GPT 6 Astra: How to Build 3D Worlds!
AI News Today | Julian Goldie Podcast
September 16, 2026
GPT 6 Astra: How to Build 3D Worlds!
Speakers Julian Goldie
Julian Goldie (0:00)
Today, I want to show you how to use Chapter 6 Astra to create 3D worlds. As you can see right here, you can basically create these amazing 3D worlds where you can navigate 3D landscapes, and you can just sort of walk around. This works in loads of different ways, and I'll show you some different scenarios. I've actually got a world library, as you can see right here. And this is just some cool examples that we've built step by step using the system that I'm about to show you. So Astra 6 is unbelievable for 3D modeling. As you can see, I mean, the world itself is beautiful, and it can also do it all start to finish pretty much by itself with a couple of free MCPs, which I'll walk you through in a second. And you can just give it a image, and from there, it will generate a 3D world that you can actually navigate. So you give it a reference, it creates a scene, and then it actually creates a scene that you can navigate as you can see right here. And you can see some example previews like so. So how is this working? Well, basically, what we have over here is we have Chat GPT, and we're using the MCP directly with Blender. So Blender is really good for creating the 3D worlds and modeling, but super difficult to actually use yourself, super difficult to learn, and most people just give up because it takes like a year to create something really basic, right? And it's very, very boring. But what you can do now with your AI agents is you can use this whole system, get Chat GPT to do it directly for you, and it's a simple three-step process. So essentially, what you can see here is the prompt we use is you said, you're working inside Blender. I want you to design and build a complete photorealistic 3D environment from scratch. The subjects are in the day, the mood are entirely your choice. Surprise me. Pick something that you think would look stunning, not something generic. And then it has a few different steps for building that out. So step number one is that it's actually going to come up with a concept. So GP6 Astra can do that, and also GP6 Astra can generate the image itself. So for example, it's using ChatGPT 5, 2.5 inside there, ChatGPT 2.5 images for image generation, right? And so when it's coming up with a concept, it can come up with some ideas, generate the image for that, and then use that as the architecture, and decide from there, okay, what materials, what light, and etc. is going to use, and it saves the image as a reference inside our project. So that is step number one. From there, we've also got the build rules, right? Now, I said work primarily through Blender using computer use, because it can basically operate Blender for us. So for example, this whole project that we generated with Blender, I didn't touch it, I haven't created it, I haven't used it at all. It's all ChatGPT generating every part of this element, and you can see how complex the builds are. If we pull this down, for example, look at all this stuff that is generated inside our 3D modeling system, and each one of these requires quite a lot of technical expertise just to operate it and understand it. And so you look at that tiny detail over here, that's been generated with Astra 6, and it's done every single step of that, as you can see. There are literally hundreds of tiny details inside this 3D model system. So that was pretty cool. And then from here, we said organize everything into collections, these geometry nodes and instances and et cetera. If you want this full detailed prompt, you can get it inside the AirProfile boardroom, link in the comments description or go to the AirProfile boardroom. But that is step number two. So just to recap, you come up with the concept, you give GPT-6 Astra the build rules, and then you're going to come up with materials and lighting options too. So for example, the way that I would look at this is it's going to compose a scene, it's going to generate the image, and then it's going to add the different elements to that image itself.
Now also, it needs like different camera angles inside the actual setup. So when we're navigating it, we want different camera angles to be able to use. So if we go back to our 3D world system over here, this is the world studio that we've created.
You can see that if we navigate, for example, the conservatory, we can walk through it, we can see different elements. We've got different views, so we can reset the view or we can switch to in fly mode, full screen, and we've got different settings here. So we can actually change the lighting system, which is pretty nice too. And we can save the viewpoint, restore the viewpoint, save it as a photograph and hide the controls as well. And then if we want to edit the blender scene, we've got the details here, as you can see right.
6 more minutes of transcript below
Thousands of transcripts fetched by people building searchable podcast archives
Fetch the whole transcript
The demo key returns a sample episode in full, no card needed:
curl -H "x-api-key: pt_demo" \
https://spoken.md/transcripts/1000651996090Markdown with the speakers named, for your notes, your knowledge base, or anything that makes HTTP calls.
From $0.10 per transcript. No subscription. Credits never expire. Prices exclude VAT, added at checkout for EU customers. Not what you expected? Email us within 14 days with 20 or fewer credits used and we refund the pack in full.
Using your own key:
curl -H "x-api-key: YOUR_KEY" \
https://spoken.md/transcripts/1000790384575