How MVRDV is using AI to design their buildings

okay so look at this project done by mvrb this is  an image of the base 3D model and these are all   variations that were generated using AI each one  slightly different than the other some influence   materiality others change spatial design ideas  and others just certain parts of the design now   mvrdv is probably one of the most interesting  architecture offices in the star spectrum and   this is because since the 90s they've been not  only designing projects that are different and   to be honest kind of crazy but because they're  also constantly investigating and researching on   the side from projects centered on roofs to City  Planning and even Furniture they're committed to   a research based design practice mdv is a firm  that does design through research not by testing   a million things or by making a nice shape it's  always by uh creating a hypothesis testing it   figuring out if we agree with it do we like it do  we not and we through design research then gets to   a final conclusion this is KAS esbach an architect  and project leader at mgtb KAS has been involved   in numerous projects and is constantly emphasizing  how the office's success is centered around one   main ingredient embracing change in Innovation 's  tradition is to embrace uh this growth and embrace   this ever change so it's the our tradition is  to change and to accept these things and to   go full force into them but with AI entering the  architectural realm how will this tool reshape the   way mbrv approaches design will it enhance their  creative process or make it dull and boring this   is a story of how one of the most avanguard  architecture firms in the world steps into   the future with AI as their companion in design  okay so now architecture is always changing and   evolving and the tools which create our future  buildings also change and evolve so when a new   technology like AI comes along and Promises to be  a super tool and enhance the architecture design   it's a no-brainer to accept it and use it to push  architecture even further but one of the first   questions that arises with a new tool is almost  always why should we choose a new tool when our   current one works perfectly if you can't do it in  the traditional way you shouldn't do it through AI   so so everything can be done in the traditional  way but AI simply increases the speed of it and   makes it more efficient and faster this is Freddy  Fork he is an architect with a masters in Building   Technology from TU Del and AI specialist inside  of mvrdv so more or less he has been at the center   of this transformation of architecture and AI so  I I really enjoyed that part where they push the   boundaries and this is what what I'm doing here  trying to push the boundaries of applying AI to   architecture here at a VDB AI is transforming  the way we think about architecture design   it's not just about the automation of task it's  about enhancing our creative process AI allows   us to experiment at a pace and scale that was  previously unimaginable the AI is is not a magic   wand but it's a power tool it just simply helps  us uh get results much faster and each new tool   throughout the years has brought something unique  to the table enhancing our ability to visualize   and realize our designs but with AI we're not just  looking at another logical step in this Evolution   we're witnessing a leap a long time ago back in  AutoCAD I don't know how long it took you to make   a photo realistic render and how much money then  slowly when fate came in okay you can do a really   nice render in about 3 days now with ncape you can  do a nice render in one day and now with M you can   do a nice render in maybe 30 minutes now clearly  for mvrdv integrating AI into their design process   was a step into Uncharted Territory in order  to fit a new tool into an extremely intricate   architecture process they would have to start the  same way they start any of their Designs By Design   through research I think about July of last year  I mean the first Whispers of M AI or mid Journey   version 3 was kind of going through the office  and people like what what is that what's going   on and then there was a lunch meeting by what  by a few of the Nerds in the office where we all   sat together we're like okay what are you doing  with your M what have you done with it and we're   showing each we're sharing these images and then  what happens is since MTV is a studi based office   so we have nine different Studios within one MTV  title uh we started sharing uh this knowledge so   slowly the from the Nerds it kind of trickled  down into the whole office until finally many   people were uh challenged to work with MIT nbb's  initial dive into AI wasn't a structured plan but   a series of organic curiosity driven Explorations  it was about shaping Knowledge Learning from each   other and collectively envisioning how AI could  fit into their work this group of people we   brainstormed a few Studio sessions in which we  then uh grabbed smaller teams of maybe 20 people   and then presented them how mid Journey works how  stable diffusion works and then uh give them a   design challenge so that's maybe an hour in total  30 minutes we give a little lecture 15 minutes we   give them to play around with themselves and then  they have to present their thing so I remember in   version three of mid Journey we were asking people  to generate floor plans sections elevations and   renders of one fake project which we made up  and then ask them to present it to each other   and then finally we had the vot on which had the  most convincing design was really great s to see   so as mvrv delved deeper into the world of AI they  began to uncover its potential and its limitations   this phase of exploration was crucial it set the  foundation for how AI would be integrated into   their future projects but integrating a technology  as advanced and as nuanced as AI into the refined   and intricate process of architecture design was  not a straightforward path it involved a steep   learning curve to learn new software technical  hurdles and the task of aligning ai's power tool   capabilities with the firm's Creative Vision what  was extremely interesting for me was actually   typing mvdz what does it mean if you ask somebody  with different type of bias AI has integrated into   it what comes out well colorful pixels was the  answer and then I thought to himself that's not   what we're doing that's not what we're about  at all of course another big problem was the   rapid Pace at which AI technology was evolving  I mean just look mid Journey launched in July of   2022 stable diffusion in August and Chad gbt in  November of that same year and some results have   gone from looking like this to now looking like  this AI is growing so fast like you're out for   for a month you're working on something else then  you come back and you're you're like oh wait what   did I miss I used to be able to type this in and  it used to work this struggle was not just about   technical adaptation it was about a philosophical  adaptation as well I mean how does a firm known   for its meticulous research driven approach  Embrace a techn technology that is inherently   experimental and unpredictable then you also see  the limits of ai ai couldn't just knock out uh   mvv design uh because it's missing all of that  other stuff that we're talking about for the the   research the cultural context the locals cost of  programmatic requirements and it was through this   exploration of trial and error that the office  started using mid journey and stable diffusion and   testing to see how the initial stages of a project  could be enhanced with AI tools if the normal   process would look something like this where you  can talk to the client sketch some ideas out pass   it onto a cat software and start refining it from  there AI would start enhancing this process and   playing a specific role in different parts many  times we use it in the first stages we sit down   we talk to the clients we go back home we're  like okay what what did they want and we just   throw a few things in there see what comes out  and maybe something that triggers us but then as   we go through the process these things get refined  and Def find more and we have to deal with things   like the widths of doors and stuff you know which  Min Journey doesn't take into account for stable   Fusion so we start to develop those and then we  reach a point in the design process where we have   let's say a blue foam model which then we try to  Reiner into this mid Journey or stable defusion   where we overlay different types of facades  and kind of without too much effort create a   atmospheric image which relays well to the design  which adheres to all of these architectural rules   the floors the doors the walls the roof so through  this constant testing and research they found that   mid journey in stable diffusion could not only be  useful in the initial stages of the design like   brainstorming and coming up with crazy ideas but  also in the more technical processes and then what   is really fun is later in the process it starts  to play a role again the moment that everything   is defined but we have to test uh you know 20  different types of window opening sizes and you   just like rock it in see what happens sometimes  horrible things come out as you're probably see   now on screen or sometimes beautiful things come  out which excite us and we're like okay now let's   develop that further this new look into using AI  to design was also having an effect on the teams   that worked in each project so if before AI a team  of 10 people use their time and effort to create   repetitive tasks like test window openings  or facade materials now they could dedicate   their time to more creative parts of the design  process that are more akin to what an architect   is supposed to do and not what a robot or AI can  do instead of having to lose people super talented   people in the productional workflow we're able  to use their brain power in a more uh beneficial   way for the process so actually efficiency just  keeps increasing as we start to implement these   tools but of course with every new technology  especially one as impactful as AI comes a set   of responsibilities and ethical considerations  as many of you know many current AI models use   information from all over with the internet  to generate images so regardless of copyright   ownership of credit anyone can create an original  image and the ethical boundaries of creativity   are ignored or blurred lots of these models now  have been trained using web scraping throughout   the whole internet and it takes absolutely all  images whether they're copyright or not so I do   think that the proper way of advancing the future  is knowing where the data and the data sets are   coming from and and being able to use models that  that use these uh these data sets at mvrdv we are   using only images that are property of mvrdv to  train the lower models that generate the design   process now the advantage of creating their own  models is that it becomes a sort of personalized   AI tool that can understand the office's  perspective Mission and principles which is   something we will see much more of in the future  I think where each architecture office has a group   of really creative employees but also has a set  of distinguished AI language models that will   separate them from the others so the the one thing  I'm really worried about is that it's going to be   kind of a cheap trick that people apply it doesn't  take too much effort to just type in great design   beautiful atmospheric light vegetation and then  applied on to just some random shape so if people   start misusing and abusing then we're we're going  to start getting posts on Instagram where it's   like render versus reality and you just see these  really beautiful over-the-top jungle Interiors or   jungle buildings floating with one little column  in the middle and then the reality is completely   the opposite but then when I'm hearing myself say  this I think to myself you know what this probably   won't happen I only see a positive influence for  the world so as mvrdv's Journey with AI progressed   they reached a pivotal point of integration this  was the moment when AI ceased to be an external   novel tool and became an intrinsic part of their  design process we have three main branches that we   started to use rank sty customization research and  AI renderings talking about M Journey everyone's   been using it and using more for this typ of  brainstorming at the initial stages where you   get interesting images then there is the prom  styles in stable diffusion and and and there we   really start typing our own specific PRS that we  start to use more often and give us more results   that are closer to what we want also tools like  stable diffusion allow you to grab your existing   handmade models and start testing facade options  or volumetric changes we were able to use these   bluef models we were with a very small team apply  a really quick facade on them and throw them into   mid Journey with just a few key words we able  to knock out beautiful images which really   excited the client to a point that we were able to  continue with the project The Next Step would be   embedding it's also trying to to teach the model  what type of graphical uh Styles or or concepts   that uh that we would like to get from the model  and and finally the most interesting one is uh   Lowa trading and the most complex one that we  actually teach it our no concept okay so think   of Lura trading like customizing your favorite  video game character you have a character that's   already design that's the pre-trained AI model but  you want to make some changes to make it more you   like adjusting the its hairstyle or outfit but  not completely redesigning it from scratch Laura   training does something similar with AI models  it takes an existing AI model and makes small   specific changes to it this is way more efficient  than creating a new character every time you want   a different look it's like fine-tuning the model  to do exactly what you need like teaching it to   recognize certain architecture Styles or to  create music in a particular genre without   having to rebuild the whole model what we tried  just recently and it worked out pretty well is   teaching stable diffusion how to generate the  valley so the the project the valley that we we   worked we feed some images tell the St diffusion  model this is what the body looks like and when we   use the low it will give us results as of the body  and then the third branch in which mvrb is using   AI is for architectural rendering where they use  stable diffusion along with an additional model   called control net to precisely modify the  renderings add materials and change style so   control net what it does is it freees the geometry  and once you have KRON set up it will follow as as   possible to the The Prompt that you give it but  there will always be surprises it's not a static   result where in in in landscape or in Lan you  really specify this is this material this you   you have to be kind of open to a certain degree  of surprise but that also means that you can get   inspired by it so we worked in a collage based  manner having very quick and dirty collages where   the project leader would share with me the these  collages and then we would with the very low D   noising strength D noising strength is is that the  other creativity that the model has we reduce the   D noising strength but then it makes the collash  keep the general view of what we want to see but   then really improve graphically or change the  style to something more like for example Studio   we or something more cartoony or so that the the  client would would like the the images rather than   doing it the traditional way so what was at the  beginning a tool for exploration that was merely   explored and talked about is slowly becoming a  structural part of the design process in mvdb   and the best part is that the Creative Design  through research process that they are known   for is not being interrupted but augmented and  improved and as a matter of fact they're already   trying to see what a more integrated AI future  would look like for an architectural office but   I am really excited about the next steps what's  going to happen with AI for example imagine an AI   integrated Bim software where you can just type  in to Rev hey all the the walls in the bathrooms   of the more expensive Apartments uh give them  this layer and it just does it so once again   it's all about removing the tedious part of the  work and uh really making it giving people more   time to design and research now in this new normal  AI is redefining the limits of what's possible in   architecture it's not just about generating quick  renders or creating visually stunning designs it's   about using AI to explore new ideas to push the  boundaries of cre creativity and to see projects   from angles previously unimagined I see that  because we start to have this conversational   chat type design approach AI could actually help  us rehumanize the design process we start to   see the images or see our architecture from the  user's perspective and because of that we could   not we not only have to focus on the technical  criteria but we can also see it from the user   point of view because we have the ability now to  to see how would look like almost instantly it it   doesn't mean you can generate your building in one  second and that we lose our jobs it actually just   means that we can do more things and more fun  things in a shorter time span leading to more   exciting designs so I'm 100% Pro AI integration  into architecture as we look into the future   it's clear clear that AI will continue to play a  pivotal role in shaping the world of architecture   driving Innovation efficiency and above all  inspiring new ways to imagine and create the   spaces around us our office's design ethic and  Beal and thought process is so based in this   research approach to design AI merely is just like  a tool that you use to enhance your your process   so I want to thank C and Freddy for helping me  create this video I love seeing how offices like   mvrdv are pushing the boundaries and redefining  what it means to be an architect and what it means   to be Innovative if you would like to know more  about mvv there will be a link to their website   down below and also if you want us to talk more  about Ai and how to use it as Architects let me   know in the comments thank you guys for watching  this video I'll see you in the next [Music] one

Channel: Show It Better

Views: 113,240

Keywords: show it better, architecture tutorials, photoshop for architects, architecture videos, tutoriales de arquitectura, ai architecture, artificial intelligence, midjourney ai tutorial, ai architecture design software, architecture school, ai architecture design, ai for architecture design, best ai for architecture design, architecture student, prome ai

Id: dvKAyTRptkw

Length: 18min 20sec (1100 seconds)

Published: Tue Feb 20 2024