Midjourney Architectural Design Workflow

Link to Discounted Midjourney Prompt Digital PDF Booklets (For a limited time only):
www.coorlasarch.com/products
This video looks at an emergent workflow using Midjourney to assist in concept design and influence fabrication drawings. We take a brief look at how Midjourney Version 4 (--v 4) uses reference images to produce advanced drawings that further influence our final product.
The 3D pavilion model shown in the video was produced by hand. We are currently looking into AI programs that generate 3D models from 2D images. There are several companies working on this technology.
The 3D Printed Pavilion project is currently in development with our local municipality as we secure funding and further develop planning and logistics to bring this to life. We are currently printing scaled models of our prototype pavilion with Coorlas Architecture PLLC.
Export STL:
help.sketchup.com/en/sketchup...
00:00 - Introduction
01:00 - Concept Design (Midjourney + 3D Model)
01:40 - Emergent Workflow Theory
02:30 - Midjourney V4
04:30 - 3D Print Model (Export STL)
06:50 - Architectural Drawings (Midjourney V4)
07:50 - Workflow (Concept Design to 3D Printing)
09:30 - Conclusion
#midjourney #architecture #artificialintelligence

[@josebranco5727]

Fantastic. No words to describe the potential. Thank you very much for the inspiration.

[@StephenCoorlas]

Glad you found this useful 🤘🏼 It’s looking like these technologies are going to have a significant impact on the way Architecture is taught and eventually practiced moving forward, so it’s definitely worth looking into!

[@fabianfarbeyond558]

Love it! The top right seemed very interesting to me. The Symmetry was very appealing.

[@StephenCoorlas]

It made some really interesting options. You need to really evaluate why you select one over the other 🤘🏼

[@jonsonnewton6402]

Starting to use midjourney in landscape architectural design work flow, thanks for the tutorial video, I have already seen much more possibilities on AI-assist idea generate. Hope this will change the industry and make our living district a better place.

[@StephenCoorlas]

I am hopeful that these technologies will help streamline our workflows. Thanks for watching!

[@flickwtchr]

As a photographer who has photographed many interiors/architecture, this is exciting to imagine what we will be photographing in the future with these emergent technologies bringing such emergent processes into being. Thanks for sharing your emerging processes.

[@StephenCoorlas]

Thanks for your interest and excitement about what these new technologies can offer!

[@ThomStanton]

As a past RE photo/video guy I wholly agree. Plus as an architectural designer I'm seeking to streamline creation of concept art and drafting of documentation. I'm psyched to see where AI aids in taking practical needs for illustrating a specific building type and enhancing time consuming media output.

[@StephenCoorlas]

@@ThomStanton Yes! Exactly, well said. To imagine how quickly iterations and final production content will be created is mind-boggling. Also very excited to see where things go.

[@_soulcraft_]

You'll probably emerge into photographing emergent structures "designed" using ai generated images that were created using the emergent LAION database full of 6 billion unlawfully stolen real photos and designs including your own (the foundation of these emerging technologies) as well as the work of other real photographers and architects used to train and render emergent processes like this. It will be quite an emergent future that will probably mostly create emerging anger in artists and designers who actually developed their own style long before emergent ai systems were "trained" using their hard work. It will be so emergent.

[@nasserahmad4653]

Great tutorial,thank you for sharing, it’s very interesting

[@StephenCoorlas]

Glad you found it interesting 🤘🏼

[@infographie]

Excellent

[@StephenCoorlas]

Thank you! 🤘

[@noemihenrytorres9149]

i TOTALLY AGREE WITH THE VIEW OF NEW CONSTRUCTION PROCESS YOU ARE TALKING ABOUT. I love organic architecture and sustainability, bioconstruction and healthy spaces. Of course the help of high tecnologies!, soooo, lets see if someday we can colaborate.. ;)

[@StephenCoorlas]

Hopefully we can help change the world just a little by continuing to push these ideas forward!

[@johnhanrahan7990]

This is Awsum, like the application becomes even more interesting when you integrate regulations and parameters of building systems into the BIM like model testing in real-time with defined requirements like brief to construction. Makes cradle-to-grave design becomes when it becomes integrated to Revit to generate models from images or text. Very exciting prompts for generations give so much room to design. Microsoft project AI integration for optimized workflow in the office and site though I will say error correction is something that is easy to identify seems to be part of this design approach. Ai still experiences basic math issues which I find quite strange seeing as they are written in code, Is this bad coding or an approach of prediction vs calculation that generates errors. Still so cool

[@StephenCoorlas]

These are good questions, and I think a lot of people are asking similar questions. The AI systems will get better, meaning more accurate and easier to work with, so I think it's important to remember that these technologies are still in their adolescent years of development. Thanks for tuning in!

[@askgato]

cool

[@StephenCoorlas]

Thanks 🤘

[@SebAnt]

I took a course at the end of Nov by Tim Fu, where he showed how to rationalize a wooden pavilion from Midjourney via Grasshopper…

[@StephenCoorlas]

I have seen similar workflows with Stable Diffusion embedded in rhino using grasshopper. Definitely some interesting and exciting progress being made in this area.

[@phillippond3459]

Hi Stephen, A couple of your videos really kicked off some productive midjourney explorations for me and my team. Glad I finally connected with someone who is explaining what I need. Thanks! I have a specific question and hope you don't mind I ask it here. When working with a reference image, or multiple (as with your white 1 point perspective staircase, separate video) how do you lock the perspective of the image? I couldn't see a keyword in your prompts that indicated "use the same point of view as ref image x" but it seems to work for you. My reference image results seem out of nowhere with all new camera angles.

[@StephenCoorlas]

Hey Phillip - Glad to connect and happy to hear these videos were helpful. I the case of that stair image, the results were just coincidentally aligned with the same perspective. If the reference images have a strong relationship in terms of perspective the results typically will follow. But to ensure that the results are from a similar perspective you can add exactly what you typed , "one point perspective staircase" and maybe add some additional qualifiers such as "centered" or "focused"

[@studiokite6403]

If anyone Needs a structure like this 3d printed CADzilla is doing a lot of this work using Recycled Plastic and combining it with steel and concrete

[@StephenCoorlas]

Great work - your Jindi house if very impressive!

[@pedrorolosilva]

This is great! I'm really interesting in 3D printing construction. Let me know if you want to start working on this

[@StephenCoorlas]

Hey Pedro - I'll reach out to discuss more. We have some key players in place already but I'm always curious to discuss with interested people 🤘

[@CryptobdSchool]

I used so many AI tools. My favourite text to image generator is Blue Willow. I created amazing images with Blue Willow and it's super cool

[@StephenCoorlas]

BlueWillow is a great alternative

[@yichiban]

Hi, thanks for the video. Super interesting. Can you clarify what is the exact purpose of generating these drawings? They are not drawings you can build off of right?

[@StephenCoorlas]

No, you cannot build off these drawings, yet. The idea here is that we will not need traditional drawings to "build" from in the future. The hope is for image > 3D > digital fabrication will be a completely automated process. And that technical documentation will be replaced by digital fabrication instructions that are for reference only. This video is highly conceptual, and based on that research.

[@vynahidnyky]

Thank you for this video and for being an inspiration. Please tell us, do you plan to potentially realize these projects with the help of 3D printing? How strong are the materials in 3D printers to realize such projects? Where can we learn more about this?

[@StephenCoorlas]

Yes, there are many projects undergoing fabrication using concrete or mud/clay 3D printing. Research Icon, Black Buffalo, COBOD, Peri, or my personal favorite WASP. I have been in discussion with many of these companies, but need funding to move our project forward.

[@vynahidnyky]

​@@StephenCoorlas Thank you for your response. It is valuable. I will look into these companies. Also, have you considered using fiberglass to create 3D elements for your projects? Fiberglass is strong, reliable, doesn't rot, can be stacked, and doesn't corrode.

[@StephenCoorlas]

@@vynahidnyky Fiberglass it a great option too. The fabrication or molding process is a bit different, but you can get great durable parts from that method

[@mahmoudmoughazy8279]

Very useful content, i appreciate it so much . but id like to ask, how could you use a single image link in v4 algorithm ? it shows up an error each time i try to . is it maybe because i have a free account ?

[@StephenCoorlas]

Yes - I believe it only works with a paid subscription. Glad this content is useful for you!

[@adityagupta7288]

hey stephen, can you please share the prompts you had given for those inspirational iterations with all the lighting, the grass, the scale and accurate proportions between humans and the model

[@StephenCoorlas]

A photorealistic rendering of an architectural open air pavilion, double barrel vault, arches, terracotta, geometric, illuminated, during a bright foggy sunrise in a grass field --ar 16:9 That should be it

[@twitterglobalarmy]

Midjourney is cool but users must pay to utilize the site after a brief period of free use. I've also experimented with Blue Willow and I'm really amazed to use it

[@StephenCoorlas]

Yes - BlueWillow is a great free alternative

[@pixiedigitalninja8260]

Amazing work, very inspiring. What is the name of your architecture firm? Very interested in checking you out and learning more about your work

[@StephenCoorlas]

Thanks for your interest! Our firm is called Coorlas Architecture. The work shown here on this channel is through our research and development program within the firm. Our primary work focuses on using advanced technology to efficiently design and document residential and commercial projects 🤘

[@pixiedigitalninja8260]

@@StephenCoorlas Thanks so much for replying! Great, I'll definitely be googling Coorlas Architecture to learn more about your outstanding work 👌🏼

[@daynelawrie499]

Really love the content, we build a lot of architects designs, inwas wondering if you have been able to take a screen shot from a 3D model or midjourney create image and create a simple line drawing in midjourney? I guess its reverse engineering to a traditional method.

[@StephenCoorlas]

That's what we've been experimenting with, but currently it doesn't seem like there's enough applicable data for Midjourney to drawn from in producing 2D plans. We're also a bit more interested in seeing if it can produce instructions on how to assembly instead of simply 2D plans. 🤘

[@daynelawrie499]

@@StephenCoorlas it's so interesting, have you found that it can produce a hand line drawing of an elevation that could be used for more marketing purposes rather then construction?

[@StephenCoorlas]

@@daynelawrie499 It goes a little rogue with the actual dimensions of the image you input. I think right now version 4 is better for making variations on more natural things like portraits of humans or nature where the features can vary slightly and still have a close resemblance to the original.

[@mlee9049]

Hi, I have a sinilar idea to your emergent workflow. Do you mind sharing the prompt that gives you the best ai variations?

[@StephenCoorlas]

I have a video where I share 16 Prompts in the video description: kzbin.info/www/bejne/qaexnaiEndSXn8k

[@OsosArquitectosbyJoseZenteno]

Hello bro from perú, i have a few question before buying midjourney. I have a small theory can we use midjourney as a motor render? Example we can put a enscape or vray premade render and then in the prompt put "realistic render etc etc"? It will work? (But i dont want to change de desing Just for "render it"

[@StephenCoorlas]

Hey, no that will not work with Midjourney yet. There are some other programs working on that such as FireFly or Stable Diffusion using ControlNet. But Midjourney will not render a shaded view from a 3D program without significantly changing the architecture

[@OsosArquitectosbyJoseZenteno]

@@StephenCoorlas thanks for the answer, the other programs should work? I Will try

[@forfran]

is there any way of importing an image to midjourney an make the ai work exactly on what you inserted ?? for example.. y have a house render made with twinmotion, now i want it to me even more real. so can midjourney work on the model and modify the enviroment in some way ??

[@StephenCoorlas]

I haven't seen it regenerate exact geometry/pixels yet - In most cases it gives you something similar, but alters the design too much to be reused for practical workflow processes. The closest thing I've seen is people using Dreambooth with Stable Diffusion to train models based on a limited data sets of reference images. That process is a bit heavy on GPU usage, so we've been searching for methods and workflows that are less demanding.

[@liquidmind]

depends on the prompt... or, a hint - dont use prompt, but instead, use 2 or MORE images OF THE SAME thing, barely different angles OR, same clone photos 2 times if you want to be STRICTLY that photo..... MAKE IT OBEY YOU..... ;)

[@forfran]

@@liquidmind ty !!

[@StephenCoorlas]

@@liquidmind nice lol I’ll need to give that approach a try

[@liquidmind]

@@StephenCoorlas i love your videos since july august. thanks for them prompts.

[@nathan1459]

Where are you planning this pavillion? Did you push it towards a Serengeti vibe? I'd be genuinely interested to know where the location came from.

[@StephenCoorlas]

It will be located in the north suburbs of Chicago. We are developing a general plan/design that may be adapted to any of the townships here, but it is significantly dependent on funding. I will have some interesting updates on this pavilion to share shortly.

[@nathan1459]

@@StephenCoorlas Cool, I'll make sure to check back. Do you see earth building in this way being adapted into general building practice?

[@StephenCoorlas]

@@nathan1459 I really hope so - At some capacity it will be adopting, but it may not become "mainstream" for some time. It will be supported by groups and communities where the value of building with earth are cherished and respected. Slowly but surely it will make its way into general building practice.

[@IbrahimKayanja-wr3we]

Do they build grass thatched houses in the us???

[@StephenCoorlas]

Not many, but I'd like to change that.

[@mahmoudramdane2848]

The sections made in Midjourney v4 are correct?

[@StephenCoorlas]

Not currently, no. Although, they are getting closer. I don't believe the AI is cross-trained to develop a 3 Dimensional understanding of the 2 Dimensional images they produce. Perhaps in future versions this will be the case and you can instruct it to slice plan and section views.

[@mrbigolnuts3041]

Can you just render your basic image into a setting without MidJourney changing the design of the architecture?

[@StephenCoorlas]

Unfortunately, no. Not yet at least. There are plenty of other programs that will do that though. Try Veras - they offer a great standalone ControlNet based platform. PromeAI also offers some free trials for this task. You can check out my “Sketch to Render” video that features some of those tools.

[@puresoul9294]

I'm learning rhino right now shall i stop and start learning midjourney?

[@StephenCoorlas]

I won't stop learning Rhino or any other software for that matter. It will take time for these AI tools to truly integrate and supplement existing 3D modeling programs, so it is still very important to learn how to model.

[@user-tb3xc4bn3k]

hello very interesting your work at 8:43 on the video there is a crazy design I wanted to translate into an amphitheater with your permission or buy a concept

[@user-tb3xc4bn3k]

about 15 meters in diameter

[@StephenCoorlas]

Hi you can contact us at Coorlas Architecture to engage our architectural design services. Thank you!

[@learqui]

Please share SketchUp style you use

[@StephenCoorlas]

I'm not sure what you mean by "style" - My background and color settings?

[@learqui]

@@StephenCoorlas yes

[@StephenCoorlas]

@@learqui It's just a black background with black fog, and "back edges" turned on to get that appearance.

[@luongvu7007]

SHOULD I STOP MY CURRENT WORKFLOW 😂

[@StephenCoorlas]

haha, no! It's still important to understand how to model, draft and develop concepts without AI

[@fafaOTR]

So far, all these AI image generating algorithms are just like architectural interns in a design office who only can make still drawings, but don't know how to use any 3d modeling software.

[@StephenCoorlas]

There do seem to be a fair amount of young architects and designers exploring this technology. Although there are also some more experienced architects researching how to develop a more direct relationship between the image content and actual 3D modeled geometry. That will be where this technology truly begins to have a significant impact on our workflows.

[@quangduongsong373]

funny how the artist community is butthurting right now about AI, while we, the architects just dont really care that much :Đ

[@StephenCoorlas]

Architecture is like a team sport, so the responsibility and credit gets diffused amongst many throughout the process (at least on large projects). Most artists operate solo, so I can understand their frustration when their credit or identity is being borrowed, shared and extrapolated upon so quickly and hastily in many cases. AI isn't going anywhere. Perhaps more restrictions will be put-in-place, but my advice to artists is to embrace it. Maybe it's the "push" some artists need to advance their style into new territories that data sets don't have access to yet. And to be careful what you share online moving forward! 🤘

[@jetvlz8592]

AI is in need of artists' artworks though. That is also funny.

[@quangduongsong373]

@@jetvlz8592 nope, we also learn by watching others do the work first, AI just do it 1000x faster than you.

[@jetvlz8592]

@@StephenCoorlas Disposable digital art will be gone, that's for sure.

[@StephenCoorlas]

@@jetvlz8592 Absolutely, I tend to agree. It’s an infatuation phase without much meaning or purpose.

[@DKlavdious]

I totally understand AI as a source of inspiration and iterations engine but other than that it's very abstract and random.... poor engineers that will probably have to deal with all that... anyways. The "drawings" as presentation layouts were quite nice, but as far as I can tell the views included were absolutely random... I might sound a bit bitter but all that philosophical approach of architecture is really getting on my nerves... or i'm just to shallow to grasp...

[@StephenCoorlas]

I agree with you. These tools are not very reliable in terms of documentation or even producing consistent designs from multiple views... yet. But that's why I research these technologies - to help nurture the integration of the tools in a responsible and reliable way with our current architectural workflows. The philosophy is also just a personal approach I use to help better understand the implications of using these tools, and in effect how they may change our profession or even society. Thanks for sharing your thoughts 🤘

[@boulderbite]

I think it is a good generator of concepts that could inspire, but right now it is very infantile in producing anythign that is usable. Having a pretty picture doesn't translate into actual 3d models on its own. Still lots of investigation to be done.

[@StephenCoorlas]

Have you tried ZoeDepth on the Hugging Face website? It offers AI generated 3D models from image inputs. Yes, still much investigation to be done, but it seems promising.