Interview: Matrox Unleashes RT.X100
Page 2 of 4
DMN: As Intel and AMD chips get faster, your system will be able to do more, as you said. Does this mean that you'll unlock these features when these faster chips are released or will it be, all of a sudden, the RT.X100 will be able to do more -- or is it a combination of these two things?
Legault: It's a combination of the two. As more filters exist and more effects exist, you may find yourself trying to do more than the system can do, and it could drop frames. But as machines get faster, they'll be able to keep up in real time. Also, certain things like unlocking more video streams is something we're going to do in future releases as machines allow us to get to that point.
DMN: So users can look forward to this benefit, if someone buys an RT.X100 now, a great benefit would be that they'll be able to download software that will suddenly enable more video streams.
Legault: Yes, at some point, they'll be able to do that.
DMN: You also made great strides in easing the installation of the RT.X100. Can you describe what the difference is in the installation? What was the problem before and how was that fixed with this new release?
Legault: We experienced some difficulty in the past with installation. It was cumbersome. For example, it was a requirement to have three re-boots that needed to be done while you were installing the product, which was a big annoyance to customers. So what we've done is integrated the new product to the Windows XP operating system, and we made some changes in the lower-level driver -- instead of the previous 1394 driver we're now using the Microsoft driver stack -- and because we did that, and also wrote the lower-lever driver, we were able to eliminate the three boots, and now require just a single re-boot of the computer. So, basically, it greatly reduces the time it takes to install the product and eliminates some of the annoyance. In our testing with Beta sites, users were able to install the product in as little as a half an hour. It's fairly simple where it used to take a lot longer. The effort we've done to make a better fit with the plug-and-play technology of Microsoft Windows XP is really what made the installation a lot easier.
DMN: Everybody is going to be glad to see that.
Legault: I think so. I think the first experience out of the box is very critical.
[an error occurred while processing this directive]
DMN: So now, Alain, here's your chance to blast your competitors, the Pinnacle Pro-ONE and Canopus DVStorm. What's better about the RT.X100 over its competition?
Legault: First, talking about the Canopus DV Storm, it at first looks impressive in terms of things like number of video layers, but you also have to wonder what you can do with those video layers. We find there's a lot more freedom and creativity in the ability to create keyframes and the ways to manipulate the layers on our platform than there is on the Canopus platform. In the Canopus product as well, if you compare the two products and what you get for the price you pay, you have to add more components to the product before you have a complete solution. For example, you have to add the MPEG-2 encoder, (which now they're throwing in for free, for a short period of time), but there are things that are native in our product. The quality of our effects are a lot better. Our 3D effects are performed with a 3D graphics chip from Matrox which includes sophisticated anisotropic filtering technology which provides very good quality video, even at difficult angles. Those are things that the two other product lines, the Pinnacle and Canopus products aren't doing very well.
DMN: What is this anisotropic filtering? Is that why your 3D effects look so good? It's like anti-aliasing, right?
Legault: It's a type of filtering, and what's special about an anisotropic filter is that it's a two-dimensional filter that works in the X and Y direction, but can provide different levels of filtering in X and Y depending on the angles. Let say you create a 3D surface which has a strong angle going deep into the image. Say the image is higher than it is wider. If you use a less-sophisticated filtering method, you apply the same filtering methods on X and Y, inevitably making one direction not filtered properly. With anisotropic filtering, you can adapt the filters in both direction to take into account both angles, and you always keep the sharpness of the image at whatever angle. It's especially obvious if you do a steep angle of attack, like in 3D graphic text. Text is especially difficult, because it has high frequency content. Other companies don't do that because they don't use as good of a 3D graphic chip, like Pinnacle, or they use old-style software like the DV Storm, where their 3D is especially low quality because the kind of filtering I just described requires a ton of processing.
There's a reason why our 3D graphic chip is about as big as a Pentium 4 -- it's because there's a lot of complexity in the calculations. If you're trying to do that on the host processor, you have to compromise somewhere. The compromises they're making to try to keep up with real time -- there are two -- quality and flexibility or keyframability. If you play with their 3D plug-ins, you'll find that you don't have anywhere near the flexibility. Everything is pre-canned, pre-calculated. They find they cannot compute on the fly because there's not enough horsepower to do that.
So, 3D effect-wise, we have better 3D effects, we have more of them in real time, and they're all top quality. Although Pinnacle does certain things with the 3D graphic chip in the Pro-ONE, I think they make compromises in terms of quality. You see a lot of jaggedness in the edges. These are things that came out in your shootout from last December. Things that used to be a strength in our previous product are still a strength in our new product. We've preserved all the good stuff like 3D. Now we have all the new stuff on the host side which gives us great editing performance. If you've played with the product, you probably saw that the scrubbing is very fast, no latency, not sluggish, very high performance.
And finally, we do a lot of stuff in parallel, like we have 16 real time capabilities all at once, I think it's fairly well described in our product guide. Our product has the most real time all-at-once capability currently. Here's a list: Two layers of video with transparency, two layers of graphics with transparency, two channels of field-based slow motion, two channels of color balance, two channels of input-output level control, two proc amps, two keyframeable 3D transform, and finally two channels of keyframeable cropping. That's the most you can get today in a product in this category.
Prev 1 2 3 4 Next
[an error occurred while processing this directive]
