In the recent months, AI visualization tools like Dall-E and Midjourny have been gaining attention from the design and architecture community. These tools use artificial intelligence text-to-image technology to generate images based on prompts entered by the users. In today鈥檚 article, we鈥檒l discuss what this means for the architectural industry, how to use the tools, and which one is better. To celebrate the launch of these tools, we are hosting the world鈥檚 first competition for AI architecture. Send us images generated by AI tools, not limited by Midjourney or Dall-e, for a chance to win $1500! Tell your friends who might also be interested too!
What does this mean to architectural industry?
What does this mean for the architectural industry? Whenever AI comes up in discussions, one of the first questions that comes up is “Will AI take over the jobs?” This question has come up every time there was a technological breakthrough. Some traditional architects considered the art is dead when AutoCAD began to get adopted. Time and time again, it has been proven that technology cannot replace architects. Even with the current state of AI is a far cry from replacing architects. In fact, they are not even able to replace a small part of it. Even though these tools can generate unlimited number of iterations and pump out original designs, at the end of the day, the designers have to make the decision on which design to move forward with. So these tools may not be able to completely replace a task in the industry yet, but they can be used as great design partners. I personally find that they are super effective against writer鈥檚 block during the schematic design phase, and it works especially well when you are drawing motif from various sources.
Two of the most popular and well known tools today are Open AI’s DALL-E and Midjourney. Both of these are text-to-image models to allow users to enter a prompt to visualize. Let’s go through each of them.
Midjourney
Let鈥檚 take a look at how to use both Midjourney and Dall-e. For Midjourney, you can sign up for free on midjourney.com and sign in using discord. Enter one of the newbie channels and type `/images` followed by your prompt. Voila! You have your first AI generated image. Another thing to keep in mind is that you can enter `–v 4` at the end of your prompt to use Midjourney’s V4 model, which seems to provide much more detailed, realistic and consistent result. From here, you can ask Midjourney to generate variations of one (or all) of the images, or increase the resolution for one of them. Keep tweaking your prompt and variations until you find something you鈥檙e satisfied with.
DALL-E
For DALL-E, you can sign up by going to openai.com. Enter a prompt directly into the website and wait to see your 4 results! Similar to Midjourney, you can request additional variations for one of them. One of the cool things about Dall-e is that they provide you with additional tools like background extension, and inpainting to fine tune your images.
Which one is better?
For Midjourney, the pros are that the images seem to follow the prompts closely, and they are also very photogenic. Om addition, there is a community aspect where they feature amazing works created by others on their home page, and you can also see what others are working on while you wait for your images to process. The only con for Midjourney is that you are forced to use Discord, and those who are not familiar with discord might find this confusing.聽
For DALL-E, the pros are that the web interface is easy to use, and they provide a kit of tools such as inpainting and background extension that are really useful. The biggest con for DALL-E is that there is an issue with prompt comprehension, where images don’t follow the prompt very closely and the images seem a little out of context. In terms of the flexibility, we need to give DALL-E some extra points because you can make some edits to the images afterwards.
For pricing, both platforms require you to purchase credits after your trial runs out. DALL-E charges $15 per 115 credits, which come down to about $0.13 per image. For Midjourney, they provide a couple subscription plans. For basic plan, it costs $10/month for 200 images, which comes to $0.05 per image. You can also subscribe for Standard plan for $30/month for unlimited number of images. In terms of pricing, Midjourney definitely takes the crown.
AI Architecture Competition
All in all, they鈥檙e both fantastic tools and blows my mind every time I use them. Both DALL-E and Midjourney have their own style and it鈥檚 up to the user to decide which one works best for them. This is definitely only the beginning of these technologies, and we are excited to see what will come next. To foster such growth in the architecture community, we are giving away $3000 as a part of AI Architecture Competition. Show us your images you created for a chance to win $1500 prize cash! For more information checkout archihacks.com/challenges