Simone Niquille: "Blood Camera"

and um okay so i'm gonna like in previous lecture uh okay so first of all if i can ask everybody to mute themselves so that it doesn't go awesome thank you um you see the negative aspect of being all in the same space [Music] so i'm gonna try like in previous lectures to keep the intro short so we can make the most out of our time together um but um i want to welcome everyone to our uh fifth and last event or a global lecture series here in florence um it's uh it's been a really interesting um i mean again i feel like i'm praising myself but i'm i was very happy about the sort of variety of voices that we brought in uh to present this year from um sort of italian architects to uh curator and uh and more sort of uh uh instructors and pedagogues let's say and today i'm i'm really excited to have um simone kill to add to the to the to the roster um i think you know something that we should all uh remember ours as we slowly transition out of this covered uh uh year or years we should say is that uh it's really exciting to see how we have learned to sort of come together at a distance and bring people that otherwise would have not been as easy to have part of our conversation so uh thank you thank you all for for your availability and for for participating in so many different ways uh i see people from from main campus on syracuse and from here on on our florence program so thank you all for for coming and joining us um and thank you also to dean speaks and associating uh julia cerniak for helping us in sort of coordinating and supporting these events throughout this semester so like i said i'm super excited to have simone joining us this semester in in the last few weeks in studio we've been thinking about the role of uh digital infrastructure in uh the reframing of the domestic space especially as we've been facing and uh sort of witnessing the radical changes that have been going into the build environment due and accelerated by uh the ongoing pandemic and i think i couldn't i could not find a better speaker to reflect on the way that these infrastructures have changed the way that we looked at the environment at our architecture as the building environment but also as the way that we start as designers to think about representation and visualization of this different ecologies and different uh bodies in space um so uh i think this is going to be a great uh conversation uh like i said i'm gonna try and keep your your specific intro short but i'm going to copy the bio on the on the chat so that everybody can take a look but simone is a designer and researcher which is based in amsterdam her practice which i'm sure all of you have spent a good amount of time online looking at is techno it's called technoflash um and investigates the representation of identity and digitization of biomass in the network space of appearance she holds a a bachelor in graphic design from the rhode island school of design as well as of an m.a in visual strategies from the sandberg institute in amsterdam um she was the recipient in 2016 of the fellowship of the new institute of rata in rastadam uh where i think the piece that the students uh the students have read uh kind of came from is that correct or that was a little the beginning of the spot meeting the pain and spot meaning it's very exciting because last year we had marina coming for uh to lecture so it's i think it's really incredible sort of flow of research across people uh so um she was a contributor to the pavilion of the 2018 venice biennial and i'm hoping i'm hoping i know that she's gonna talk more today about the relationship between machine vision and architecture so uh thank you again for joining us uh simone um uh please help me welcome her to our community well thank you daniela it's always very awkward to uh hear your own tv thank you very much um i'll just take it from here and again i think conversations are always um [Music] very welcome and much easier in digital space than a sort of monologue so i'll um right away start with the presentation um so that we have time for conversation afterwards and um what i'll speak about today is indeed the research starting with um spot mini and going to a work called homeschool um i mean my background also like you've just heard but it's mainly graphic design and i came to graphic design through photography so i think like design as a field wasn't really anything that existed for me going to art school and i only mentioned this because um in the work that i do but now since a few years sort of the camera as an object has become sort of the main focus or sort of the the main reason through which i do research and so it's been sort of a strange tragic story from the sort of obsession with a photographic camera to then being educated in um design mainly graphic design and now sort of finding back to the camera as a tool um but i think within all of this sort of graphic language design as an architecture is still really important and that might be something that we can pick up on a little bit later um blood camera sounds disgusting but very much is a term from a video artist tipi lotiris a swiss video artist and she refers to our eyes as a blood camera sort of you know manifesting that there's many kinds of cameras many kinds of seeing devices um and i personally hold this fascination with the middle ages also the drawing that you're looking at is sort of you know early renaissance um geometric studies studies in um perspective drawing this is already an area where we see um sort of remnants of architecture coming in this sort of needs to take three-dimensional environment and copy them onto a two-dimensional surface onto a paper onto a drawing onto a computer screen now um and one of the things that i'm interested in is what happens in this translation right is there information being lost um and also the obsession of the middle ages really comes from this idea is there really only one way to copy to look at a three-dimensionality on three-dimensionality is there really only this idea of perspective drawing as it's been developed like the 15 1600s mainly in europe or other other ways the small sort of a bigger question that i have in mind while i do this research at the same time also this quote by walt whitman he's an american um poet uh important as a queer poet um and stuart contradicts myself very well then i contradict myself i'm large i contain multitudes and for me it's always been sort of a yeah a red thread for my work um sort of starting very much with an obsession in face recognition identities um celebrity look-alikes and this weird space between uh face recognition technology trying to um sort of define or identify yourself whereas there is also such a thing as a look-alike or a margin of error and you know doesn't matter i think even in the work i still do now um to search for multitudes or we also can call it complexity like what is the world we live in how complex is that and how can you actually tear that down simplify it or reduce it to teach it to a computer vision system and of course within again this this there that translation i think there's either a lot of chants and challenges and sort of absurdity poetry beauty or there's actually you know a lot of violence um and sort of control and i think that sort of intense um dichotomy is the space i sort of find myself in um to start off the sort of the work with um computer vision and mainly architecture you know for me it started with um i mean more or less i think looking back there's always this sort of clear research tragic story uh if you're in it obviously it's not as logical um but this sort of headline was one that that was crucial for me um it was in the garden 2015 and it read that um south korean woman's hair eaten by robot vacuum cleaner actually slept and for me it really sort of rang bells where i was like you know sort of the way it's written it gives the robot agency as if it sort of purposefully sort of drove to that woman while she was sleeping and like started eating at the hair whereas what really happened is right it's a roomba a vacuum cleaner has three wheels um and while cleaning it essentially just got sort of caught up um in her hair and so for me what was important about this is much more um the assumptions that it reveals about how the robot is is engineered so the fact that you know the the cycle with which the robot operates so the fact that it started running um during a time where probably people weren't supposed to be sleeping they were maybe supposed to be outside of the house at work so that the house is empty for it to clean or the assumption that you don't sleep on a futon or on a floor but that you have a sort of western bed so all these sort of assumptions about behavior about interior about domesticity that actually led to the hair being eaten and not so much the robot attacking um a human being and at the same time this uh poop destroys which was uh posed by chessy newton in 2016 on facebook and sort of got picked up by the by the media afterwards and essentially this is what happened where you know the roomba again the vacuum cleaner doesn't recognize dog poop um so what happens instead of cleaning it just like drives over it and it gets like spread all over the uh house apartment whatever space it's supposed to clean um and so again i mean you know also humor i think for me was always important in being able to address these topics um because the way they uh sort of find themselves in society i think are really drastic um but if you start with that it sort of lose people's attention right away um so i think having sort of a sense of of reality and for me that normally is humor is important um and so getting to to the spot mini which is this thing um you know it's yeah for one i think um the thing itself is quite fascinating right the fact that a robot is walking on four legs is already um a sort of fascinating thing um but at the same time i was more curious by the fact that boss dynamics would have announced it as their first i mean by then quick sort of factory embossing dynamics to the ones that aren't familiar but they sort of became sort of internet famous by posting youtube videos of robots they've developed um doing sort of funny things or um amazing things one of them was like a humanoid called atlas and he would like be sorting boxes um but he would also sort of dance um so they were playing with slapstick also playing with humor actually to make these machines they could seem really creepy and really scary quite um well make them look kind of silly um and so you know most of their funding they're a private company but most of their funding either comes through um sort of academia or also the american military one thing they were interested in is for example developing robots there's a bigger model of the spot that was designed so it could carry large loads so that the soldier for example doesn't need to carry all of that and it's essentially a second body instead of a vehicle um and because it has four legs it can go over completely different terrain than a vehicle can and so um boston dynamics sort of departing from that uh strand of research um you know sort of decided hey we want to make a robot that's attainable something that we can actually launch to market and so the spot mini was their first product that they were interested in actually being able to make it available to quote-unquote normal people um at the same time they were acquired by google um and they sort of mentioned that during neither acquired for softbank again but they mentioned that um during their time of being acquired by google they were inspired by the google offices and for me it is also architecturally interesting right sort of google offices wanting to be a domestic space they want to be as welcoming as possible they don't want to be sort of office cubicles so designing a robot for a google office at the same time also has hints of domesticity within it or so you would assume so here we are at the spot mini one thing that i sort of really love about this product photo is the table probably is a sort of standard ikea table but you see that the legs have been elongated um to fix the um ergonomics of the robot itself so that's already awesome that you know you can't just like lower the robot it has to look a precise way and to do that they need to make these legs longer so you see a little um slits where they've cut and added another leg right here when do we get to the next slide let's see he really likes this slide we're gonna do this manually all right it's not playing the video i think yeah here we go um this is a video that uh one of the ones that boss dynamics posted online on youtube to show what plot mini can do um and what struck me again is the sort of this this house or like the sonography it walks in um because once i knew that they sort of you know are trying to design this for for a market um knowing that it should be domestic so the assumption was that this is supposed to become a robot that can help in your house and so the question i ask myself is like how do you define what a house is or what a home is um such a global scale right if you want to launch a product you want this to be something that um as many people as possible can buy um so that within itself sort of has a has a need for universality right so like now it's picking up a glass um but what is a glass right all these questions that are quite banal or philosophical actually become really well quite real uh when you speak about computer vision where you actually have to define these things um and when you speak about the glass it's still quite um straightforward but now it's like putting this thing in the trash how did it decide that it's trash right like i if i have to clean up my room it's quite difficult for myself even to sort of decide what i'm gonna throw away and what i keep um again here also some sort of slapstick right some sort of mario kart moment of having this thing slip on the banana um but so ultimately this video served as sort of my only guide to a possible training data set like to impossible um yeah sort of main home that spots dynamics would use to train or to teach these spot mini what a home or a domestic environment was and what i realized is that um i started modeling this home with its furniture in blender and the thing that i realized is that i didn't really need to model anything um because the stuff was so standard that i could just download it from sketchup right so like the the house was sort of a levee town sears catalog house with a front porch um and then the interior very much which is ikea so all of that is available on the sketchup warehouse and you know that sort of pointed towards again another sort of strand of standardization right the fact that if we're all gonna sort of at some point live in ikea homes that really solves the problem for computer vision you have to define what a glass is or what not because you could just train it on the um ikea data set and strongly even you know knowing that ikea catalogs now for you know more than half of um the photos are cgi rather than actual photos so ikea is actually creating highly detailed models of all of their furniture right i'm like wait can't they just like collaborate you know that can become the data sets uh to sort of train everyone and the one hand is handy at the same time i think it's pretty scary um thinking about the sort of unification of what a domestic space is supposed to look like right you can't sort of decide what you want to sleep on you have to sleep on that kind of bed if you want the technology to function in your home um the renders we're looking at are ones that i've made of this house these kinds of renders are a depth path i don't know how many of you are familiar with that it's also sort of c depth in um computer rendering and normally they're they're rendered in gray so from white to black and they show the distance from the virtual camera to sort of the closest object or the one furthest away so it's a way of measuring distance in virtual space where there is none technically um and so yeah one thing i like to do and i'll get to back later as well is to also use these various ways of how cgi or computer generated imagery actually works um to create imagery that aren't meant for the human eye and so the depth image is one of them right it's not supposed to be an end product it's supposed to be something that you use for example to create mist or to create a depth of field in your cgi workflow it's not supposed to be your final image and this is like an update of 2021 um you know nypd has acquired now um spot minis as one example i think they're also using them in singapore to control social distancing for example and so yeah i don't think they've entered the domestic space as of now and sadly they're more surveillance robots um but yeah this is where we are i mean interestingly as well for this one they've used it to enter a home in the bronx as far as i know because of a hostage situation so still there's an element i think of home and domesticity and that is to be discussed here as well right i think like how does it navigate at the same time also it's used a lot for construction inspection so to make sure that construction goes according to plan but especially on the huge building side threat where you wouldn't want to have a human being go through the whole construction site all the time um to the sort of next chapter of this research um which was a short film called homeschool i mean also in my practice sort of what i show is the visual output but normally i do a lot of writing which ends up being the research to then you know try and pack research into writing so i can make visual work i think trying to combine both sometimes feels like then i just overload or like try to do too much um in a piece of film for example um so i yeah for me it's important to have both um so that i can sort of be different voices or different roles in them and also that they can address a different audience so homeschool now is trying to sort of take that question of the trading data set sort of saying hey for spot mini i didn't have access to a training data set for computer vision um which one would i have access to and so what i did find is one called cnet rgbd and that is being developed by imperial college in london by a group that is sponsored by dyson um specifically because tyson at the time just like 2016 was saying that you know we have the technology to create autonomous domestic robots but what we don't have is the intelligence um so specifically that you know you don't have uh the sort of you have the mechanical text this thing can like ride around or you can walk um but what doesn't exist is the fact that it can for example whatever it needs to do like start the trash or do the dishes or fold your laundry um i mean it's it's a similar idea of hygiene i think that has appeared in the 50s or 60s as well right when sort of a washing machine or dishwasher was introduced or a vacuum cleaner for that matter um so i don't think anything has changed but that aspiration of sort of more leisure time even though none of that happened it was just a heightened idea of what hygiene meant um now it's just it's not the dishwasher it's much more sort of robots that suddenly will sort of enter your home and magically do like elves right like do everything while you contend to your life so what you're looking at is part of that training data set um what cnet is is i mean it's a little bit of a mouthful but what really you have to imagine is essentially a bunch of folders full of 3d files um i mean training data sets for computer vision can either be a bunch of photos um right so like actual photography digital photography whatever um or it can be 3d data and the trouble is always where do you find enough examples for whatever you're trying to train right so like if you're trying to train a computer vision data set on um on dogs for example you have to assemble as many photos or examples of what a dog is uh whatever that category um is to be able to sort of say this is a dog and this is not a dog that's a cat or that's a lion um same with the domestic um data set but here of course the problem is like how do you enter people's homes right like how do you gather information about how the inside of the private space looks like um i think airbnb was a little bit an answer for that but of course um you know then everything also starts looking the same i think airbnbs are also a space that has become um pretty standardized so one other answer to this was to say hey we can sort of tap into another workflow of design of architectural design which is often three-dimensional and happens in 3d programs and just use that data rather than the actual lift spaces and so a lot of this data comes from the sketchup warehouse again and also sort of obscure data sets like gaming um gaming warehouses and stuff like that and you'll see the effect of that in a little bit and so what i was interested in is like you know for one where does this data come from but also how is it organized um and so one of the things is that there's these uh five categories and so the all of these objects that you just saw you know they're sort of put into um five different room categories that then a house has and already here i'm super interested in right like who lives in a house that actually is that large right that already sets such a standard or an assumption about what a house is um that isn't universal at all um and so these five are bedrooms office scenes kitchens living rooms and bathrooms and then the number of models um that there are and this slide is all of the objects and here um you see a little bit uh some of the gaming um you know gaming assets that have sort of found their way in here i think i can't find it right now but somewhere it says pistol where is it i mean there's there's a bunch of weapons um in here which points to you know well that could be domestic i guess it depends on what country we're talking about um but again i think this is one of the sort of the political questions right that actually is part of such a data set um oh yeah there's the gun yes what happened there we go um one thing i also really liked i was speaking to one of the engineers um of the state of fact and he was quite upset that's during like what they do is sort of they generate rooms that i have folders of rooms and then they have folders of objects and with a script that they write sort of saying there's a possibility that these kinds of objects will appear in a kitchen and then that script generates a bunch of kitchens um and this script sort of you know assumed that yes of course the category box could be in all of these rooms because the box is just like a container that holds stuff but what happened is that actually a mailbox got placed in all of these rooms um and to them this was super frustrating because they couldn't figure out why mailbox has suddenly appeared in all of these um spaces they really just makes no sense right mailbox absolutely is outside of the house it just turned out that right like the category box had tons of mailboxes inside again i think for a human being it's like you know sometimes it even feels stupid explaining what happened because it's so obvious but having to relay that into a computer vision system is a very different thing because there is no thing as a sort of human logic and the second um i wanted to point out is that this data set also comes with textures and yes this sort of looks like the sims and the 90s but this is really sort of the standard um at least in sort of 2017 that this data set um was released now it already looks different right with game engines um this sort of development goes super fast um but back then three to four years ago um this was the standard and so i was fascinated by the top right image which is this one here and what we're looking at is a office scene and um it's fox news which is on the projector so there is one folder that's called tv or media screens um and projection surfaces and all of the textures in that folder that are automatically chosen for all of the tvs are fox news so again like i'm not like saying that that's by design it's more what i am um saying is that a training database if it's accidental or not is a piece of design right so if it's not treated with um care or with intent um that's actually an issue because if you don't right there's i think a lot of um well there's a lot of subjectivity um that will be part of it anyways and that's probably um completely unavoidable so then sort of trying to to create something that's objective or that's general becomes not only um difficult but also dangerous one of the design steps i took is to try and take all of this content you know i was like this is all digital how do you even make this interesting to people it's sort of all of this um the training data set is purposefully never visible for the human being right it's like the one thing you want to make the technology that then is visible and it functions awesomely but you never actually want to see the training data set um so i was making this sort of these well sitting things furnitures wasn't too happy with them i just ended up making um the short film instead where i took the contents of this training data set and just created my own sort of 3d stenography um my own house and littered it with well mailboxes and all kinds of other stuff from the training data sets to then make a movie where a first-person camera with a child's voice is sort of trying to orient itself in this space and trying to figure out what the hell is going on um and the voice over here the video is also using the vocabulary from the training data set it's using all of the object categories for example as part of the vocabulary that it has to orient itself at the same time the way the video is rendered sort of referencing again the depth images i showed before where there's like what we're looking at here for example is a way of denoising like in 3d rendering right you get quite grainy images if you don't render with a high sample count if you don't render them for a long enough time essentially and to forego that and to render faster you can de-noise and if you denoise on an image that is really crappy looking especially now with ai denoisers you get something like this and i was again also super fascinated by ai denoisers that are trained on interior images and sort of realized that the interior images of this ai denoiser are actually exactly the same interiors that this computer vision system is being trained on so it's trying to approximate um what it sees without actually getting a lot of information this was pretty much a black image before it was denoised and at the same time these again these sort of depth renders the distance in the virtual space as well as this which is an ambient occlusion path um so what you're seeing is actually shadow they're not objects with a white material if that makes sense it's the it's a shadow between the material that's rendered and by that you see the objects um so yeah one thing i'm interested in is more like what kind of ways of looking or seeing um are there in the process of creating computer generated imagery that surpass um the human eye and in what way are they beautiful in themselves right they don't necessarily have to only serve the purpose of making something that's photorealistic they can be something else as well and ultimately this was the installation so um making a space that is approximately um like to the human body feels yet again like it could be a domestic space so the projection was um quite large on purpose i think the relationship with your body to the projected screen is very different um i also don't mind if people watch the video on the phone screen or whatever at all but it's more if you're already going to go through like you know the ordeal of making an installation i think it's quite important um to really bring a physical element or a spatial element um to the work as well and one thing that might be interesting to say is that if there's four speakers um for their work and each speaker has its own track so there's also a sort of a spatial simulation um that's part of the video where suddenly sort of in the corner left you hear dog barking and what's so nice is that people really turn around even though they know there's absolutely no dog but it's just it's so intuitive um to orient yourself but that information a robot doesn't have well yet um one thing they're working on and this is a trailer which i don't know if you hear the sound no there is no sound okay i'll send it to you later so i'll send a link to that otherwise pretty strange without sense but um yes thank you that's awesome thank you thank you simon yes of course um uh i could like i keep saying this at the beginning of the q a session i could have a very long conversation but i would love to hear from the students and see if there are any questions from uh from from from the group yes please i mean you know i don't really mind what the question is about it can also be about the studio or about how to make this work or a work all good yeah so i asked them to compile a series of questions prior to prior to the reading yes sorry prior to meeting and prior to the presentation so uh this is all and we have a very long list of questions here so i don't really understand why i think this is just zoom zoom embarrassment exactly i get it yeah we just need someone some courageous uh soul to jump into the mix oh what's the vastness here hi bass she's a thesis student since syracuse feel free to just jump and ask the voice verbally okay um hi simone um thank you for the lecture i really enjoyed it um that's really similar to what i'm exploring for thesis um and i wanted to ask um uh because we're trying to develop like data sets as well that um uh like instead of really aiding computer vision they sort of like try to disrupt computer vision um and so we're trying to create catalogs that involve um like more participatory models where we start create curating a data set and other people can add to that or um uh see how they interpret those models in their environments and things like that um so i was curious when you said that um with airbnbs we're creating sort of like the standardized aesthetic um and so and you're using sketchfab models and other like uh model platforms so would you imagine that it'll be possible for um residents to partake in this or do you think that that would um i don't know like create difficulties for machine vision or other challenges yeah no thank you um yeah i mean i think one thing computer vision is really good at is specificity right so like um i think sort of a standard example is always like if you're developing a drone that has to sort of you know surveilla an oil pipeline um for a leak you just it's pretty straightforward um to try and get examples of what a hole in a pipe looks like i'm gonna teach that i think as soon as um generalization is introduced it becomes difficult no matter what the area right is domesticity is just one that makes it really obvious that that is difficult um but i think one thing is more uh specificity i think is important and also actually really beautiful i think as soon as in any sort of design process you arrive at a sort of general object even it becomes really bland i think the designers are kava and bint for example are super is like a really important practice for me in terms of inspiration um and like their design has always struggled to be architectural even right to like become like part of the built environment even though that's what they wanted to do but they were struggling with like building regulations and safety regulations and stuff like that that you know doesn't even have anything to do with computer vision like let alone the fact that they would probably never be recognized by computer vision um sort of devices as a domestic um domestic space either so i think what we're dealing with has less to do with sort of the new technology and it's very much part of um well other kinds of technologies like yeah safety regulations for example already so already i think there is yeah i think in terms of data sets and disrupting them i think it's much more about showing what else it could be um building maybe very personalized data sets that also answer a specific question right like for whom like if it's a community-built thing um what does this community want out of it um rather than um and i think this is where open source technology might be an interesting um yeah way of looking at it right like a software like blender to not drift too far into another field but has suddenly become super popular but it's been around for so long um but it's community built and i think it's really become so strong because people have built what they wanted and because of that it's also sometimes difficult to enter because you're entering someone else's brain um as how someone else wants something to be done within a software within the interface um but given enough time i think it actually makes it um weirdly enough actually quite attainable again not to say general but it makes it quite um something that is relatable on a broader scale but it started as something that was um really specific and really small and communal so i think that's something um yeah something to keep in mind yeah no that makes a lot of sense and i think uh is how we want to also posit our own research thank you for that answer oh and one thing i wanted to add is um sort of my uh sort of the second chapter i'm working on from home school is much more focusing on language um it's just the language used within these data sets so for example taking the five room categories um already what are these room called right quite when i'll put like why is there such a thing called room even and why is it english so i think also there that's something that's interesting to think about when you build these categories so the the visual side like the object itself um but also how you refer to it um and i think their context is always lost right like if you think of an object to sit on my examples i was like then why isn't the toilet also part of the chair category but of course there's a context in the behavior that's that's part of the object that's lost if you only look at the thing itself but that is very much part of how yeah you experience a space so that might be interesting to look into as well everyone's very timid anyone else i try to learn and uh really dwell in these moments of silence in zoom and have people get a little bit uncomfortable with it rather than just jumping the gun because otherwise okay no questions so far maybe i'll ask one myself um really it was yeah i'll give i'll give it some i will give it some some some time for them to to enter the conversation but really this was super inspiring simone i took so many notes and i think you know if there is something that resonates in the conversation as a continuous threads i think you also preface it in the beginning with the with a quote um about may i contradict myself or how sort of a reflection on identities and the possibilities of containing multitudes i think that's really an incredible and super valuable investigation especially when dealing with uh technologies or modes of representation that seem to present themselves as objective or as a matter of factual uh in in the way that they are assembled in the way that they appear in public to the public eye etc so so really uh you know maybe this is not necessarily we didn't take this topic ahead and sort of face in front but i think it's very much part of the uh you know the pedagogical ambition of of the studio or many other seminars their students are taking in terms of how do you how you kept asking during the presentation how does a home looks like or how who decides how a home looks like and when we start thinking about systems of organization how do we develop a sensibility or some sort of agility to be able to understand and recognize the moments where bias are embedded within the trainings that files are embedded within the imaginary of of the house or of a certain function of sorts so so i think those are really really exciting and of course i really appreciate the the way in which your um you you are embedding this ambiguity into the the aesthetics of your projects as well right so when you take uh you know just simple tweaks like the the the z-pass or the depth pass where a grayscale gradient is translated to this sort of like blot camera or some sort of like sort of really uh moody moody environment um i think it's it's really exciting and and so i guess i guess my question is uh you're it's always a matter of sort of not always but sort of like it it seems to me there is in your practice you're trying to sort of always deploy uh techniques and and aesthetics onto themselves sort of as a way of of uh feeding the the biases of the of of the training instead of the biases of the ai denoiser into an image that is supposed to fix or supposed to standardize or to make uh legible in a sort of homogeneous way and rather as a way of breaking that system or to create some gaps where uh that what where where some questions arise and so i wonder like how important like uh i guess my question is how do you like how important it is for your audience to be fully aware all of this of all of these mechanisms that are happening in the background uh to be able to um you know process and understand the imagery that you're that you're producing yeah um the hope is always that it functions on various levels um i think this is where animation um or cinematic sort of techniques um yeah have become really important in my work for one because of temporality like just the fact that it's like in graphic design i always found it super difficult to make a poster um it's like the most dreadful thing instead of working in versions or series um makes a lot of sense to me but making one was difficult and i think the same is with videos so literally getting the space and the time to move around um to add a voice to add um sound and to build on a narrative um really helps at least in my brain in sort of creating a world um and so i really always um try and also hope then you never quite know how the audience receives it but um that there's different entry points right but there's one where you might and this is also why i use a child's voice in most of my work because i think it lowers sort of the the threshold of accessibility sometimes it also renders to work a little bit literally childish right where it's like why do i need to care i've sort of solved all of these problems as a ground up because the questions of for example navigation and of learning about the environment of learning language are quite similar for a child but i think so there you know one is maybe someone only just can hook into um the voice over and get some sort of get something out of that i mean there's also within the voiceover itself sometimes i make like references to a philosophical quote like in home school um it's sort of bricking on the philosopher wittenstein and sort of paraphrasing he sort of said um that your own language i mean i'm paraphrasing this in a very bad way now but like that your own language limits your world so like your own vocabulary sort of defines what your world is like and i'm sort of paraphrasing that for the age of computer vision sort of saying what the size of the training data set is going to define or limit um the world of that computer vision system and like if as a viewer you're not going to know that that's a paraphrase like i don't really care um but if you do you know it's almost like an easter egg in a computer game then it's kind of cool that you recognize that and i think it's the same with the cgi technology where like someone that knows and that produces those kinds of images will have a familiarity with the process um and someone that doesn't um well they don't but i don't really mind um i've i sort of don't need to explain all of the ins and outs so i'm much more interested in yeah in talking to people in different ways but still producing a thing um that can live in in many different places um i think now it's also only mostly art installations but i always joke like you know if i could do like a sunday morning kids tv like i'd be happy to you know that'd be awesome um so it's sort of yeah i'm also it's more like it's an artist practice now but i still always see it as a design practice like that's part of the design like what do you make and for whom and so creating these very layered um works in terms of sound in terms of um voice over in terms of images like also the with the images i borrowed so much the end scene of homeschool is um sort of taken from antonioni's uh films a risky point where in death valley you see a huge villa being blown up and all of the contents of the villa are in this like really beautiful antonioni way sort of in slow motion um in the sky and it's like a seven minute um video sequence to a pink floyd song and i used that same pink floyd song in homeschool and it sort of chopped and screwed at the end scene is also that same sort of um you know deflection of capitalism of antonioni again if no one knows it doesn't matter but for me there's sort of an inside joke in using all of these um references and sort of breaking like using them as samples as part of pop culture to produce yet another object but yeah it doesn't always have to be unpacked i think but no no but yeah no that's that's great of course it's it's sort of like in in itself is sort of like the same way of like acting in disguise like the the idea of like producing an image that could potentially leave um unobserved undetected within the aesthetics of social media and then under close uh inspection you realize the the the easter egg or the kind of subversive message that is embedded within it absolutely yeah that's it so i think that's really awesome for sure uh and i think that you know uh with the students when i think this is something to sort of keep in mind especially when we think about ways of representing a project or sort of a design where uh like like the the ways in which your your ways of representing a project can be explicit and always made uh sort of like a singular narrative where everybody is capturing the same content or sort of populating this sort of uh subversive and and sort of contradicting at times sort of narratives where a space or the possible inhabiting bodies of that image are affected and engaged with in different ways so i think that's that's really awesome i think to to to for for everyone to keep in mind and also specificity there i mean i've used that word enough now but like i think like i've always for now chosen that my videos for example are in english um but i've recently wondered why you know like i grew up speaking swiss german there's not that many people speaking swiss german all over the world but like why not just make a video that's in swiss german without any subtitles right i think there's always a sort of anxiety of like oh but then no one sees my work or like oh it's not accessible um on social media things like that um but there's so many i think levels of accessibility that we miss anyways um i mean it's there's always a certain assumption of who the audience is i think like i don't have subtitles on my videos and i think i really should because yeah that's also a level of accessibility that i'm missing there so you know i think that also in student work i think don't be afraid to make things that are to an audience that you choose um i think there's only strong basically yeah i thought that that was a really interesting point that you made about making it in uh your native language versus you know english for accessibility purposes i mean you know because we just viewed that video without any sound you know i think it gives a different type of opportunity to to gain an understanding of what you're trying to show when you you don't have the assistance of the words right so i i think yeah i think it'd be really interesting to see your work um in perhaps the language i don't i don't mean to speak yes absolutely i mean there's also a lot of reference there's a video called metahaven that some of you might know their design studio in amsterdam they've also moved into the world of video a lot and they've played with that as well where they have voiceovers in different languages but then also subtitles in yet another language that might you know not be commonly represented and so yeah it's super interesting um to use those elements of moving image but it doesn't have to be right can also be written text can be the way an image um can be read by whomever um yeah i think it's such an important consideration to make yeah another thing that i think is really interesting about when you produce work uh in the language you're most comfortable with is that um i guess the best example for me is that you know i'll watch a lot of uh like foreign films like i mean from the us perspective they're foreign and i and i've been watching something that had you know voiceover and it was it was so distracting from what the film was supposed to be that it was it was you gained so much more of the the intent of the emotion and what was trying to be said just through the tone of the voice even if you don't understand the words and you have to read the subtitles i think there's really something to be said about you know the message that you're conveying through what you're most comfortable in because you know i find when i try to speak another language it's much more robotic you know and sophomoric because you know i'm not very comfortable in that language so if you have you know someone trying to put something into english even if it's really good it might not have the same emotion and tone sure so i think that'd be really something i'd love to look at yeah sure yeah i mean for me it's that english really is working language um i think when i'm asked by german speaking places to give a lecture i still always want to do it in english because it's so difficult to do it in german because it's not i it's like family and friends that i use that we're gathering for um but that's also something that i think you know people with other languages are familiar with um but no absolutely you know i think your brain functions on different levels depending on what sort of terrain you enter and i mean i'm also celebrity lookalikes i know for example in italy a lot of the films are dubbed rather than subtitles and there's for example specific people that always do specific actors like just like the brad pitt in italian etc so that's like a whole other world that's super interesting in terms of how familiar you are with someone's voice and how you identify them and also yeah what does it mean to voice act anyway is like a whole better territory but yeah absolutely you know that's absolutely uh crucial i mean we we started this semester with uh with this um um how can i say this this collaboration where we have speakers coming in and do presentation only in italian with english subtitles to so create this flexibility or this mental agility to be able to put multiple cultural elements within the same conversation and i think you know i think your presentation really speaks so much into this desire to contaminate this this this language or this visualization uh which perhaps makes me you know like i was a little bit curious uh if i can ask you one more question like about you start with your um your trajectory or your sort of uh um sort of you say you came into design from photography and uh so much related to the use of cameras and uh photographic cameras and in in your work it seems like as much as of course conceptually the data set the the way in which this engage with the physical reality of robots and spaces and architecture in in sort of uh um in in other media or in physical matter i think all of your work is sort of like all of these the visualization of your work are sort of remain kind of contained within the cgi or sort of rendered environment and i wonder if like at any point you find sort of moments where the need of sort of introducing sort of multimedia uh engagements or sort of moments of switching between one and the other or moments of sort of misalignment between the two i think could be uh if it never ever comes up or how or or whether like at the idea of staying within the cgi setup it's helpful in any way for you yeah i think there's two aspects one is um like when presenting work i think for me it's always um necessary to step beyond like to really think of like not just put the work on a screen but really think of the physical elements that are part of the work as well so for example home school i think the size of the screen was important but of course you don't get that kind of space everywhere but there it was possible um at the same time sort of gathering all of the chairs they had in that exhibition space rather than getting something rented specifically so sort of making an inventory of the exhibition space itself and using that as the furniture to watch the video in um are sort of tactics that i do with every work i mean they're different um depending but um like for venice for example i made this a huge uh jumping castle like an inflatable jumping castle that was based on a um uh 16th century perspective drawing to watch a film right so i think it's always really important to have that relationship um i mean there's also a reason i sort of stopped doing physical work simply because of like what you do with it after like for a while i've done much more also sort of 3d printing um experiments and work like that and i found it really difficult to just have it in storage afterwards honestly you know it was like now you've like produced all it is plastic and it's just here and because of that i really use more the moment of insulation of a work to sort of realize that rather than sort of making work um that's physical and i think in the world of cgi i mean i've done video before and then sort of entering the world of the data sets it's become i mean before that also there was a work where i started working with um scanning a celebrity lookalike because the idea of scanning a lookalike and what are the rights of a digital scan was part of the question there and that sort of slowly um brought me into the world of cgi but i think there's so many other avenues there right like with photogrammetry um we're starting to use a camera together with the cgi camera um that i think yeah it's not as limited so i think it's two aspects one is like how do you produce the stuff that's on screen and i think that can go beyond just the cgi uh sort of 3d program um and at the same time then it's like if you are showing it somewhere then how do you do that and i think that's another opportunity to really think of the spatial configuration like how the body of visitors whoever they are sort of relates to the work itself so that it's um like i think we're so used to seeing uh especially cgi on screen so how do you yeah ask a question again how do you need to present it so that it doesn't look commonplace and yeah that's something i'm trying to figure out yeah that's that's great um i assume there are no other questions today students have been long day long day of studio today both luca and i exactly i can't believe it red line that drawing um so good um but again i wanna we're trying to be as respectable as possible to everybody's time uh so uh otherwise this zoom life produces like you said just another surveillance robot or in this case another uh monitoring 24 hours device of where you are at but so again simone thank you so much for uh for your availability for joining our series was really incredibly inspiring and i'm hope will get a chance to meet and catch up here in europe soon otherwise good luck with the your future project and i hope to see you soon definitely and if there's no any burning questions come up then uh well you have my email [Laughter] thanks for having me that's great thank you simon thank evening hi everyone thank you thank you so much

Channel: Syracuse Architecture (Syracuse University School of Architecture)

Views: 10

Rating: 5 out of 5

Keywords: Architecture, robots, florence, syracuse university

Id: o32iutc3Ui0

Length: 62min 15sec (3735 seconds)

Published: Tue Apr 20 2021