Infinity by Thomson
Back in December 2008, we left Richard Everitt after he told us all about the Thomson camera factory at Breda in the Netherlands. Now we continue with Ronny van Geel, Product Manager Cameras, and Ronny is the Infinity specialist.


Ed: Ronny, would it be fair to say that Infinity has been a “work in progress”?
Ronny: Yes, a lot of things have happened to the Infinity DMC over the past years. At the moment we announced the Infinity series, a transition from SD to HD was about to take place and there was a lot of fuss around the transition-race worldwide. At that very moment, we announced Infinity with an appealing mindset and an appealing message …
Ed: Which was that it was capable of SD and HD in the one camera, so it could cope with that transition?
Ronny: Yes. We positioned it in the market as a product in a world of transition, like it’s for instance doing both SD and HD in one model. We also have made available both 1080i and 720p in the same model. Interestingly enough, a fairly large number of potential customers also were still in the phase of deciding themselves what it was they were exactly looking for. Because our message was so early compared to the mindset of the market, initially the Infinity product was compared to a 3rd generation SD camcorder product. Only later, the customers started to compare apples with apples, like two-thirds inch acquisition rather than half inch acquisition.
Ed: So Infinity was really the first two-thirds inch HD camcorder product?
Ronny: Yes – with the full capabilities of what you would expect as a customer when making the move to the next generation of camcorders.

Ed: And that was why it was relatively hot and it was heavier?
Ronny: Compared to the SD camcorder again, yes. HD video processing is one and a half times the SD power consumption by default. It was only until the competition showed their preliminary specs that customers really understood the impact of going to HD acquisition. And with the progress of technology, the history will repeat itself. Five years from today, the figures on power consumption will be far different from today’s figures, just like it developed in the early SD CCD camcorder days. Of course, we’ve received a lot of feedback that we’ve incorporated into the development process. Many of the philosophies were different to any other camcorder in the world. For example, we initiated a whole different way of controlling settings. Audio routing, for example, historically was done via tactile switches. All of a sudden we had this touch-screen graphical user interface because it’s impossible to catch 4 channels of full random routing in tactile switches. The initial response was “it’s different”, directly followed by “but we like it a lot”. A lot of feedback went back into the design and this wave of initial feedback made the product we have in the market today. We also completely changed our imaging solution. In the labs we were working on the new Xensium imager – our own in-house development and it became something absolutely beautiful.
Ed: And this is a CMOS device, or something special?
Ronny: It’s a CMOS device, the first CMOS device that we’ve built ourselves. Of course, we have been developing imagers for more than 40 years and we took a completely different approach than any other CMOS imager available in the field. Competing CMOS imagers are typically developed by silicon manufacturers rather than by people who understand imaging. So the Xensium imager turned out to be a very, very good imager. At that moment, we came at a very important decision point at the end of December 2006. Do we take the time to incorporate Xensium into the camcorder or do we continue with the product as it is defined today? We decided to make the change to the new imaging technology and we’ve incorporated a lot of other feedback that a lot of customers had given us world-wide. So basically, what people see today is the Mark II of the Infinity camcorder without us ever having released the Mark I. That basically explains why the initial announcement and the first customer shipment date were relatively far apart. But most important and interesting: the message we brought at the very first announcement still has the same high level of appeal as today’s Infinity message. We’ve delivered what we’ve promised, and the story of openness, versatility, transitions and future-proofing resonates with customers. Those are the real key values of Infinity and that’s what we’re standing for.
Still a significant amount of broadcasters who are broadcasting HD today, gather their content in SD which gets aired upconverted. What you typically find in these stations, is that they have an HD camera in the studio, they broadcast HD, but the majority of the content is upscaled SD.
We believe the majority of the transition on the camcorder part is only starting right now. Today you have the real comparison between the competing models and the Infinity. The best part of it is, yes, of course competition has taken over some ideas we’ve announced early, but Infinity still stands out clearly. The competing models don’t have 720 progressive formats, we do. We have 24p available today and at least one of the competing models doesn’t. We have SD and HD in one model, the competing models don’t; etc, etc. So there are a lot of features in there that people really need in this world of transition right now and that’s still part of the Infinity story. Rather than focusing on forcing decisions, we offer answers to the questions customers have today.
Ed: It’s still early days yet, obviously. In terms of the camera manufacture, we’ve been here at the factory, we’ve seen the two cameras going through the assembly process side-by-side, but it’s not an LDK8000 in disguise is it? It’s actually quite a different camera, for a different market. Can you explain the difference between the two?
Ronny: The Infinity DMC is designed from the ground up. The whole architecture inside the camera is different from the other cameras in our portfolio. When designing a new camera, you can choose to create a spin-off of an existing model, or to make a radical design change. For Infinity we’ve changed everything you can think of, which means that the platform is completely different in a lot of ways compared to our other cameras. For example we took into account the modern PCBA technologies and progress in those areas. Just like we’ve worked together with a lot of customers world-wide during the definition phase of the DMC, all people in the factory have been involved in designing the new Infinity. And basically it is the next step we all believe in, in every way you can think of – so architecture, download management, PCBAs, mechanics, everything has been looked at again.


Ed: So it really has an IT architecture because of the way a News gatherer wants to get their material from the camera to the broadcast situation?
Ronny: It’s interesting that you say that, because the architecture is not so much IT, but what you can do with the camera is IT enabled. So a lot of people see it as a PC with a lens. Your PC can do USB, can do BlueTooth, can do Ethernet – all those kind of things; but try to do that at full speed all the time with your PC, and you will find it can’t perform to the expectations. So we’ve made the IT based interfaces available, but to get there in a product that everybody wants to rely on 24 hours a day and that has to work under every condition on batteries, it’s a complete different starting point from a PC that is connected to the mains power 24 hours a day.
Ed: Now in terms of comparing the two camera lines here – the Infinity and the LDK8000 – apart from having different chips, different sensors in there, they’re quite different cameras, built for different purposes?
Ronny: Correct. The Infinity camcorder is really made to fit the camcorder market; the LDK8000 really is there for top-quality multi-camera productions. On the inside, today there’s not so much in common, but the same of course can be said of the generations of cameras before the LDK 8000. So it’s like with the tube/CCD transition; this is the first CMOS camera, but the technologies will find their way also in the other ranges of cameras in the future.
Ed: Now it’s interesting that you bring up the future, because what do we see as the future for Infinity? Is it just going to develop as it is, because I guess one of the major interests people have, is that you have gone for a totally different recording medium. You’ve gone for the REVPRO drive, which is a hard disk in a box, and some people might think “Hmmm, that’s a step backward” whereas other manufacturers have gone solid state and have gone laser disc?
Ronny: So we, of course, think laser disc is a step backwards! Infinity has both support for spinning disk and solid state storage solutions, so you can record both to REVPRO which basically is a hard disk split in two and you can store your content on Compact Flash with the two built-in Flash slots today. So while the other manufacturers are quarreling about what’s best, we offer both spinning disk and solid state solutions. And comparing apples to apples, today REVPRO is the most economic option available. In addition, Infinity can record to external hard drive solutions, where you buy half a Terabyte today for €120.
The main difference in approach is that the other manufacturers have closed the media. So the P2 has a PCMCIA form-factor with Panasonic on it, and the professional disc is Sony’s closed version of the blu-ray. If you look at SxS media, again it’s a standard form-factor that has been closed by a manufacturer of camcorders. The main theme in the approach we took is “openness.” So the people who buy the Compact Flash cards from the stores know that the market prices are driven by the digital still photography market – we don’t invest one single Euro in solid state storage and our end-users benefit from the economies of scale. This also completely fits in the mindset of, for example, our EDIUS editing solution, which just takes any format you can think of on the timeline. So the whole open approach is the main mindset we have with Infinity, as well as within the company as a whole. We don’t say there’s a 720p and a 1080i camera and you have to make a choice when buying equipment – it’s all there and available. We don’t dictate the decisions, but offer the possibilities.
Ed: Just in terms of the Compact Flash, is it capable of those top end HD uncompressed data rates?
Ronny: By far. For example, if you look at today’s SanDisk solutions, they go up to 400 Megabit per second and higher.
The highest profile we offer in the Infinity today is 100 Megabit per second in the case of JPEG2000 compression. The main question you have to ask yourself when you go for solid state is “what do you consider most important?” Is it ingest time, or is it capacity? If you go for the capacity, you could choose the biggest capacity available, which is still very well capable of handling absolute HD quality at 100Mbps, but the ingest takes longer compared to a faster Compact Flash card. If the time it takes to ingest is key, you typically find media at half the capacity but with a two times faster ingest at the same price-level. So, yes, the Compact Flash media available today are very well capable of handling HD content. The only one that you might find to not work is a 64 Megabyte Compact Flash card in your kitchen at home, but anything else that can be bought today, is very likely to be able to handle HD content.



Ed: So really, there’s no issue with Compact Flash currently being of too low a speed for higher data rate sources, it doesn’t matter whose camera it is?
Ronny: Correct. P2 as well as SxS are closed media by choice. There’s no issue with choosing the much lower priced commercially available Compact Flash solutions today.
Ed: Now I guess one of the other major points about the Infinity is that it’s almost codec agnostic. You can pretty well record any codec, but in there Ronny, you have a favourite one, a favourite flavour?
Ronny: My favourite flavour is vanilla ice-cream and the JPEG2000 is my favourite codec in there.
Ed: Speaking of vanilla ice-cream, have you ever tried hokey pokey ice-cream?
Ronny: I don’t think so. Is it available in New Zealand?
Ed: Only in New Zealand, so you’re going to have to come and try it. But anyway, back to the codec?
Ronny: JPEG2000 my is favourite for a number of reasons. One of the big reasons is, of course, the 10 bits …
Ed: 10 bits?
Ronny: It’s a 10 bit 4:2:2 codec, yes … and the most important differentiator is that it is wavelet based. When you start with high quality material, like from an Infinity camcorder all the way at the beginning of the chain, you know it will be handled a few times by codecs before it is put on air at a very low bit rate. So it needs to be absolutely perfect at the acquisition-stage. I’m not saying that MPEG is doing a bad job; but one of the drawbacks of MPEG in general – MPEG2, MPEG4 – is that it is based on DCT blocks.
DCT means that the image is encoded by making small blocks and encoding those. It also means that when you have a very demanding scene or use too few bits to encode the content, we all know the little block-artifacts that can pop up in the scene. Now if you introduce those artifacts at the very beginning of the production, and take those effects all the way through the production chain, those blocks are encoded/transcoded over and over throughout the chain,using up valuable bits to basically encode artifacts. So it means that every artifact you add to the image, makes the next step in your chain less efficient. JPEG2000 is built on wavelet technology which can be compared to what an equalizer does with audio; you have frequency bands which give the full symphony when available all together at the same time. We have three layers, or frequency bands if you like. The first layer is the more or less soft, low resolution layer which is also quite small in bandwidth demands; then the second layer adds more detail to that. And then the third layer completes it and brings it up to the highest quality level. So, the picture is divided into frequencies rather than blocks. It’s like with zooming in on a Google Earth map. When you zoom in too far, you don’t see blocks appear like you would get with bitmaps, but still see a smooth picture. There are no sharp edges introduced in any of the compression/decompression steps when JPEG2000 is used. This approach means that the real picture is the only thing that counts and what’s played back after decoding. And if you, for instance, work with EDIUS, you can even choose up to which layer EDIUS should decompress the content for pre-viewing while working on, for example, a laptop. With this so-called built-in proxy, you can edit 100Mbps HD JPEG2000 on a standard laptop straight from the REV drive in the field, without having to worry about the logistics of handling separate proxy-files.
Ed: Now Ronny, you mentioned “wavelet” and these days as soon as I hear the term “wavelet” I think Red camera. Are you connected in some way with Red?
Ronny: No, no, we’re not connected in any way with Red, but they also use the wavelet codec in their RedCode, which is based on the same technology as we use. Of course they only have one imager and use a Bayer-pattern approach, etc, so it’s a completely different league they’re playing in, but yes, the codec is based on the same technology.
Ed: So basically Red code is JPEG2000?
Ronny: It’s using the same wavelet compression, yes.
Ed: So that’s why EDIUS can edit either?
Ronny: Well EDIUS is as codec agnostic as it can be, so EDIUS takes any codec anyhow in any resolution on the same timeline. So if EDIUS wouldn’t be able to take it in, then I would be surprised!
And that, dear readers, is that from the Thomson camera factory at Breda in The Netherlands. If I had to choose one impression to come away with it would be that of confidence that they know what they are doing in both camera design and in their quality of construction. Check out the product for yourself and see.