 Hello, I'm talking about visual code evaluation. Maybe it's something everyone does, but I thought it would be nice to highlight some aspects of it that I find interesting, and that's me and Daniel. First I want to jump approximately 100 years back to 1915 where a couple of the Gilbriths, Frank Bunker, Gilbrith and his wife, Lillian M. Gilbrith invented what they called Cyclochronographs. It was a pretty cool technique because it was like the beginning of management, the invention of management, basically, and it was about measuring the motions of people. So when someone does a working task, they thought if we optimize this, if we find the ideal way of doing this, it would help a lot. People wouldn't get tired so easily because one could optimize it perfectly. In order to scientifically figure this out, they built some devices and that's one of them. It's one of the first Chrono-Cyclocrafts that they made, but once he's there, it's like, I don't even know what it's called in English, but they're like this sound fork basically, like it makes one sound if you hit it and it's like tuning fork, yeah exactly. So this tuning fork vibrates in the right rhythm and through this it creates a pattern basically. So it's like a timing mechanism. Maybe I'll show you first what it's good for, but so they also invented some variables like a fancy ring back then. This ring has an incandescent light bulb in there, so if you imagine using this hammer and having a photo camera and having long exposure photography, you can generate the motion path directly by just having this long exposure thing. So you see exactly the line, but then they thought, okay, this is maybe not enough, so if they have like the timing information in there, they could also have precisely the duration of how long it takes. So if you, for example, have like a blinking light every second, you would be able to tell, okay, this is after one second, after two seconds or something in the image. So I just wanted to go through this quickly. So they started with a tuning fork, was too expensive for them. They reconfigured like a doorbell, so this mechanism is like from a doorbell. So this was their second iteration. Then they had a third with like a circular motor, basically, and like what they were doing was like simply like trying one iteration after the next and like seeing the results. So they saw that it's like then they figured out if they start with a with a bright flash and then it decays, then you can also like have the direction in there. So they already had like three valuable things in there. And then this kind of a side note, but I found interesting, like in order to illustrate it, they created these wire models. And so this is like a representation of the motion path as a wire model. It's also like stereographic, that's why you have two. And the process of like creating these was, they took these, these pictures from different angles and then a little bit like in a 3D CAD software, they really like someone was like taking this one perspective from the front with a stereo photography, trying to bend the wire in a way that it matched from this perspective. Then they switch perspective, look from the top, try to attach it like to, to do it from there and did this with all the different, different angles. And then they had like a perspective view to check if this matches exactly or not. So this was their process. It's basically the same, but more like a newer thing. Like I found this visualization of like a 3D printer's motion path. So it's someone who just put like an LED onto the printer also does long exposure photography. And I think it's pretty nice because you can like A, you see the line, but B, you can like, without printing any material, you can like do a real life check of and visually check if this is working the proper way and like the way you want it. Now I have to switch to some video. It's not my computer, but I'll figure it out. I'm sure. Oh, different. Where would be that now? Oh, how did I do that? It's just advertising from some company. The last bit was the thing that I'm interested in, like, like this visualization of this motion path and like, I find it really interesting how like there is like a visualization of an algorithm. It's not, it's not explaining how it does it or something. It doesn't unveil like necessarily what the inner workings are, but you're like, you see exactly that it's like far more efficient because it just goes the pathway once and it doesn't go like random and like it's not hit and miss, but it's like there's some plan behind it. And I think it's, it's interesting how it's like presented in this way. It's like a company that is like, it's Nito, I don't know, they do the same like Roomba, like they try to have like a unique selling point basically, like, but they say they're four times faster. So, so that's there. And I saw also a lot of like Roomba visualizations of like Roomba's going through the room. This is like a thing that I did myself. It's like, it's just like a like Piazzo element, one LED stone. And like long exposure photography. And like then when you like, like start scratching this stone, like it generates electricity, this becomes like translated into light. And then you have like some kind of, yeah, offset visualization of the structure, basically, because like, you can see that like, like around the corner, this edges is kind of like creates energy here. It's very, very like flat. So, so it almost didn't do anything. Even though I went over it over and over again. And like, down here, you also have like that thing. And you see that there's like here, there's like this offset of the, of the Piazzo disc. It's like same concept, it's like a microphone, a color LED. So, so as soon as you add like the color dimension, you can like say, okay, like different amplitude is a different color, you go through the room with long time exposure and can try to visualize whatever in this case sound, but it could also be like other things. They're nice examples of like thermal, super cheap thermal cameras. There's like a thermal flashlight, what's a really cool project. You're basically like have one thermal sensor, you have like bright color LEDs, and you just point it into the room, for example, the kitchen space or something, and you just go over it like, and like as soon as it detects like a hot water boiler or something. It's like the color turns red. So then you have like a red spot there and because you're like shining the light on there directly, it's kind of like you have this one to one relation, so you don't need any like further computing or you don't need to like remember how you're moving your hand or like what the device is doing or something. It's just like, yeah, sees from the same direction, sends the way back, basically colors the thing. It's like, yeah, this is just another thing where, where like, I think we all do this probably, like, like just aiming for something and like, like trying, like in the life coding example, you'd like do something, it doesn't quite fit yet, you do it again, you try it again. So it's kind of, that's what I mean with like visual evaluation. So it's, so there's like this constant, like back and forth checking. This was like neon, no, not neon, fluorescent tubes that I basically made like, like flash the light in one direction. So it's like the star shape and it goes like out like this again. And it's like very dependent on the outside temperature. And you just have to like, like play a little bit with the values, but then you can create this effect. And it's like, so it's also this constant feedback loop, basically. It's a little bit of a switch, but like, this was like a kiosk that, that I had like as a breathing thing. So it's like a, like a building that basically does the same as like the Apple sleeping LED or something, but like, like on a bigger scale. So it was like, like this breathing kiosk in Weimar in Germany. It was kind of interesting how it like transforms the whole space because like, yeah, kind of like really, like, if there's like light change over time, it says very like strong, strong visual attraction kind of. So like changing color over time or like brightness over time or light intensity over time is like thing that I got interested in. Like, so this is like basically this actually like the input is or the output is generated from the input in this case from processing. It's like, it's like the motion of a pendulum and it's like when it's when it's on the outer sides, I don't even know how I mapped it in this case, but it's like, I think when it's in the middle and not moving at all, that's black in this case. But then if it goes to the outside, so it gets gradually less and like basically attempts to try what you can do with only like light intensity and time. So now it's, but you can also like, so this would be kind of like a frontal perspective to the pendulum. But you could also imagine the pendulum swinging like this and then if it's far away, it's maybe dark and like if it's very close, it's white. So if it was like this white pendulum, so this would be a different way of mapping it and there's like a different effect probably. Ah, so I also did some like like video tests and there's one really amazing piece of work that's Tony Conrad's The Flicker from 1965 where he basically just was using light and no light as his medium. And so I tried to recreate it basically. I'm sorry, it's not my computer. Here's this warning, epileptic warning in the beginning, but like after that, it's basically really just like rhythmic patterns of like light and no light. You just used like black frames and like bright white frames. And I thought like this is one of the few cases basically where the quality of the movie should improve by compressing it to the maximum because as it's film, it's kind of like it has all these like artifacts in there and so on that probably shouldn't intentionally be there if you're trying to have this like one and zero thing. And this is what also got me to talk about this topic today because I wrote this like bash script. I don't know, it's probably like a stupid way of doing it but like it's like takes separates all images of this movie, puts them through image magic, creates one value between zero and 255 or something and then makes this long list and then from this list I can recreate the video. So the video on the left is like the recreated version of the film on the right and then in the beginning like the wall seemed to work well but then I was like splitting them because it was like getting too big. So they only had like one minute sections or something and so I put them next to each other and that's where you like immediately see if there's like one frame off or something because like you see exactly that the flickering is not in sync and this is something that I think is really good and maybe sometimes there are ways to translate something into such a visual thing in order to just check if the output is correct because like, yeah, looks proper if you just look at it not in parallel but like, yeah, so this is a long thing. There's only excerpt of it but like, so I tried other stuff with the same software like two more things that I would show to you. This is one where like Tato is like a German criminal series or something and like, so even though it's like out of proportion you're kind of like, like instead of using like gray images I now use circles so I put like all the, I created these, generated these circles put them in the folder so for each image it's trying to find like one with a matching, with a matching value and I find personally that this is like a, like really a different way of showing something like in this scene that comes now like this, here and this, yeah, there. It's finding a value now from the size of it. Sorry? How's it finding the value? Ah, it really like just takes each image. So this for example, see, like looks what, what light intensity value the overall image has. So, I don't know, 200 and then there's like this one is also like 200 in total and then makes the match but like here's like this, this car going by and like the second one here there's this, this, this sudden, sudden thing in the circle and I think it's like, like in the circle it's far more apparent that there's something like quick going by or something where, whereas like in the, in the video one cell phone wouldn't notice it. So I think it's like it also changes perspective a little bit and probably finally, of course one can do this also in a different way and this is like far less readable I guess but this is the same image again or the same video snippet but this time on the left it's like other video footage that has the same values but like it's not as apparent because the, there it is kind of but like because the shapes are very different it's kind of like hard to, to recognize it at first but this is kind of like a, like a light intensity equivalent movie. No, no, no, there's like a last slide. Thank you.