- Architecture Insights
- DALL-E 3 Release & Review (Comparison To Midjourney)
DALL-E 3 Release & Review (Comparison To Midjourney)
Welcome to the newsletter where architects and designers come to stay up to date with the developments of artificial intelligence in architecture.
In this weeks 5 minute read:
AI News for Architects
Parametric-architecture continues to host AI creative challenges that bring architects and designers together to compete by showing off AI generated images, check out the most recent one here. The winner receives $300, a free publication, and more.
Adobe Firefly has improved their own image generating AI with Firefly Image 2. For now it is only available on the web, however it will soon be released into photoshop and other creative cloud apps. Check it out here.
Europe continues to impose strict rules and regulations on the developments of AI. Asia on the other hand has taken a more “business friendly” approach. Read more about it here.
As of October 2023 we can all access the newest version of DALL-E 3. All you need is a microsoft account.
For architects, this opens a new door of image creation that can be used in the professional setting.
If you have any experience with the older version of DALL-E you might have found it underwhelming, especially when it came to creating high level renderings and perspectives that could be used for presentation.
DALL-E 3 brings us one step closer.
*DISCLAIMER: this is not a promotion for OpenAI nor is this newsletter sponsored by them*. This is simply personal experience and a overview of the features and how they have usefulness to the practice of architecture.
How to access DALL-E 3
Accessing the software is different than just logging on to your OpenAI account as you would to use ChatGPT. In order to access DALL-E 3 you need to visit bing.com/create and login with your microsoft account. As of October 2023, there is a free version you can access.
How good is it?
Check out one of four images DALL-E 3 created for us using this prompt:
“Create a victorian style castle in the forests of the amazon with an overcast setting.”
Was this result accurate to the description of my prompt? For the most part.
Can you tell it was taken in the Amazon? Somewhat.
How is the quality? Amazing.
This example was just to show you how the quality has improved and how creative the AI can get.
If similar results can be obtaining for more site and project specific requests could this be the end of manual image renderings? Only you can answer that for yourself by exploring DALL-E 3 and its capabilities for your specific project work.
Let’s try a more architecturally realistic prompt:
“A bungalow by the beach with a wooden deck leading out to the water.”
From an architectural perspective, this is still a very basic and high level request. Architectural renderings have many more features.
Let’s try one more time and be a little more specific:
“A bungalow by the beach with a glass roof that features a hammock and leads to the beach through a stone pathway with bollard lighting.”
These were the 4 images generated.
Once again they are all high quality, fairly accurate (one of them doesn’t feature the glass roof), and offer unique variations to chose from.
For concept design, we can see this being a great place to start.
While we don’t yet have a software that can convert AI generated images from jpegs or pdf’s to editable photoshop files, DALL-E at least provides a solid foundation for concept development.
When we dive into site/project specific designs we see AI softwares fall behind. It is not yet possible to reference exact locations via google earth and have AI use that as a base point for your images.
Try out DALL-E 3 for yourself since every architect will have it’s own demands of this AI. Hopefully from this however, you can see the improved quality of images, the level of accuracy, and the power of prompts.
Read how OpenAI improved DALL-E 3 from a computing and software perspective, Link.
In the near future, video creation AI will reach the capabilties that image generating AI is at. For a look at software that has started this development check out Moonvalley.ai
Check out adobe’s prompt guide to help with writing better requests and deliver better results when using AI image generators.
AI Image of the week
Incase there wasn’t enough AI generated images for you in this email already, here is one more.
Created with DALL-E 3
Thank you for reading this weeks issue, check past issues here. Sharing this newsletter with colleagues, friends, or anyone interested in the combined world of architecture and artificial intelligence is greatly appreciated.
Until next Friday,