5 articles

Hugging Face lists over 90,000 text-to-image models alone. For architecture and 3D, the list that actually matters is six — and the trick is knowing that 'a render' and 'a 3D model' are two different problems needing two different tools.
Read article →
I ran ChatGPT and Gemini's Nano Banana Pro side by side through a real design pipeline — same images, same prompts. The winner isn't the one that makes the prettier picture. It's the one that does what you asked.
Read article →
The honest answer isn't yes or no — it's a question about which half of the job you think the job actually is. AI is coming for the drudgery. Whether that's a threat depends entirely on you.
Read article →
ArchiCAD has no AI button. So how are architects using AI with it every day? The answer is a workflow, not a feature — and once you see the shape of it, it's hard to unsee.
Read article →
Text-to-image AI has one maddening flaw for architects: you can't tell it where to put the walls. ControlNet fixes exactly that — and the trick behind it is more elegant than you'd expect.
Read article →