XCam XRay

by Rick LaBanca

XRay USB to Video Cable

The USB to video cable is a plug and go way to get video into your computer. Once the drivers are loaded, you can plug the cable in and you're hooked up, saving a slot in your PC. It's a simple thing... USB on one end, box in the middle, RCA and S-Video connectors on the other end.

While USB is very convenient, there can be disadvantages. USB throughput is 12 megabits per second, but that's not high enough for uncompressed video. So the device has to compress (as this does) or reduce the frames per second. Also, I'm not sure how other USB devices would contend with a full video stream going or vice-versa -- I don't have any other USB devices (if you try this, let us know!)

But X10 only market this for snapshots, so bear that in mind, the motion video is a bonus, and none of the software that X10 provides will capture live video. I'm not sure if any third party software could, perhaps the OEM of the cable will come up with something.

The company that makes the internal chip is NogaTech, and they make a test program that installs in a different program group than the X10 XRay software. The program tests live video and can take snap shots, but that's about it. It's all I used for testing.

To test the cable on even ground, I fed it from a Sony TVR-65 camcorder, rather than the XCam, or wireless transmitters. The software offers sizes from 80x60 to 352x288, I left it at the highest. The computer is a Pentium II 400mhz. For comparison, there is a screen shot from a Matrox Rainbow Runner video capture card, via the composite input.

Note on video source. The pictures are a bit grainy because they're inside at night, and that's what you get in low light with this camcorder, but they should show the relative differences accurately.

The only problem with the video is a jaggie effect because if low apparent vertical resolution. To show you, I've captured the image from the desktop using printscreen. The screen shot below shows this better than I can say.

USB cable's video captured with printscreen to show live video quality.

Note the top of the window corner, the staircasing. It doesn't look as bad in a still, when the image is moving, you really notice the stairs moving on angular lines. For comparison, here is the same output captured from the video capture board.

Captured via Matrox Rainbow Runner.

Now I've been talking about live video, X10 actually is conservative (shocking!) -- they don't mention this. They only sell it as a video snapshot device, with the client software delivering the snapshots via e-mail or Internet.

When a video device captures, it can do it over more than one frame, improving quality, often increasing the apparent vertical resolution. The test software claims a capture of 640x480, witch if true won't show those jaggie lines as badly. Here's the result.

Capture using test program's capture feature.

As expected, the snapshot from the software is much improved. I'm guessing that the X10 client software uses the same technique (will verify this once I get the software to run and review it!). I've compressed these to JPG for sake of the net, but the BMP size of the capture was around 250k. Hopefully that X10 client software has options to save as JPG to save bandwidth.

I tried to find out more about the features by going to NogaTech's site, and they list two USB/video chips. Both say they do 352x288 at 30 frames per second, and 640x480 at 15fps. No setting to go over the first size though.

Finally, what load does this put on your computer? My CPU when idle was around 17% and once I started viewing the live video it went to 52%. Exiting the app brought it back to 0%, so the test program is doing some kind of monitoring even when doing no video.

One thing that's not so good are captures of quickly moving images. If they move quickly, you get an interlaced effect. It takes two frames as I suspected, and combines them, but if there is motion, the won't align too well. Here is me moving a palm controller in a circle, note the two images.

USB capture with a bit of motion.

If you saw the full 640x480, you would notice one image on the odd lines, and the other on the even lines. I was moving the controller at a moderate speed (not sure how to benchmark that!), but I would say that you would only run across this is if the camera caught someone walking across the room, or moving quickly. Slower motion just blurs the edges a bit.

Overall impression: B