 Hello, my name is Richard Hughes. I'm talking about the colourhug spectrum as a type device. I've been working on open source for about, well I guess over a decade now, I work for RedTact and the desktop team and I maintain the colour D, a load of the colour management tools that I'm a colour manager, also a package kit, nomosafware, used to be HAL, bits and pieces of basically up the stack, top to there, top down. And in my spare time I kind of took up photography, I'm not very good photographer, but I didn't notice that all the colours weren't right when I started printing stuff and then went down the rabbit hole and started discovering that there are no open tools to actually create colour profiles to fix this properly. So all kind of went from there. So we invented this thing called colourhug. Colourhug's a tri-stimulus calorimeter, which is good for profiling screens, basically to measure colours accurately on a display technology. So far we've sold 2,300 devices, which is about 2,250 more than we thought we'd sell, and that's been done the last three years. We've been upgrading it, we've been fixing bugs, we've been adding features as the years went through. And we've made a lot. I've made an initial batch of 50 and kind of hoped I hadn't wasted a lot of money, hoped that I could sell maybe the majority of that batch. And I think in the first three weeks of sale we sold 800. So it was a project that sort of massively blew all our expectations and a lot of people have been very happy with. So when I say we made the devices, we literally did. Me and my wife were sitting there making these things by hand, making the tools to punch the holes in the plastic boxes, sticking the stickers on the boxes, trusting the envelopes, packing in DVDs into cases. So we literally did it completely ourselves. So we had no capital to start up with. Many weekends and evenings have been spent making colour hugs. The first thousand we made without making any profit, we had no money for time, we were working for free. Now thankfully a little bit is kind of working out with volume discounts we're getting. But it was kind of a labour of love. Now I guess a confession. This thing about colour hug spectro that I kind of talk about, we can't call it a spectro anymore because I got a very nice legal letter from Spectro GmbH saying you can't use the word spectro. So we're now calling it colour hug plus. So many times I do slip and say the word colour hug spectro, just kind of in your head right down colour hug plus. That's kind of just the word we already had a trademark on in the UK. So it's like an easy way around it. So just when I miss talk, just replace it in your head. So why do we need a new device? Why wasn't the colour hug good enough? Well, for a few decades, our only display technology that most people had was a CRT monitor. And the way a CTM monitor works is you have an electron beam that hits a lot of phosphors. And if your electron beam is turned on when it hits the phosphor, it glows. And there's very few phosphors for this kind of technology that are stable, cheap and easy to handle. So consequently, a lot of CRT monitors actually had the same primaries, the same red, the same green, the same blue. So you could design tools which were targeted for a whole class of equipment. And then something happened. Display manufacturers began selling displays that could display many more colours, much more deeper blues, much more vivid reds, much more natural greens. But in doing so, they changed the phosphors that they were using. Now, it was going to happen in any way with the shift to TFT and then LED. But it's happened now much more quickly. And now displays you can get in the shop are much, much better in terms of colour gamelets than you could have got with a CRT monitor. So the upgraded screen technology kind of drives us to do something slightly different. Colour hug one tried to solve this problem using correction matrices, which you could apply to a device. But to create the correction metrics, you also need to have a very expensive photospectrometer. So you're kind of in a situation to buy a cheap colour meter. You also have to buy an expensive spectrometer to get an accurate result. So as time goes on, colour meters, targeted specific primaries begin to stop working so well. This is a kind of a good example. So you can see this is the spectral output of different screens. So for a long, long time like the iPad 2, Motorola, Zoom, et cetera, we're all very same thing. So the same kind of blue, the same kind of reds, et cetera. And then OLED came along. And OLEDs are very vibrant, very much wider colour gam... a range of colours that you can see. And really if you try and use an original colour hug, or it's not matter of Huey or DTP 94 or something on one of these new types of display, you just get completely the wrong answer. You get completely invalid readings. So really we need to go back to the physics and try and think how can we fix this with all these different display technologies in existence. Now colour gigs kind of recoil in horror now. But this is my way of kind of showing the horseshoe is all the colours that you can see and the triangles are all the colours that the display can display. And so you can see that the difference in colours isn't slight, it's dramatic. So with the advent of OLED, it's really increasing number of colours that we can display by a long, long way. So how do we solve this? Well, we can spend millions of pounds and do a massively parallel spectrometer type device. So these, I think this is one from the NPL. Millions of pounds, you squirt photons into a fibiotic, it splits them apart, hit a diffraction grating and you measure what the frequencies that come off it. Kind of impractical to sit in your design room. But you can't get lab equipment. Now this is, I think it's about €35,000 worth of lab equipment, which is basically doing the same thing slower on the bench top. Very capable piece of equipment, but its price kind of takes out of the range of most people in this room. So we need a much sort of smaller and cheaper version of this. The maybe isn't quite so accurate or quite so precise, but still does a good enough job so we can create an ICC profile. Then X-Rite really bust open a market. They produce this thing called a colour monkey, which it lets you profile printers and displays all for about €300. And it's a really amazing piece of kit. If you take one apart, obviously you void your warranty, but if you take one apart you see the design and the plastics. It's all a very high-end piece of kit. Now we actually broke one down and stuck it under lots of microscopes and stuff. Between me and a different consulting company, we worked up, they'd spent several million dollars developing the spectra unit inside this unit. So they've got obviously a very big R&D budget and they've produced something that is technically very good. Not without its drawbacks. The main drawback I guess from my point of view is we talked this morning with Chris about additives you put into paper, which make the paper fluoresce. Now this can't measure the fluoresce in the whitening additives in paper. And so you can get a very odd result when you try using it with UV light sources without. So it's an awesome piece of kit for the price, but it's still not without a fault. And then there's the people that say, well you can make a spectroscope using a bit of broken DVD and a defocussed webcam. Yes, yes you can. You really can. If you want to explain to kids how diffraction gratings work and how a microspectroscope might work, it's a really good way of showing kids. But it's neither accurate nor precise. When you're talking about nanometers being the difference between an accurate result and a non-accurate result, just the glue you're using to stick the DVD onto the camera will change it by 10 nanometers on a warm day to a cold day. So you have all these thermal effects and all these non-linear effects and the fact that the DVD is curved. You've got all these so many unknowns. It's a miracle actually if it works at all, let alone you should be able to take an accurate reading from. So, yeah, a cute trick but not so much for creating ICC profiles or having any confidence in what you're doing. Also this doesn't really have an illuminant so you can't shine a light on something and get some sort of spectral data back. Another issue with a lot of equipment including to some extent the colour monkey is that not all monitors are producing a nice sort of outwaves. So the blue primary is very clear. The green primary is very clear. But if you try and sample this at say 5 or 10 nanometers resolution you're going to miss a lot of the spikes which means that the profile you get isn't accurate to actually what you see in reality. So you have to actually have a piece of quite precise accurate equipment to produce a profile that's actually worth using, let alone any good. Also people nowadays are buying high end printers with high end ink and they want to be able to profile what they're producing as well as what they're seeing on screen. So they want to be able to soft proof and they also want to be checking that what they send to the printer is actually what they're going to see when the print run is done. But with printing it's not emitting light. So you have the issue that the ambient light is affecting the colour you see of the ink. So for something like a sodium vapour light obviously you've seen everything's yellow. With LED light it's much more like daylight. Still spiky but much more kind of the colours you expect. So one of the solutions for this I've found is that if you create a spectroscope with a wide band illuminant and a UV switched illuminant at the same time you can mitigate some of these effects just from a couple of dollars worth of equipment. So I have not the experience nor the optical ability to design something that's thermally stable, accurate and precise enough for the results that I need. So I emailed and spoke on the phone with about maybe 10, maybe a dozen companies. Getting some quotes back. The most expensive quote back for the NRE the non-refundable engineering costs was I think I can't remember the name of the company but it was €350,000. So I said no. Most of the companies refused to be it at all and this was the only company that gave me any quote. The initial quote was for NRE €50,000 but we agreed that it was an open hardware project they would absorb four-fifths of the cost of any product on the assumption it would be not tied to me as a manufacturer which kind of makes sense from an open hardware point of view too. Now I think that's an awful lot of money. I don't have that much money. My wife says that's the price of a new car. Obviously to invest £18,000 on a piece of hardware that you're selling to basically colour geeks there has to be some sort of profit back to be able to make this worthwhile otherwise you're spending a lot of money on a piece of hardware and actually when you do the sums when you actually add up all of the stuff that it takes to ship a product like the elastic strap, the label on the box, the plastic screw that holds it all together not including any label at all and then you take off a vat and you might have tax as well. You're left with £14 per unit if you do an RRP price to match the colour monkey. Now £18,000 divided by £14 is an awful lot of units to sell on the assumption you're not getting paid to build any of them. So it's from a business point of view makes no sense. The only thing logically to do would be to raise the RRP but then £300 is a lot of money for a measurement device let alone £400. So am I marketing this thing wrong? I'm marketing this so far at colour geeks and so far 83 people have said I'll buy one for £300. Now 83 is not enough to make this commercially viable. I would be spending £18,000 and getting back to £7,000. So it would be a ridiculous thing to do. So am I marketing this to the wrong people? Should I be marketing this at Windows people, at OSX people? If I'm doing that do they care that it's open hardware? Do they care that they could access the firmware, they could write patches, they could do this kind of stuff? I think by changing my market away from Linux kind of geeks I'm pricing myself out of the market. But I've been playing I've been prototyping. This is an original colour hug 1 PCB with stuff bolted on until I can make it work with it as a spectro device. So this is all hand sold at multi level. It's kind of crazy. So imagine that bottom copper wire is 0.7 millimetres wide. So you can see how I find some of the soldering is. So basically the prototype works. Obviously the optic stuff would need money for other companies to do who have experience in this field. But the electrical side, nah, that's no worries. I can do that bit. This is a linear CCD designed in the 80s for photocopiers. Happens to work quite well as a collection device for a spectro meter. The specs I'm aiming for is kind of close to the if not better than the colour monkey with a UV switched illuminant. So I'm pretty sure I can do this and from a hacker's point of view and from a sort of open hardware free software kind of point of view I'd love to do this. From a business person and someone who's married to a wife I don't think I can do this. It's too much risk and it's too much money. So it's really in a role. It really is. I've got space prototypes. It's just a prototype basically saying for the box can I physically fit all the chips in. The bits of cardboard and the spectro devices are plastics that I had to spare. But it's basically showing what would it look like how big would it be, where would the USB port go all the stuff you have to work up quite early. So that's my talk. Thank you very much for listening. If you do have any ideas or feedback on ways that I could make this happen it's obviously a money thing really. Technically it's possible Should we do this? Could we do this? Any ideas? Kind of welcome. So thank you for listening. Any questions please? Wrong question. It's kind of an obvious question but since you're asking should you market this to Mac and Windows people I'm just wondering what experience you've had with other OSs already. Have you heard from people who are using the original Colorhug or want to use the original Colorhug so you have any way to gauge So very few people use a Colorhug on either OSX or Windows. It does work now. There's various firmware bugs we've fixed to make it possible. But the kind of people that are running Windows Vista or whatever don't really care that it's open hardware. You can get a cheaper version which isn't open hardware, they don't care. But that said like someone said this morning 1% of the Windows market is more than 95% of the Linux market. So I'm not a marketing person I'm a geek I'm not good at the business side of this stuff so if any of you do have any ideas or kind of can help I'm really all ears. Talk to him. Thank you very much. Thank you. Cheers.