Stable Diffusion: Will It Revolutionize Architecture?

In this tutorial, learn how to download and run Stable Diffusion, an innovative AI tool that generates images from text descriptions. Tailored for architects and designers, this video shows how Stable Diffusion integrates into the architectural workflow for visualization and concept development. Explore its powerful image-to-image capabilities and see how this technology can transform your projects. Additionally, join a broader discussion on the role of AI in the future of architecture.
00:00 Intro - Why Stable Diffusion is interesting in architecture
01:40 Stable Diffusion web browser demo
02:14 System requirements
03:10 Install Git
04:05 Install Python
05:13 Install Stable Diffusion - WebUI
07:06 Install model/checkpoint
10:17 Set up WebUI.bat
11:00 Launch Stable Diffusion
12:02 How SD works & how to operate it - PART 1
25:11 Controversial discussion about AI image generators like Midjourney
27:56 How SD works & how to operate it - PART 2
30:24 A different approach to concept development in architecture
33:49 AI specialists in architecture
35:10 A "ping pong" between specialists as always
37:12 How SD works & how to operate it - PART 3
51:11 Landscape architects and SD
51:57 How SD works & how to operate it - PART 4
57:39 Upscale & manipulate an image in SD
1:02:25 Showcases
1:05:55 Conclusion
Resources:
Try Stable Diffusion in a webbrowser:
huggingface.co/spaces/stabili...
Install Git:
git-scm.com/download/win
Install Python:
www.python.org/downloads/rele...
Stable Diffusion WebUI GitHub link:
github.com/AUTOMATIC1111/stab...
Download model / checkpoint:
huggingface.co/runwayml/stabl...
webui-user.bat edit. Paste the following text at the top and save:
git pull
Models/checkpoints:
civitai.com/
or
stable-diffusion-art.com/models
Related videos:
• DALL-E vs Midjourney: ...

[@AdnanDarwish3]

thank you for a great video keep up the good work

[@ArchViz007]

Thx @Adnan!

[@KILABANANA]

Just name the stuff in negative prompt, you don't have to write "no human", that means that it has to have a human. Also try using a blender model for shape, vertex colors for masks... you can keep the shape that way just change the texture, i'm rambling but you can do a lot with 3d and control nets etc...

[@ArchViz007]

:) yeah I noticed that also after upload! lol. I know you can do a lot with a 3d model also. Thx @KILABANANA