 Our next talk is on the Axiom open source video camera from Rasmussen Pickelhofer. Perfect, perfect, perfect. Excellent, so please give him a warm welcome. Thanks so much. So what I'm going to do now is take you on a journey into filmmaking, a bit away from coding and all the technical details, into the world of motion picture art, you know, the dream factory of Hollywood. That's where we're going to dive into, right? Axiom, that's a worldwide community, a product, a kind of movement over time. The goal is to create a professional film production tool, right? So we want to actually make films. I want to make films. I never wanted to make a camera, but some things led to the other. And that's where we are currently. We have high end requirements. So it's really about high end cinematography. And why do we need another camera? I mean, aren't there plenty of options out there already? What you see on the left hand side, on the left hand side, I'm not going to point to the screen, is a camera from the 60s or the 50s, a 35 millimeter analog film camera. And on the right hand side, you have what's a modern kind of representation of a digital cinema camera, Sony F65. And the paradox situation is that even though the technology itself and the quality and everything, no matter how you look at it, it has evolved. Of course, it increased performance, the image quality and everything. But the paradox situation is that the accessibility and the way you can actually influence the images has quite decreased in contrast to the technology, right? So where the old analog camera was kind of hands on, it was all mechanically, you can think of it with the motors and everything is like open the box and you have the mechanics inside, easy. What you have on the right hand side is more like a black box, a computer, no idea what's happening inside, only Sony knows. And you can push the buttons and see what happens. But that's like all you get. So what is the Pertos project? What is the Axiom project really about? And I think there's one quote that summarizes it better than anything I can come up with or anything I could ever say. And it's a quote by Lee Parker, an IMAX cinematographer. And he said, I grew up as a photographer in the dark room and I missed the intimacy of watching the image slowly appear on the paper in the rate of the safe flight. The beauty of an open source camera to me is a step backwards towards the dark room in which making the tools is part of the joy of making the art. So that's why we're here now. So how did it all start? So how did we take the first steps? This is the so-called shoe box prototype, a proof of concept, large cumbersome strange form factor, a bit of wood inside, very transparent enclosure. And it, well, economics wasn't really the focus at the time, but it worked. It actually created a prototype that output images and that we could use and to shoot some sample footage and to actually kind of show the world that it's possible to build a camera like this. And then we took it one step further because we knew that this is something we couldn't do alone, we couldn't do it in a small team, we couldn't do it in our garages, we needed to scale it up. So we ran a crowdfunding campaign to build the next generation, the XM-Beta, in a way more compact, more modular, easier to produce so that we can actually give it out to people, that people could build it themselves, we could ship it to them, something that's like available, accessible and ready for production, basically. And the result of this is the XM-Beta developer kit. It's basically the entire electronics, the complete stack of functional hardware inside without any enclosures or more accessible. You can modify it more easily, no protection for outside filming in harsh conditions, but for development that's perfect. That developer kit. This currently retails for around 4,000 euros, so you can actually buy it from us. Sounds like a lot of money, but if you compare it to available high-end industrial cameras or cinema cameras, it's a bargain, right? It's all high-end. We looked at image sensors because that's like the heart of every camera, obviously. The plane that transfers light into digital image in the end. And we looked at the market and there weren't too many options to actually get as a normal person if you're not so neo-canon. And then we chose the one that was in terms of quality, in terms of performance, the best we could get, and of course it's not very cheap, so that also drives the price, obviously. The next iteration with the prototype being more for the lab and for development is what we now have here as first time prototype for about one or two weeks now. The XM-Beta Compact. It has a full metal enclosure, all CNC-made. It's a bit heavy. I mean, you can carry it afterwards a bit. It's not as heavy when you hold it in hand, so it's quite ergonomic, but it's a very solid shell, a very robust kind of enclosure and turns the hardware into an actual camera. It actually looks like a camera already now. Let me take off the camera. We don't know yet how much it will retail if the electronics are 4,000 already, probably in the range of 5,000 to 6,000 euros, we'll see. And this is the first prototype, but it will still take us, I don't know, half a year or so to actually get this production ready. We still have a lot of relation and testing to kind of take it into the field and actually see how it performs in terms of ergonomics and so on. The big picture or the long-term goal is something even bigger, where this is kind of reduced the camera to the essential components that you need to actually record images. And the X and Beta Extended, as we have the concept currently, is more like the full feature of everything you could want or need in the future, so that's the long-term goal. It's a concept, it's not a reality yet. But you can see that some elements from the front don't look that much different in the concept. So how do you want to control it? Currently, quite easily, I think this is the audience, this will properly appreciate the way it's controlled through a Linux terminal. Yes. Filmmakers, I don't know why, but they're kind of not really into Linux terminal applications and so on. They always look strange when we tell them how to connect and login and so on. So we came up with a physical device called the XM Remote, which connects to the camera or pretty much anything you want over USB, which features a small display, a couple of buttons, styles, everything that provides a kind of haptic experience for controlling camera in a sophisticated way. Currently, the way we built the camera, we've shipped around 50 developer kits so far now, is completely by hand. So actually taking every electronics component, putting them on the boards, soldering, it's very cumbersome. Unfortunately, it's very labor intensive, but I hope that will improve in the near future with a fully automated industrial process. But our approach is also that we kind of want to keep innovating and every time a board or electronics gets improved, there's a new iteration. We want to move that into production immediately. So it's definitely not a mass production product. It's definitely not going to be produced in a number of thousands or so. So that is the kind of the challenge for us. How can we always bring the latest innovation, the latest innovations, iterations into the actual production process? Because obviously, high prices and low volume and making it cheaper, but always keeping innovations concurrent, it's a bit of a challenge. I don't want to bore you with too many technical details, but the heart, the image sensor, it's a 12 megapixel sensor, so we have 4K resolution by 3K in the height, so it's 4x3 aspect ratio for anamorphic shooting, for example. Anamorphic. 12-bit color. It has a global shutter and can do up to over 300 frames per second at full resolution. There's a monochrome version available with the same device color version and the near-infrared version as well. And that's currently the way we actually produce hardware. Some of the boards, I have some stuff here, if you want to take a look later, have over 500 components on them. So, yeah. Lots of fun. Who wants to build hardware for us? Yeah, excellent. We'll all bring you into the lab afterwards. And yeah, that's the process of actually assembling the previous prototype with the mechanical parts. And the result is like this, all now with the top handle and the shell and the wooden grip around here, added as well in the second generation. Now, beside Hollywood or beside the filmmakers, who would want to use this? Well, there are quite a few people who do high-end imaging, machine vision, scientific imaging, and having a completely open and adaptable, extendable platform available is quite interesting from what I heard. So, there's a company, for example, or not a company, it's a university. They're astrophysical laboratories in Marseille. They use the XMP to develop their own curved image sensor based on the image sensor we are using currently. There's a company in San Francisco that created a system using three of the cameras that have set near-infrared sensors to map agricultural land because plants reflect a lot of infrared light. For development, there's something new as well. They're called XM micro developed by two German students. And they met at CCC, I think, last year or two years ago and said, I want to develop on this, but I can't afford it. So what do we do? Yes, we build our own camera, of course. And the XM micro is, in that sense, very much reduced in complexity in terms of cost. It has a small image sensor. It has only one piece of hardware attached to the main processing of the shelf board. In contrast here, we have a stack of five PCBs. So everything's kind of reduced to the sensuals. And that way, the bill of materials comes down to around 200 euros only. It's not available for sale. You actually have to build it yourself currently. But all the files are, of course, available online, open hardware and free software. We're doing Google Summer of Code now for a couple of years. So if you want to join us for this year, please do. The tasks revolve around FPGA development, BHDL, embedded Linux, running on the camera, CCC++. So if you're a student, then that would be much appreciated. If you're not a student, no problem. We have plenty of things to do. It's a very diverse community of artists, software developers, filmmakers, cinematographers, hardware, mechanical, optical engineers, everything. So you would be much welcome and very much appreciated if you want to join us. And this slide reminds me, what would a camera be without actually showing you what it can shoot? So if a bit of luck, I will now manage to move this video, which should load in an instant over there. No, it can't. There it is. It has audio, but that you have to imagine. I'll turn it up very loud from here. It's actually audio. Don't worry. Also forgot. It's just music, so... And it's gone. No, but it's going to be on your video. Excellent. You will suffer for posterity. So yeah, just look at pictures. On this huge cinema screen here. It's just to be... Yeah, why not? Two minutes long. But it shows the image sense in all its glory. Yeah, that's it from my side. Many thanks. More details on the website. We will take our babies outside after the talk and you can actually hold it if you want. I don't know if we move to the tables with the job offers or so, but meet outside at first. Are there any questions? Yes. I want to have first the MCCC or the golf recording on the next camera. Yes, I remember to repeat the question. When will our cameras actually record here? I have good news and bad news. We actually met the people from the video team before and who run the network and the streams and they were very interested in using open hardware, open source camera. The bad news is that doesn't alone solve the issue so we will still have to build a lot of cameras and they will have to build or buy them or build their own version for lecture recording. Why not? But it's not something that will happen immediately. I'm afraid. But as long as people are interested in doing this, contribute and push it forward, definitely. How do we... Yes, very good question. How do we deal with sensor manufacturers? There are not that many, so indeed it's a difficult choice. All the manufacturers who require an NDA to even get the datasheet, not just preliminary information but the datasheet, were already left out. The remaining ones, we benchmark, test the communication, how is the performance, what does the datasheet contain and that is the result of our choice now. Can you view the names? The image sensor is made by a company called CMOSIS which is not in action anymore because it was bought by AMS. So it's now called AMS Image Sensor Belgium, actually. How long does it take us to build one prototype? The electronics, it's like five to six spots in the XM Peter developer kit and we ship one every five weeks roughly. So it's not 10 hours a day every day but still that's the time kind of we need. Currently everything is manual. There is a project already where we try to reverse engineer the lens communication but it's not there yet. Is HDMI involved in your camera? Is HDMI involved in our camera? There is an HDMI output plug-in module at the back indeed. Do you have an HDMI license? It's not actually HDMI, it's just a connector that looks very similar to HDMI and somehow works on some screens that also have HDMI I heard, yes? Please cut that out at the end, yes? What does your image processing pipeline look like? Like your ISP and then how do you get the data out for 4K by 3K and 300 frames per second? How does the image pipeline work or look like? The entire image pipeline all the processing runs in the FPGA in real-time and the advantage we filmmakers kind of have is that we want a very pure and very raw and very unprocessed image so we actually don't do that much with the image. I can show you the link on the website in the wiki we have an actual image of the pipeline with all the processing blocks. Currently over a connector which is at the back is largely compatible with HDMI devices. Currently an SDI plugin module is in development, dual 6G, but it's not finished yet. Yeah, it has 4 connectors and you can kind of dynamically patch them into the FPGA if it should provide sync or timecode or with your output input, yes? A last question. Yes, how do we fund the project? How do we fund the crowdfunding? We got the European Horizon 2020 grant a couple of years ago and now we're kind of bootstrapping our way up by selling developer kits and by that we're again affording a bit of prototyping but currently we have no employees and only the people who actually built the camera in hard labor get any money from it. Thank you. Thanks so much.