding and never months, so it's probably not a large part of the curve

Absolutely. How many concurrent users does it currently have?

The idea that you need three years of engine math and shaders just to ship a playable prototype is officially dead. Stacking Claude Code, Meshy, and ElevenLabs lets solo beginners move faster than bloated 20-person indie teams.

Steam will be full of these in the next year or two. They won't have any value but I guess it's fun to mess around.

Making a game is easy. The ongoing game and community management is the hard part.

I have seen almost this same game highlighted on streams for playing and laughing at terrible games. And it was deserved.

it's alright, it's something you could do in a night. it's impressive for AI though, so kudos.

Wow that looks so cool. Btw which engine is used here?

AI把入门门槛压低后,原型验证会变成更高频的学习循环

The compressed learning curve is real, but the interesting engineering question is reproducibility: can the same agent recreate a working build when asset versions, prompts, and generated code change? A pinned toolchain plus deterministic smoke tests would make that progress durable.

Looks excellent man!

impressive stuff. ive already been experimenting with AI and unity, using it for scripting, folder structure, etc. Havent really tried using all these tools though. Very interesting.

The flashlight cone on the brick looks so sick

Nah it's not gonna scale. Try making a couple of more scenes and watch it explode

It is a shame that people don't want to pay for this kind of product.

There are a billion games that look and play exactly like this.

same

If you don't understand the code, you didn't write it

Nice, reminds me of

That’s wild. AI is making game development far more accessible for beginners.

Super cool shows how far Ai has come in game dev. That being said in my experience once you add a bunch of game mechanics the ai just starts making poor choices on how to do stuff.

Gamrig allows you to get 3d models, SFX, music and voices all that for free! Let me know what you think 😉

one week to something that looks like this is real, and the wall comes at week three when you need save systems, input remapping and a build that runs on someone else's machine. the demo is the easy ninety percent.

Wonder how much this demo cost

But why? I have already played this game many times bedore. Token wasting

Where can we play it?

We havnt got much time fellow devs , the end is near .

AI is definitely bringing the barrier to entry down, making a quick MVP has never been easier. a week to something playable with zero prior experience would've been unthinkable not long ago

Very bad to be honest, It would be great if you used Sol.

This is the real AI disruption in game dev: not replacing developers, but collapsing the learning curve. If you can describe what you want, AI can now help you build what used to require years of specialised skills. And we’re still ridiculously early. 👀
