 Hello and welcome to the wikipakka stage on the dining room. It is 9 p.m. prime time We are live with translation and bleepdrag and blinry are here and Have now the third edition of operation mindfuck and it's going to be very fun funny Have fun. All right. We also say Good evening. We are so glad that you're all here. Hello to the stream Hello to the wikipakka living room where some people might also be sitting. So this room is a pretty full So it's nice that you're all sitting together We have art computers and curious things here. So we'll Switch sides Every now and again and we'll start with blinry. Okay. I am starting with chess variants So a few months ago. I was at the MM CD in Darmstadt There was a presentation about a medieval board games and one of the board games that was shown was this one It has Spanish roots. I can't speak Medieval Spanish, but maybe it's can't I see directs or something means grand chess and If you know a just of today and compared this is a 12 by 12 board. So it's larger and the Figures are a bit weird. So there's a giraffe and I think this thing here next to the king is the Elephant bird for example, and they have very weird movement patterns, which we don't know with our standard figures So the giraffe for example moves to Fields diagonally and then once straight note three fields diagonally and one Or a lot of these so that's kind of like the horse today and The elephant bird goes one Diagonal and then continues to slide as far as once which is pretty weird And now there it turns out there is a whole community behind chess variants like this Which includes fairy chess pieces. So there are lists where people think up these things and Throw it together and make up new chess variants and I thought that was pretty impressive This game so here the Rooks move normally and the row at the bottom has these weird rows Weird figures and to accelerate the game a bit. There's also the rule that you have an extra dice Determining with which kind of figures you can move Another chess variant is losing chess and the objective is to lose all of your own figures as fast as possible And if you have to if you can Take another figure you have to but you can choose which one if you have multiple options and the king doesn't have a special status It's a normal figure like the others. It can be beaten and there's no Jackmate and Whoever loses their figures first wins the game which has some weird effects So if you make a move you create weird chain reactions because both players have to Beat another figure and they can't stop and in this gift that's playing all the time The Rook is played first. So I think D3 and That's a very bad move because then black can always win it turns out So people modeled this and did the math and there's a sequence for black Which forces that black is going to lose all the figure. So there are some openings That are a really bad idea right infinite chess Has the usual figures in the usual arrangement, but the board is infinitely large So that's more a kind of theoretical model. I'm not I don't know if people actually play this It's kind of difficult in the real world, but people think about how this works and if you can find out things about it so You have a weird situation where you have to move one million fields with the queen Upwards and then it will take the Rooks a while to catch up So it's weird to analyze. There are people who investigate this and found out that given a certain arrangement and Number and you can determine if there is a possibility for the current player to win within and moves Which is a pretty cool result, but it's not very useful To write Artificial intelligence or something because what you really want to do is to figure out if from this position you can win at all So if you know I can win in 10 moves That's kind of not that useful if it means like you have to move a million Fields up with the queen so it might take a while until things happen So it's a theoretical Has theoretical value There's also blind chess Which is so you might know blindfolded chess where people actually blindfold their eyes and then they're told Which moves happen and they have to memorize where each figure is and what they want to do so blind chess is different so both players are in front of a board and So they don't see the figures of the other color only their own and they don't get told what happens So they have to infer that Cautiously move their own figures and try things out and then they're told if it's an illegal move then they can Try something else so they can explore the fields what their positions of the other player might be and Try to find your own good strategy. So that means you need a third person who Communicates between the two players and watches out that nothing illegal takes place I checked and there's a very Defined a set of words which this third person can use which is for example, no if it's not a legal move like if you try to move across another figure or if you Make a move which is independently so not possible independently of the other players Figures then you say hell no And This is a small website. Oh, I have you have to click out of this presentation. So let's take a look This was written by a human called Pippin bar who also builds other interesting game prototypes and so these are chess variants you can actually try out and one I really like is Quantum chess So you don't get told what the rules are you have to find out by trial and error So let's try to make a move with the rook. Oh We have two rocks now So it's blacks move. So there's no artificial intelligence. So two people move against each other and the system checks that the rules are Followed so if you make a move the figure moves to all the possible positions And now we have a lot of queens for example, which might be useful. Maybe it's also a disturbance So that's quantum chess. I also wanted to show you gravity chess So let's try to make a move and you're already laughing Yeah so and so on and then all the figures fall down and you can play this and someone tweeted this and other people try to actually play this and Analyze final positions of this game where you can not do very much And it's really cool. I think and last but not least. I wanted to show you chess boxing You might have heard of this one. It's an actual sport a series sport Started as an art project because someone thought how wouldn't it be funny if people had to box and play chess at the same time or Alternating Lee and it became really popular and there are now chess boxing associations which organize championships So it works like this. It's in rounds you play three minutes of chess and then you box for three minutes and You alternate and you can win by knocking out your opponent or by checkmating them and the rules also say that you can give up and It's it says you can give up in the chess part or in the boxing part. So, yeah, that's funny All right one thing that occupied me a lot in the past few months is Twitter bots mainly artsy Twitter bots a Twitter bot is normally simply a Twitter account that's Where posts are written by a by program sometimes once a day sometimes several times a day And I'm going to show you some of my favorite Twitter bots One of them is very simple. It's endless screaming and it never posts anything other than are in varying Lengths, which is very important another one is Big Ben clock. It's quite similar it Simply spells out the beats of Big Ben Of course at the right times Another one that's very nice is choose to accept it gives you secret missions that it Randomly generates one of them is your mission should you choose to accept it is to sneak into The casino of Colin the hair stylist there You must steal the horn of plenty Melvin the renter mob inflate Colin and finally escape using a secret tunnel It tends to be a bit confused And it's generated from a grammar and from Building blocks There's the cocktail button. Well, it the bot cocktails and it builds builds funny cocktails from Random ingredients and it's interactive so you can tweet to it So you can tell it makes me something and then it thinks of a of a cocktail name and No, we don't currently want to update Linux and it gives you a recipe for a cocktail and another fun thing is that you can Ask it to Surprise you and then it'll tweet you it'll generate a random cocktail and then there's Mazing bot or amazing bot and it generates mazes and you can play them It looks like this and then underneath the tweet you can reply with the directions So for example Up or down or the only directions that currently makes sense and when you tweet at it then it will reply with a With a picture of the new state So you can collaboratively Corrupted collaboratively solve a maze such as this and Another one that I read like is the Nixie bot I'm not sure who built it, but somebody built this Nixie tube setup and you can Tweet at the bot with oh no With the text and then It'll be shown on these tubes it's It's filmed by a raspy cam it and then you get a get a get a moving image of your text It usually takes a few minutes and It's very nice because the person who owns this Nixie bot likes to Decorate decorate the tubes and you can also see what time of day it is it's very much alive and the bot likes to post a daily review where everything that happened over the course of a day is Collected in a in large video and it'll be posted once a day. So I Can very much recommend that you tweet at this spot Finally, I want to show you one of my bots which I Built it's not on Twitter. It's on Masebon. It's the to read half a bot. So the two re Touristic sign Bot these those are the signs that you know from the side of the Autobahn or the highway and they tend to be a bit obscure and tell you what a Particularly region particular region or city might be famous for and I built a bot that generates this kind of sign It fetches data from two databases Wikidata if you if you're in the Vicky paka Vicky then you probably know Vicky data It's a sister project of Wikipedia. It's a machine readable database and I fetch City names and object names from Wikidata. So these are real objects and cities or places and then I Fetch an icon from the noun project, which is a collection of SVG graphics that you can use and then my bots Bot builds an image that looks roughly like one of these signs and These tend to tend to be quite obscure there might for example be the chips or fries museum Actually, all of these happen to be museums There that's not usually the case You can also get something like the like the Verical city MacBook or the bowl city MacBook and this bot is on Mac the book due to a very interesting story a while ago Twitter ended has changed its API and it's become a lot more difficult To get an API key for Twitter. You need to fill out a form of what you want to use your Twitter account for and Twitter would is very reluctant to hand out these keys and this Really shook up the scene because it's become difficult to build these bots and because older bots are no longer functional and it brings up debates of How we should maybe deal with archiving if things we built once upon a time can Stop working at some point and the Solution that many people found is mastodon, which is a decentralized Twitter alternative and there's a Separate instance for this kind of bots is called bot in dot space and this has become a home for Lots of these artsy Twitter or not Twitter mastodon bots. I can highly recommend clicking through these There are lots of fun projects that if you fancy building your own bot then there are some products I can recommend there's bot wiki which is a wiki that collects What variants you can click through it, but it's not very complete yet If you find a bot that you very much like then you can you can add it to this wiki But otherwise you can just click through it and find get inspired by what other people build. There are bots that Write poems or music and there's lots to discover if you want to build one yourself then there's a cheap bot done quick which is website that lets you host bots without a lot of prior knowledge and you can also use tracer which is a tool that Makes it very easy to build grammars for example to generate texts or images If you're even more interested then I'm afraid I don't have a link for it, but you can check Blindry's blog He gave a workshop on this that was unfortunately not recorded But you can't find the slides on Blindry's blog and he explains where exactly how exactly this works Mixed up will be ray marching and sign distance functions. That sounds rather scary, but I'll tell you how I came about Working with this We had been around three years ago on a demo party in Cologne It's kind of a meet-up where several hundred people come together and sit together in dark Horms halls and sit on the laptops and code and they are coding videos. They're coding little snippets That Generate or do rather interesting graphical and graphical things. There are several categories in which they can Sign up for and one of them is a restriction for the program that you're writing That the program must not exceed a certain size Certain files as for example five kilo four kilobytes which I Think is very very little just a few lines of code and the challenge is to do cool things with this size restriction What might come out of that is like this year like a 4k demo from the evoke Which contains rather interesting complex geometries. Let's look into this. You will see those haze effects fork effects Interesting line of sight for the camera. You are flying through a fractal geometry But let's skip ahead. There are things moving in this world And oh by the way, there's music in there. That's also been generated by the code in this moment It's all running live So there's the surface is kind of shiny and I was sitting there Mouth the gape and thought how are they doing this with such little code generating such complex Shows so I looked into it on the way back on the train and what I came across was fragment shaders This is an open GL thing which is about writing a rather little a rather small program and That is executed for each pixel on the screen so you Put in the coordinates of the pixels and the program Calculates where the but the program ever the pixel have to go So if you've got to if you do this correctly, this can be highly paralyzed on your graphics card So that's what this technology is for So I found this book online the book of shaders Which calls itself a gentle step-by-step guide through the abstract and complex universe of fragment shaders I can highly recommend that Which takes you by the hand and shows you how everything is done and how to generate more complex geometries But this only generates two two dimensional graphics, so you cannot create the demos that we've just seen So I went on and found some other ways some other ways to generate 3d graphics What I've heard of already in university, but some of you have may have heard already also is ray tracing So if you want to generate a picture that's that consists of pixels You shoot a ray of light from the camera to each part of the scene where there might be little balls or more complex geometries and you look at where this ray from the camera Comes upon an object and then you can calculate. Where is this object? Are there shadows? it's this light in a particular way and You end up with pixels. That's a pretty cool way to render 3-dimensional pictures for example glass objects or soft shadow effects That's a technique that's been around for quite a while but it's got a problem because If you've got a more complex geometry those Intersections are quite quite hard to find out So that's that's not really feasible for live demos so what we're doing here is a Technique called ray marching. I've never heard of that before So what you're doing is to define a mathematical function and for each point in the room It gives you back the distance to the next point to the next surface In this example, it's two-dimensional, but it also works in three dimensions So you insert some point and the function returns you that the next point might be one meter away And if you shoot a ray in there and the next surface is one meter away So it is somewhere in a sphere with one meter measure one one meter radius So I can go one more meter without colliding with anything so I move on and Execute the function for the Next option And it's half meter away and then I repeat that until I know I'm there and this is stemming the 2D case Where you start it with P zero and then you get the distance Which corresponds to the radius of this circle and you repeat this until you've actually hit some surface and you can also do this in 3D and Through this very elegantly find some kind of way to phrase this geometry Which you saw so describing parts of the room and periodically repeating them for example, and then you can just repeat this function as well And if you can do this very well, then you can do cool shit like that So beginning of last year. I also did a small workshop to that So we're going to link the slides later if you want to take a look and I'm also going to demonstrate how that works and Start with the 2D case first how you draw individual shapes like defining circles for example and Ascribe colors to them and then later you can have different 2D shapes and kind of merge them Let's see if I can find a relevant location like this one So this is also very easy with this distant function to define this so you have the centers of the two circles and then Have this more complex kind of bouncy or silly putty Surface and then from there you can progress to the 3D space and do pretty much the same thing To have like a cube here with round corners and then Kind of build some surface and give it waves maybe and then at the end you Repeat this very often and then you have this flight through a complex 3D world with like I don't know maybe twice as much code as you can see there To write this and so both fragment shaders and this ray marching are two Approaches to do graphics, which I didn't know before and which kind of blew me away And I thought those might interest you as well if you can do this really well then something like this might happen So this is a person called Iku And does this for many many years and I'll jump ahead to the end So this is a video where he explains how he works and how these demos work how his demos work and So you can see this is almost six hours. It's not a trivial program So I would like let's see if I can see an animated version of it so it's a kind of small cute figure jumping through a landscape and The ground is moving and the figure is moving and so Iku did the same thing defining a function Which gives you like the distance to the body of a small the body of the figure And that's how you can read me this and in general this person also has a website with very very good resources to this subject All right, so much for that Some years ago. I started to draw Small picture every day and publish it again And I called them schnipsel snippets small snippets That are created every day And I just wanted to tell you about this a little bit. I've I can't recall really why I started with this Was a bit about reflecting a bit better about what you're doing every day You might feel the same as me I'm I tend to start projects and but not finish them really and take them up again a year later and I had the feeling that I Lose the overview about what I'm doing every day So I found it quite good to reflect at the end of each day What was the most important point in the past 24 hours and since I like drawing and During studying computer science. I didn't draw as much anymore. I thought I just Draw snippets was important for me to finish those rather quickly I did a little screencast on Twitter lately how I'd go about drawing this So this has to go quickly just scribbling. I don't take much time Just takes about 10 minutes, maybe quarter an hour And I've got a thing that that's got a Integrated digitizer so I can do this digitally. I don't need any paper. I just try about I Don't even use colors I I found that I'm really bad at colorizing. So this is really sufficient just working in Black and white with a little shading some shading in here The resolution is just 250 by 250 pixels which restricts you kind of a bit Because you cannot really go into details retrospectively, I would have liked to have some some Bigger printouts of that but for that the resolution is a little bit too too small And I started posting some of the snippets on Twitter because I Post those to my website and I thought nobody's looking at my website And I thought would somebody look at those snippets if I post them on Twitter and People thought yeah, they would look at that. So I started Posting the other my Twitter account and I got quite positive feedback for that Because from from time to time I didn't have time to to do us daily snippet and People wrote me. Oh, what a pity that you're not drawing because I'm really looking forward to your daily snippet Linry tried that too one for one month You can look at that. We'll just click on that so we can Size it up you that is in a circular way and you also use taxigons at one point We're playing and experimenting with formats Blinry Aldrich also did a monthly review. That's all drawn by hand and I find this rather aesthetic And as I said, I had this running on a bot and a short while ago I Integrated another API and Which didn't handle one kind of error. I went to bed and the bot ran amok and posted the same picture Every ten minutes and the morning you wake up and look something like this here Because my inbox was full. Oh my oh my god your box running or bosses are running amok my whole timeline is full of that and Some some post below that ha she's still sleeping and the longest been long and this was even a day where I slept in until 10 and it wasn't so good, but The cool thing is if you if you draw a snippet every day You can do a little intro like this With a bot holding a snippet. I like this schnipsel a lot And this is what I show now I would recommend to everybody of you to introduce a reflection technique like that because you've got days like here at Congress. There's thousands of Occasions So you have lots of things going on and you might but still find positive things In a day and I can still find I can remember much better what happened a month or two months ago Because I still know which snippet I drew on that day So try that out who knows if you draw schnipsel or maybe you'll just write a word a day But I think that's a very nice technology to See what kind of things you're actually doing Okay, next I want to tell you about mathematical paradoxes, which I find really interesting For example, there's something called the interesting number paradox which says Every number is interesting Which is a bit surprising to me at least so let's prove this together So it's a proof which goes through reductio ad absurdum. So we take an assumption So let's assume there are uninteresting natural numbers You usually limit this to natural numbers. So not negative whole numbers But numbers which which you can count and let's say there are some which are uninteresting so you can imagine the number line and There are the ones that are uninteresting are marked and Now we can look at this and see there's one of those uninteresting numbers, which is the smallest one Which is the one that's the most on the left of this line. So that's kind of weird It's a number which is simultaneously not interesting But this property of being the smallest uninteresting number is of course totally interesting so there's a contradiction there and If that happens that means the Initial assumption was wrong So this shows that the Assumption can't be right. We assumed that there are uninteresting numbers and the result is every natural number is interesting So that's a proof. It sounds a bit weird, but there's no trick floor here There are some proofs with which you can deduce that bond equals to or something That's rubbish, of course, but this is a pretty solid proof It's not meant completely seriously, of course, but it holds up to critical scrutiny at least and Well, I also wanted to say there are also people who Try to find this smallest uninteresting number and they use different criteria There's this online encyclopedia of integer sequences and they look for the smallest number Which is not in there for example or the smallest number which does not have a Wikipedia article of its own So I checked recently. That's currently the 262 is that interesting. I don't know there's nothing not much to tell about it so far the next paradox on is Something about this geometrical object, which is called Gabriel's horn So this surface is defined such that You take this function one divided by x which is this red number here So these boxes are one unit long and we look at a certain area So we cut this off at position one and look at the part to the right and We take this and rotate it around the x-axis So that's how you construct this object So what we get is this very very pointy Going to infinity object which kind of looks like horn and it has a very interesting property Which is the surface area of this object is infinite Which kind of makes sense because it goes infinitely far to the right and If you try to paint with color you would need infinitely much paint because it never runs out but if you put it Rotate by 90 degrees and try to measure the volume by pouring color into it you will determine that the Horn is actually full the volume is finite, which is a very interesting combination So if you fill it with paint within You can say it's kind of touching the whole surface but if we paint it it's still not enough and Well this paradox with the paint that sounds a bit weird, but you can resolve this by Imagine we don't paint it from outside. We paint it from the inside Then eventually as we go to the right into the point we would get this area where the color Layer is so thick that it doesn't even fit because the color they would also have to become Infinitely thin and if you calculate it like that, it's again finite So this kind of filling with color works, but painting it on the outside needs infinitely much paint You can resolve it like that and also this property of the Area being infinite and the volume being finite you can calculate this You can write down the integrals with this function like 1 over x integrate over the duration length and Calculate what the area is and you see it's infinite and can get the volume and it's finite And it's not really intuitive and I could show you the formulas and you might believe it or not but what I also found is Comparison which kind of makes this easier for you. So if you imagine you have a piece of city putty and you form this to a Little worm or a cylinder So cylindrical shape and then you look at the surface of that thing then the surface is going to be roughly the Circumference of the form of the cylinder times the length So if you like if you put paper around it if you roll it up Then you have this rectangle and one of the links is the Circumference and the other one is the length of this snake or worm and now we take this city putty and we Roll it up. So it's a bit thinner and what happens is the height is halved, which means the circumference is also halved but the Area cross area Decreases by a quarter But because you haven't changed the volume because you haven't taken anything away. It's going to be four times as long So what did we do? We have four times the length and we have half the circumference So the surface of this thing grew by two While the volumes stay constant so that means the more thin you make this snake or worm the more this Ratio of volume and surface changes and that's exactly what happens with this object that you make it thinner infinitely much and so this ratio also grows and that's how this weird property happens and Similar to that also with infinity is the coastline paradox so that's about measuring the length of Codesline or the borders of a country and That depends a lot on how exact you do that how accurate you are so imagine you have a very long ruler and you put this against the Border and you go around and you get a number and you could say that's the length and they could someone could come and say You didn't measure this Sufficiently exactly you would have to use a shorter ruler and then you can go into like the base So that's like the difference between the left and the right example where the right one uses a shorter ruler and use measures this The air the border more accurately and it depends on that What number you get for the border of this country and you can continue this as far as you want So if you're at the beach of this country and you want to measure the length of the coastline of the line between the beach and the water you could use a 30 centimeter ruler and Lay it down and lay it down again and say okay. That's 60 centimeters and then some some life form comes along which is tiny and says you should have measured around each sand corn and Measure the gaps and you should have gotten a much longer length and you can continue this as far as you want because the sand grains maybe have little irregularities and if you get down to the atomic level it gets a bit weird, but the concept of the length of a coastline or Border of a country is not well-defined. So mathematicians say that This shape has a fractal dimension larger than one So the more accurately you measure it the more longer it gets and I don't find that especially intuitive and The final paradox I have is the birthday paradox So some of you might know this So basically it's about if I look around the room here and look just at the right side You might be around 30 people I would guess and the question of this paradox is how likely is it that two of you have The same birthday the exact same birthday and if you want and if you don't know this yet You could try to estimate just in percent how likely that would be do you have any ideas? so let's assume it's 30 people here should be pretty high someone's saying it's high and So About 50% well I Introduce this as a production of course. So you might expect that something unexpected. So if you're naive you might think So there's 365 people and for some people to have a match shouldn't be that likely But if you add one person to a group you have a lot more pairs All of a sudden so this probability Grows larger than you might faster than you might think so if you draw the graph like this And the x-axis is the number of people and on the y-axis the probability that two people have a birthday on the same day And with 30 people we're well above 50% already so like 65% of something which I find surprisingly high and So 23 people is the limit where the probability is already larger than 50% So if you want, I know if you want you can try See if that actually the case in your group and try to find the people who have birthday on the same day Like group by month or something. We're not going to orchestrate that but I find this result interesting last summer before camp I wanted to build a fancy chair and I Thought you should have a Nice pattern and I looked around on the internet and found an interesting Paper it's called modeling and visualization of leaf ination patterns. It sounds a bit confused But it is about the way veins grow in leaves And they describe an algorithm algorithm that's fairly simple And I want to show it to you because it gives really great results This is the kind of result they achieve with it in example, they Let the veins grew rather sparingly, but they become more and more complex and Eventually they start closing this kind of loops and with the Parameters you see in G and H you actually see these kind of larger or smaller cells appearing and it's not actually this difficult they Assumed that the shape of the leaf is given and The vein that has grown so far is represented by these black points and the red points red dots are or since they are the chemicals that Determine growth in a leaf and the veins want to grow towards them so you Place these in the leaf and then you find the closest vein point to a given auxine point and each of these connections Determines or the direction in which the vein is going to grow so The point up here is influenced by two points, so The mean of these It's actually always influenced by three So it's going to Grow towards the mean of these That's this dot here. It's placed and in the next step you Check this radius, which is determined by a parameter and if there's a vein point within that then the auxine is Destroyed it's basically eaten by the veins and in the next step the leaf has grown and you scale the shape of the leaf up a bit Place new auxine points They initially place randomly, but then you check if two of them are too close to each other There's another parameter for that and if they're too close to each other you You You delete it. It's a bit like croissant disc sampling and then the whole thing starts from the beginning You check how many points there are and which of them influence the veins they can eat these The veins can eat this auxine and then you see these branches growing, so I implemented that and It looks like this for example if it eventually decides to start playing and What you can see here is that the veins Scale up as well. These are two modes. They Have in their simulation and one of them they scale up as well and In another they don't and depending on that you get these lightning shapes or you get these highly branching shapes. I am implemented that it's not it's not a big deal and You see them in different colors to make it easier to see where these Segments are and in the final step I can calculate the size of these veins So if they're more highly branched and of course they're going to be thicker because there's more juice running through them and towards the end they will get thinner and These are results they published in the paper as well I found them interesting on the left they have a photograph and on the right the rendered model bend based on the same Modus and you can see how similar the results results are. I think they simulated it fairly well and It was supposed to be a chair. So I implemented it and Put it on my CNC It's a one I built myself. So it takes a while to Machine these and it took several hours and I needed two of these parts to create entire chair So it ended up looking like this. I built these squares or I added these squares by hand It's where the legs of the chair are going to be attached And I closed the veins around them to make it more stable and this is what the result looked like The files are online on my website if in case you want to build this kind of chair But what I found out at the end if you have this kind of Vane growth then you maximize the surface and this means that the Using sandpaper on this is going to take a very long while But it turned into a nice chair. It was a fun project for the summer summer for generative art Do you mind if we go over time a bit all right Hi, thanks. I would like to tell you the story of the illegal primes and all this has to do with DVDs We've got a few young people in the audience somebody of you who of you had a DVD in their hands and Who of you didn't have a DVD? Oh, nobody that's interesting. So we're all on the same page here. So DVDs You might have come across the situation that you had a DVD and wanted to make a safety copy of Your DVD and the program of your choice gave you this error message here that the DVD is copy-protected and people thought about that problem so that people could not just copy any content and they came up with a With a procedure that's called content scramble sequence that's that works with secret algorithms and scrambles the information up according to a complicated algorithm that I have not looked into and They do have They gave the information on how to encrypt this information to the vendors that provide the DVDs and The instructions for decrypting to the manufacturers of playback devices and then they hoped for a while that that people would Not do bad stuff with this for a while And then people sat down and try to reverse engineer a DVD player how this Thing works and what the keys are and they re-evaluate this in C CSS de-scrambling so this is the main function of this procedure and Around it in the source for there are some longer tables of numbers Which are here called CST to and something and there are some lookups to those but this is the core of it and Well, other people looked at this and wrote a graphically user interface that made it very easy for people to copy DVDs And so some people did not like this So there were several court cases against the person who wrote the GUI which was a Norwegian I believe and Well, it became tricky What exactly you want to make illegal so how the algorithm works or the keys or what? and then people Actually became more creative Where the border is between things that should be protected and not and for example They produce DCSS the movie So that's kind of the title scroll off the source code Scrolling through dramatic music through space like Star Wars So is this an artistic project which is should be protected or is this illegal or people Did a dramatic reading of the source code? I thought we might listen a bit to this. That's seven minutes So let's just listen for the first seconds And so on so you get the idea What else happened People wrote high coups So not just one but like I think 600 of them or so which describe how the algorithm works and So the keys are also part of those so this is a core part of it So, you know high coups five syllables several seven syllables five syllables So it's a Japanese accessible form of poetry and you can adhere strictly to that if you enjoy that And so someone wrote all we have to do is this copy our D key into I am one use the rule above that they Crypts a disk key with I am one and it's friend I am two as input thus we decrypt the disk key. I am one And so on over many many pages. Is this art? Is this a program is this little algorithm? Who knows and And people also printed this onto shirts and onto ties and started wearing this and stuff and a mathematician sat down and Determined this number which is interesting because if you write it in hex decimal notation, then it's a zip file which Contains this C program And it's a prime which might makes it so interesting. So the trick With which it was found is so with a zip file You can add null bytes at the end as if much as you want without changing meaning of the file. So the they abuse this to With kind of brute force and some rule like what are prime candidates and then checks out of them are they actually Primes and with one number of their luck and had this prime which might now be an illegal prime so this was the first one you found and Later also found one that's approximately twice as long and that one's interesting because at that point And the point that where he did this which was shortly after the year 2000 or something It was one of the ten largest prime numbers which were known at all which made it just through this fact already interesting for Publication and it ended up in high score lists of large crimes and ended up on various websites And it's a very nice hack to disseminate this information Finally a little bit of art art from the AI Field I Don't want to talk about neural nets, but just give you a little bit of feeling for what's happening there This is a process or So that's one part of the network where you can throw in an image and so Let's we end up in a space which is called Latin space Which is a kind of very basic representation of images and I can reverse this to get the original image back And I do this in the training phase So I throw in an image and it process it Reconstructs it and I look at the difference and can determine Learn from this and optimize the procedure But what I can also do is not throw in an image, but directly go into this Latin space and Pick a vector in that Latin space a random one and let it reconstruct that one and then let's see what it reconstructs so those are the procedures used there and all I can also do is Supposing I have two images here on the left this dog and on the right this is the leopard I think it's a leopard with two weird ears. So Something went wrong with the generation in the model, but what you can see I can create intermediate images, which is funny So what you did in the 90s with morphing software now you can do it with neural networks So I look for these two images in the Latin space and then interpolate between them and then it looks much more interesting than just morphing on picture pixel level because if you Morphe on the pixel level you do a kind of blend from A to B as you know in a video game It's kind of stupid and here we have a kind of On the content level some what correct image and you can render this as a video might look like this And we don't need sound for this not that exciting so the model here is called big gun and So the images were classified in different classes and dogs and flowers But it didn't use a very big didn't use a very national auto coder So that's different neural structures, but they still have this latin space So you can still do this latin this interpolation and this morphing I call it Which happens there looks extremely fascinating and what you can also do all of a sudden is Vector mathematics in this space so you can check other special vectors where I can Which you can move around this point where things happen or I can look at points in this space and cluster them and say for example here we have a clustering of Ladies who are smiling and ladies who are looking neutral and now I can say I take the one point and Subtract it from the other one. So smiling woman minus neutral woman becomes neutral man Because the woman attribute has been subtracted away, but I can also at the bottom Do that with sunglasses or with glasses and say men with glasses minus when without glasses plus wound with glasses That's moon with glasses because I've also subtracted this away and you can have a lot of fun with this and Because earlier we had it with Twitter bots. There's one more here, which is called the smiling vector bot which looks for images and Looks for the matching image in Latin space and then adds a smile modifier or smile vector or sometimes subtracts it depends and The results are often kind of creepy, but you can scroll through them. So let's click the whole account Maybe and see what it posted recently oops Okay So it also posts regularly. I don't know where it gets the images from Sometimes also those videos and morphs back and forth there So what you might have seen as well is deep dream. So that's kind of Over optimization of an image so you can see here. I think we should also disable sound It's a bit weird. So here it was done on the video. It's very interesting And you can already see this was this neural network was trained on images of dogs so now it thinks of dogs everywhere and I don't know I Have not tried LSD does an LSD trip look like this. I would imagine it looked like this I would say so it's very trippy and very abstract and it's very interesting to see Where faces and patterns appear and in this video, you can also see that it's very consistent So if you do a pan of the camera, then a dog stays in the same place With the first attempts it didn't work like this if you when you try to apply neural nets to videos You had to do some tricks that it stays consistent. Otherwise, it would wobble around way too much Oh, yeah, and something I found a few days ago, which is also this kind of interpolation But only on cat images. I also thought this was very nice. I also always feel like you can see How their positions change like especially if the body starts wobbling That's those different foot positions with the cat can have which are and it starts permuting through those And I can generally recommend this Twitter card rolled runner zero one. It's kind of doesn't say much but Experience a lot with neural networks and post videos and they're very interesting and we were at art So I want to show you more art projects, which I found very impressive in recently So this is from Helena Serrin. She does a lot of things and it's very impressive That the data sets with which she trains her neural nets They're always always her own pictures and drawings You don't see that very often because people scrape data sets from anywhere or take large picture sets and she does this with her own data and then I think she mainly use guns and then Generates very abstract images. Most of those projects are called gun reaver So they kind of go together like a puzzle. I find that very interesting and in this case I think yeah, it says that here that was a data set from different things with flowering trees and with book pages containing high crews another nice project is from Tom White Who exploits? Neural networks or image recognition software and created this interesting architecture to Create these abstract images which can still be recognized from image recognition software So, I don't know. Can you guess maybe what this thing in the middle upper middle is what this looks like? Louder a plane. Yes, it could be a plane, but it's something different That is based on the model of a shark trained with a model of a shark Something else that you can guess the dark blue thing here at the bottom. Yes, that's a hairdryer Another thing I remember this was cabbage and this down here was jack o lantern. So yeah, Halloween jack o lantern And I think it's interesting because this is very abstract, but as a human you can still kind of recognize what it represents and He experiments in this direction as well a lot Oh, that was too much We're still here. So three small things at the end from Duo of artists from South Korea, and I'll try to pronounce it correctly Shin Song-bak Kim Yong-hoon are they called and I hope they don't mind this So I want to show three show projects The first one is non-facial mirror, which is a mirror as you can see soon with facial recognition Which doesn't want you to look at it at yourself in it Yeah, so does it rotate away from the person? I think it's the other way around track someone so that's the Other way around if you couldn't hear someone said in the audience we have an installation like that here That's right. I walked past that I think that's very nice The examples I have now are all with facial recognition, but they can do other things as well So here for example, they put a camera with facial recognition pointed into the clouds So you might play a game what can you recognize in the clouds and they always remember saved the image when somebody was recognizing the clouds and printed it and it's nice because you can often see faces there and The final project is also very interesting So they asked other artists to paint a portrait and the challenge was that a camera is pointed at the canvas from the top with facial recognition and The goal was that you cannot recognize it as a person Which is not that easy and the artist Has a disk and laptop next to them and could see what was recognized how and there were different algorithms there So you can always see like in the green red green blue frame there were three different algorithms trying to recognize faces and the artist is Suddenly trying to place eyes differently and sometimes they just put the lines straight through it So you out of frustration that they're still a face recognizable and they did this with a lot of artists They all got the same portrait and they're very interesting Paintings that happened out of that All right That for the subjects we brought you today. I hope that was fun We have the slides under this URL and They're also links to the previous versions of the two previous versions of this format if you find that interesting Yeah, otherwise talk to us if you see us walking around we are on mastodon and Twitter and Thanks for being here and have a nice Congress