正在加载视频...

视频加载失败

Only if education could be this interactive ❤️‍🔥 I've had a looong wish to build something genuinely useful through vibe coding, and I finally did it. A 3D human anatomy application built with Three.js using GPT 5.6 Sol. It all started with a single design image that I created...

2,099,446 次观看 • 1 个月前 •via X (Twitter)

67 条评论

Three.js 的头像
Three.js1 个月前

❤️‍🔥❤️‍🔥❤️‍🔥

The Bugged Dev 的头像
The Bugged Dev1 个月前

Thank you. Let's go ♥️

Dilum Sanjaya 的头像
Dilum Sanjaya1 个月前

@threejs Looks amazing! I started making those science demos hoping more people would build science related apps with AI, so it's really nice to see demos like yours

The Bugged Dev 的头像
The Bugged Dev1 个月前

@threejs Thank you so much. Let's keep building such more science demos for better education for better future using AI 💪

Guillermo Rauch 的头像
Guillermo Rauch1 个月前

@threejs So cool

The Bugged Dev 的头像
The Bugged Dev1 个月前

@threejs Thank you so much 😃

Timur Yessenov 的头像
Timur Yessenov1 个月前

@threejs The useful part is the boring asset chain: design image → organ images → 3D models → Codex. Most 'vibe coding' demos hide that middle, but that is exactly where the app becomes believable.

The Bugged Dev 的头像
The Bugged Dev1 个月前

@threejs Maybe we can automate that "boring" asset chain using autonomous agents? But either way if AI can help us produce this kind of results, it's all WORTH it, right?

Alec Zakhary 的头像
Alec Zakhary1 个月前

@threejs 900 MB → 28.6 MB is the real lesson. AI made the idea cheap enough to build; the browser forced it to become a product. Next test: give it to ten anatomy students. The first thing they misunderstand is worth more than the next hundred prompts.

The Bugged Dev 的头像
The Bugged Dev1 个月前

@threejs I've already started sharing it with people I know who are studying these biology concepts 😄

Alec Zakhary 的头像
Alec Zakhary1 个月前

@threejs Don’t ask if they like it—watch where they pause, misclick, or need an explanation. Those moments are the roadmap. One confused student is more useful than a hundred developers saying “cool demo.”

Nick Sawinyh 的头像
Nick Sawinyh1 个月前

@threejs The 900MB to 28.6MB drop is where the actual product work starts. I'd still test decode time and memory on a low-end phone. Asset size can look solved while the browser quietly chokes.

The Bugged Dev 的头像
The Bugged Dev1 个月前

@threejs I've tested it on my iPhone too and it's just freaking smooth. No lags.

Tripo 的头像
Tripo1 个月前

@threejs Incredible execution. Seeing Tripo used to bridge the gap between 2D generation and interactive 3D education is exactly why we build these tools!

The Bugged Dev 的头像
The Bugged Dev1 个月前

@threejs Thank you so much. Tripo is so easy and super amazing!!

Hridoy Reh 的头像
Hridoy Reh1 个月前

@threejs This is so cool, man. Loved it… 👋

The Bugged Dev 的头像
The Bugged Dev1 个月前

@threejs Thank you so much 😃

Priya 的头像
Priya1 个月前

@threejs this should get adapted in our school education and syllabus..

The Bugged Dev 的头像
The Bugged Dev1 个月前

@threejs Yess yess, this is how everyone should get to learn

Poimandres 的头像
Poimandres1 个月前

@threejs React and Three, but no React Three Fiber? 🤔😄

The Bugged Dev 的头像
The Bugged Dev1 个月前

@threejs Ah, my bad prompts. I miss it every now and then but I love React Three Fiber and drei ♥️

Poimandres 的头像
Poimandres1 个月前

@threejs It is curious to me that the agent wouldn't try to go this route naturally and reduce the token usage.

Poimandres 的头像
Poimandres1 个月前

@threejs Yeah that’s interesting. From an entropy reduction point of view, it should trend towards R3F. A good question for one of the harness engineers.

Evan Bacon 🥓 的头像
Evan Bacon 🥓1 个月前

@threejs Great work!

The Bugged Dev 的头像
The Bugged Dev1 个月前

@threejs Thank you 🫡

Catatan Remote by Ade 的头像
Catatan Remote by Ade1 个月前

@threejs This is the kind of vibe coding demo I actually want to see. Not just another AI wrapper landing page, but something that makes a difficult subject easier to explore.

walden 的头像
walden1 个月前

Not to be a downer, but are we sure all of the anatomical imagery and data are 100% accurate? That’s important to say the least. This is very cool but it’s only useful in the real world if experts in the field validate it.

The Bugged Dev 的头像
The Bugged Dev1 个月前

We can definitely achieve high accuracy than this. There are lots of of research focused tools, like @Apodex_AI, that can help with that :)

Inflowzy 的头像
Inflowzy1 个月前

@threejs This is exactly the kind of stuff AI should be used for.

The Bugged Dev 的头像
The Bugged Dev1 个月前

@threejs

Saul Flores Jr. 的头像
Saul Flores Jr.1 个月前

If I had a tool like this during school, I would have definitely taken anatomy and become a doctor. Unfortunately, I just became a simple aero engineer lol. The visuals are so amazing. I love this so much. Definitely gonna share with my mom who's a nurse. I think she'll definitely enjoy this.

The Bugged Dev 的头像
The Bugged Dev1 个月前

@threejs Same here. I just couldn't understand visualize feel these concepts back in school, so I chose Math instead 😂 Do let me know what your mom thinks about it :D

Saul Flores Jr. 的头像
Saul Flores Jr.1 个月前

It's like electrical engineering. You study electricity, you can't really see it, but it's there. That's why I went into mechanical engineering instead with structures and things that I can actually feel in my hand like 3D printing and machined parts. I keep thinking about a human heart. It's not something where you can just casually go and you know touch it with your hands, right? Lol. For sure. I think she'll definitely like it. It would be nice to see a cardiovascular system too.

Anvar Basharov 的头像
Anvar Basharov1 个月前

@threejs The UI is beautiful, but generating organ images with AI and converting them to 3D creates something that looks anatomical, not something students can trust. This is exactly the hard part we focus on at Easy Anatomy. AI can build the interface — it shouldn’t invent the anatomy.

The Bugged Dev 的头像
The Bugged Dev1 个月前

@threejs Hmm, I haven't verified everything in detail yet, but I think this could be useful for school and college students to some extent for sure :)

Anvar Basharov 的头像
Anvar Basharov1 个月前

@threejs As a prototype, it’s impressive. My concern is that beginners are the least able to spot anatomical errors. Once it’s used for education, every structure and label needs a trusted source. Happy to share what we’ve learned building Easy Anatomy.

Siddhartha Saxena 的头像
Siddhartha Saxena1 个月前

@threejs this might be the start of "programmable textbooks"

The Bugged Dev 的头像
The Bugged Dev1 个月前

@threejs We're just getting started :D

Daksh Tyagi | AI & SaaS 的头像
Daksh Tyagi | AI & SaaS1 个月前

@threejs Compressing 900MB of 3D assets down to 28MB while preserving quality is incredible work.

Nico 的头像
Nico1 个月前

@threejs Genuine respect. Making anatomy learning this interactive is the kind of work that matters.

𝙳𝚊𝚟𝚒𝚍 𝙲𝚊𝚗𝚎𝚜 的头像
𝙳𝚊𝚟𝚒𝚍 𝙲𝚊𝚗𝚎𝚜1 个月前

@threejs Not to crush the project it’s great work, but you know those diagrams are not anatomically, correct

The Bugged Dev 的头像
The Bugged Dev1 个月前

@threejs It was just a simple image-to-3D conversion without using any of Tripe's advanced modeling features. I'm sure we could get a much more accurate result by providing multiple reference images from different angles, along with some additional details about the object

𝙳𝚊𝚟𝚒𝚍 𝙲𝚊𝚗𝚎𝚜 的头像
𝙳𝚊𝚟𝚒𝚍 𝙲𝚊𝚗𝚎𝚜1 个月前

@threejs All high quality reference images are copyright protected afaik

Daksh Tyagi | AI & SaaS 的头像
Daksh Tyagi | AI & SaaS1 个月前

@threejs This is a prime example of what vibe coding can actually achieve when executed properly.

Zico 的头像
Zico1 个月前

@threejs Added an AI voice agent to this. It controls the entire 3D interface and provides real time anatomy tutoring

Creed Hardcastle 的头像
Creed Hardcastle1 个月前

@threejs when I click cross section on the brain its empty which is exactly how my brain is - 10/10 recommend

Ofentse Nglazi 的头像
Ofentse Nglazi1 个月前

@threejs Very impressive demo but dangerous for serious use. Medical tools especially learning tools are not something you can just vibecode. Most of these diagrams are inaccurate and should not be used by any self respecting school.

whacka 的头像
whacka1 个月前

@threejs This is amazing. If you could simulate surgery on it it would be so meaningful

The Bugged Dev 的头像
The Bugged Dev1 个月前

@threejs Simulate surgery? 😂

$CAGE 的头像
$CAGE1 个月前

@threejs The human body is now 28.6 MB. The soul is still loading.

Bala Venkatesh 的头像
Bala Venkatesh1 个月前

@threejs 🌐 Demo: 💻 GitHub:

Eyisha Zyer 的头像
Eyisha Zyer1 个月前

@threejs this is where AI becomes a real learning tool, not just a demo

The Bugged Dev 的头像
The Bugged Dev1 个月前

@threejs AI is the best thing happened to humanity, tbh!

Command Code 的头像
Command Code1 个月前

@threejs wow so good!!

The Bugged Dev 的头像
The Bugged Dev1 个月前

@threejs Thank you :D

David J. 的头像
David J.1 个月前

@threejs Built a 3D anatomy app. Codex handled the optimisation. Nobody had to understand WebGL.

The Bugged Dev 的头像
The Bugged Dev1 个月前

@threejs Yep, I told it to optimize the website to improve performance and it just did it :)

Dev Patel 的头像
Dev Patel1 个月前

@threejs cool! we built something similar but using 3d models with poly-mesh expansions to learn about specific parts of any model a while back. any model turned into an interactive learning exp try it out with some preloaded models: cc: @kushpatell_

Jianping Liu 的头像
Jianping Liu1 个月前

@threejs Really cool. How much does it cost to generate the 3D models?

The Bugged Dev 的头像
The Bugged Dev1 个月前

@threejs Good question. Cost was between $1-2. I had bought starter plan of Tripo AI

Aditya Pattanayak 的头像
Aditya Pattanayak1 个月前

@threejs you have unknowingly created an useful medical tool!! on top of it ,its super smooth

The Bugged Dev 的头像
The Bugged Dev1 个月前

@threejs We can make it even better further 🔥

Sundeep 的头像
Sundeep1 个月前

@threejs This is unreal amazing! Are you using Cursor and then accessing GPT 5.6 Sol...or is it a different stack?

The Bugged Dev 的头像
The Bugged Dev1 个月前

@threejs I did all of this in Codex using GPT 5.6 sol :)

Sundeep 的头像
Sundeep1 个月前

@threejs Solid!

Sachin Amrev 的头像
Sachin Amrev1 个月前

@threejs Wow cool sir very interesting ❤️

The Bugged Dev 的头像
The Bugged Dev1 个月前

@threejs Thank you so much 😃

相关视频

Probably I vibe coded a lil startup here? 😭 It has been such a loooong wish of mine to build some kind of 3D experience where I could customize a T-shirt in literally any way possible and it’s finally here. Built with Three.js using GPT 6 Astra, this is a full 3D T-shirt customization studio where you can visualize and customize a realistic shirt directly in the browser. That T-shirt itself was modeled by Astra using Tripo right inside Codex through Tripo Plugin. And you can pretty much do anything with it. You can paint directly on any side of the shirt using different brushes and colors, or even spray paint it in real time just like you would spray on a wall. There are also stickers generated using GPT Image 2.5 that you can place anywhere on the T-shirt, resize, reposition, recolor, layer, and customize however you want. On top of that, there’s support for things like fabric customization, sizing, layers, colors, and even wind simulation to push the realism a little further. Once you’re done, you can export the entire design as a 3D view or export individual images of the T-shirt so you could technically take the design, print it, and maybe even sell it. One of my favorite parts is how the spray painting effect works directly on the 3D T-shirt in real time. I also loved seeing how Astra managed to keep the whole experience performant across devices using a custom BVH implementation along with several CPU side optimizations. And the process of building it was super simple. I generated the initial studio design using GPT Image 2.5, gave that image to Astra with the Tripo Plugin enabled, and it basically handled everything from there. I didn’t have to separately generate a 3D reference, upload it to Tripo, download the asset, give it back to Astra, or manually coordinate any of that. Astra handled the entire flow on its own without needing any additional input from me. Really happy with both the process and how the final result turned out. Live:

The Bugged Dev

50,862 次观看 • 6 天前

How many of you actually know how "what you eat" grows? 🤔 I never learned crop science in school. I learned it by walking around our farm with my father, who showed me different crops and explained how they grow. But not everyone has access to a farm, so I built a 3D Crop Science application in Three.js using Claude Opus 5 to make that learning experience more visual and interactive. It is an exploration of different types of seeds where each seed has three different stages. The first stage shows the whole seed. The second stage is a cross section that shows what is inside the seed and how it grows. The third stage shows the germination phase through a 3D model. There are 24 models in total. The original models were 1,349.6 MB, which Claude optimized down to just 11.6 MB in total. All of the 3D models were generated using Tripo. Claude also created visual information for each seed showing details like seed type, storage tissue and germination phase. Each seed also has a YouTube video that shows how it actually grows in the real world. There is also a section showing where each seed is commonly grown using both a 3D globe and a flat map. You can explore different regions and click on specific areas to see more information about where that seed is grown. The entire experience is fully responsive and also includes quizzes that you can take in both normal and 3D formats. I strongly believe education becomes much more effective when you can see, interact, watch and explore instead of relying only on textbooks. Live: Code:

The Bugged Dev

45,081 次观看 • 26 天前

I started using Blender through MCP about two weeks ago, and I quickly realized that you can build almost anything with AI. This model was created using Blender, Hunyuan3D, Gemini, and ChatGPT. Here’s how I did it: I opened Gemini, uploaded an image of the Gundam model, and asked it to generate a clean front-view image. I uploaded that front view to ChatGPT and asked it to generate two additional angles: a back view and a 45-degree front-left view. I went to: I signed up with my email and translated the page into English. Then I opened Image to 3D and selected the multi-image option. You’ll see a diagram of a whale from several angles. Upload each reference image in its corresponding position, such as front, 45-degree front-left, and back. I selected the 1.5M-face option. This produces a very high-poly model, but don’t worry, we’ll fix that next. Once the generation is complete, download the model as a GLB file. From the Hunyuan homepage, open 3D Studio using one of the dropdown menus. Select the topology or retopology tool and upload your GLB. I chose the High setting to preserve as much detail as possible. After a few seconds, the model was retopologized. It kept most of its visual detail while using far fewer polygons. The original head and rifle didn’t look very good, so I generated them separately. I returned to ChatGPT and created dedicated reference images for the Gundam’s head and rifle. I generated each part individually in Hunyuan at the 1.5M setting, then ran both through the same retopology process. Next came the textures. Open the texture section, select the multi-image option, and upload the same reference images according to the whale orientation indicators. However, instead of using the original clean textures, I asked ChatGPT to recreate them with wear, rust stains, scratches, and other surface damage. This gave the Gundam a much older and more authentic appearance. Once every part was textured, I imported everything into Blender. You can ask Codex through MCP to remove the original head and rifle, or you can do it manually. Select the main model and press Tab to enter Edit Mode. Press 3 to enable face selection, then press C to activate Circle Select. Paint over the faces belonging to the original helmet or rifle. You can press X to delete those faces or P to separate them into another object. Then position the newly generated, more detailed head and rifle in their place. I demonstrate this process in one of my older videos. And that’s it. You now have a very cool 3D Gundam model! Afterward, I created the cockpit, separated the model into movable sections, and rigged everything for use in my Three.js game. That process deserves its own tutorial, though. If anyone wants to see it, let me know. Or just ask your AI, I guess. They seem to know everything these days. xD

Spectro

80,433 次观看 • 1 个月前

I just compared Claude Code vs Codex vs Cursor CLI The task was to build a Next.js app with Tailwind 4 and shadcn components to collect customer feedback and showcase it with a widget. I gave all three the same prompt and let them go for 30 minutes to see what they came up with. Claude Code with Opus 4.1 Even though I told it to set up the app in the existing project folder, it tried to create a directory for it. After I interrupted and told it not to do that, it built a demo form and landing page with no errors. I had to ask it to make the demo interactive so users could submit a testimonial and preview it. The landing page looked like AI and was pretty basic, but it worked and it was done in a fraction of the time of the others. Total tokens used: 33k Codex with GPT-5 At the end of the 30 minutes I just could not get Codex to produce a working app. It got stuck in a loop of not being able to set up Tailwind 4 and despite many, MANY, attempts, I ended up with a "failed to compile" error. Total tokens used: 102k Cursor Agent with GPT-5 This was the slowest agent by far and a couple of times I actually thought it got stuck in a loop and was close to Ctrl+C'ing to cancel it. The TUI is really nice though, especially how it shows diffs and it did eventually build a working app (after one or two slight errors that needed fixing) The demo was interactive and it had a very minimal design that looked bare but also a lot less like an "AI generated" app than the Opus 4.1 design. It also wasn't too chatty and just did what it needed to do! Code quality was on a par with Opus 4.1, but it did use 5.5x as many tokens to get there. Still cheaper than Opus on a direct comparison but not when you factor in a Claude Code Max subscription. Total tokens: 188k I'll be able to do a proper comparison and record some videos when I'm back from holiday but for now, Opus is still the more capable model out of the box and Claude Code is the more complete CLI product. It will be interesting to see how Cursor evolve their CLI though with commands and subagents because I think with GPT-5 they have a real shot at providing competition for Claude Code if they can optimise output to get similar quality with less tokens. Jump to 0:40 in the video to see the two apps. Which do you think is which? ;)

Ian Nuttall

195,173 次观看 • 1 年前

I got this custom scorpion rigged, animated and running in my game in one full day with AI. GPT-6 built the entire rig. My part was preparing the model and motion references, then giving feedback. After trying a few approaches, I found a workflow that worked: 1. Give it a model that makes sense. I generated the model with Tripo P2.0, then cleaned it up myself. Low-poly, optimized, split into logical parts. I think that preparation mattered a lot. GPT-6 works on the mesh through code. A manageable vertex count and sensible parts give it a clearer starting point for bone placement and skinning. 2. Show it the movement you want. I took a reference image of the character and used Seedance to generate side-view animation videos. I decided which animations the game needed and selected the takes with the movements I liked. Then I gave GPT-6 the model, the videos and that animation list. There was no existing rig to start from. 3. Compare the animation frame by frame. This was the idea I wanted to test: ask GPT-6 to compare its Blender animation against the reference video frame by frame, from the same view. That gave it concrete poses and timing to work toward. I could point to where the feet should be planted or when the tail should strike. 4. Expect to give feedback. A generated run can look like hopping. An attack can start with a jump you never wanted. Choosing the references matters, and a side view still leaves depth and hidden legs to figure out. It took several iterations. I explained what looked wrong and how I wanted it to move, and GPT-6 kept refining the animation. Once the walk worked, I also had it create left and right strafe animations procedurally from that walk. Those didn't need their own reference videos. By the end of the day, the creature was in Unity. I've recorded the full workflow, including the Blender setup and feedback process:

Stefan 3D AI

12,141 次观看 • 4 天前