http://translate.google.com/translate?u=http%3A%2F%2Fwww.repaire.net%2Fsite%2Ftournage%2Fcompar_Z1Sony_HD100JVC%2Fcomparatif_Z1_ Sony_%26_GYHD100_JVC.php&langpair=fr%7Cen&hl=en&ie=ISO-8859-1&prev=%2Flanguage_tools

Some of you may have already seen this.

I find it interesting that the HD100 is tested at 800x700 lines of resolution and the Z1 is at 700x800 lines of resolution.

The 800 is equal to other test results from the Z1.

For the Hd100 getting 700 of the 720 pixels is pretty damn good. The thing I find interesting is that the horizontal only gets 800 of the 1280.

This makes me think current 1/3" HD lens option can only really resolve 800x800 lines no matter how good a camera is. Maybe the CANON XLH1 will get 800x800 since it is 1080i like the SONY but uses true 1440x1080 chips instead of 960x1080 with pixel shift. Of course this is for 1080i mode. The CANON may get fewer vertical lines depending on what it does to get to a 24f frame.

Anyways I don't know much about the lens details and how it deals with detail I just thought it was interesting. One thing this does show is that in terms of detail 1080i doesn't really give you anymore than 720p. 700x800 = 800x700.

This ties back to something Graeme's been saying since the beginning: the HD100 seems to be recording as much actual detail as the Z1 does. The Z1 uses 60% more pixels, but doesn't deliver any more discernible detail from it. Probably speaks more to the limitations of optics and diffraction than anything else.

I don't know how much you can read into that 800 number though -- that res chart is designed for a 4:3 camera; numbers for 16:9 would probably be different. It would have been better if they'd used something like the Accu-Chart or the CamAlign, which are 16:9-shaped charts...

Maybe the CANON XLH1 will get 800x800 since it is 1080i like the SONY but uses true 1440x1080 chips instead of 960x1080 with pixel shift.Plus the XL H1 also has Pixel Shift as well.

- TV lines of resolution are not the same as pixels. That you knew.

- TV lines of resolution are based upon a circle whose diameter is equal to the vertical frame size.

Example:
A camcorder has 700 TV lines of resolution vertical and horizontal. The maximum number of lines you could resolve horizontally is 700 times the aspect ratio, or 16/9, thus 1244 lines horizontal, 700 vertical.

That you didn't know.

800 lines horizontal for a 1280x720p pixel size is a slight misreading of the chart, since it's not possible.

Plus the XL H1 also has Pixel Shift as well.


Please correct me if I am incorrect but I think we are describing two different things as pixel shift. As I understand it, pixel shift in Canon and others is done by having a slightly offset CCD, usually green, to interpolate new information and theoretically increase resolution.

As I understand it, the Z1 does not have any pixel shift but instead use pixels that have an aspect ratio of 2:1 and from that (960*2=1920) records 1440 anamorphic to tape. Doesn't sound like the cleanest way to get to 1920x1080 but it also doesn't sound like pixel shift. I say this not to nitpick, but merely to be on the same page as everyone else as we all know the havoc that slightly differing terms can have on an otherwise coherent technical discussion.

My understanding is that the Sony FX1/Z1U use three 960x1080 sensors which are offset from each other to generate a "pixel shifted" 1440x1080 data stream. In constrast, the Canon XL-H1 reportedly has three 1440x1080 sensors, which would give it an edge up in that regard compared to most HD/HDV cameras. So the Sony is pixel shifted and the Canon is not.

(Remember that HDV 1080i is recorded at 1440x1080 resolution with non-square pixels, and then "stretched" to 1920x1080 for playback.)

The Canon, as well as basically *all* cameras, is pixel-shifted.

Just about every camera made since 1987 uses spatial-offset CCDs. If they didn't, the video would have significant fixed-pattern noise and "stripes" in the video.

Offsetting the green CCD slightly allows for there to be CCD coverage in the physical space between CCD registers.

Whether they take advantage of the spatial offset to increase resolution is one question; apparently in normal video they do not, but for still photos they do.

But basically all video CCDs employ at least a little spatial offset.

I think the lens and diffraction limits are very real, and that by Canon going to 1440x1080 for their chips is "optimistic" at best, and probably more for marketing reasons than any technical reason. 2/3" sensors are really too small to get the best out of HD - probably why both HDCAM and DVCProHD don't record the full resolution to begin with. To go with such resolutions on a 1/3" camera is just for the specs and not really for the resulting visuals.

Graeme

Well, whether the XL G1 1440x1080i CCDs are marketing hype or not, the picture looks very detailed to me. What I don't like about the Canon clips is the overexposure, whether caused by default settings of the camera or custom settings of the user, and also the chromatic abberation. It's all I have to judge.

What I do like about the FX1/Z1 is the consistently good latitude and exposure of the clips. The tonality and texture is very good, if a tiny bit less sharp than the XL G1, it more than makes up for what it's lacking, if it is. The concern I have about it, is the very amateurish looking videos where it has been whip-panned, jerked around. I'm not talking about mpeg2 breakup. My observation is about the optical image stabilization system. To be able to hand-hold one of these and make it look great, is something the big buck cams of HD Net and Discovery HD Theater do with aplomb. But all of the HDV cams seem very herky jerky when not on tripods.

Does anybody have an observation about this? The advantage of the Canon that interests me most, is the potential that it might have a more robust optical image stabilization system, or rather if what it has meets the expectation of the marketing hype directed at it. The other feature that has some appeal to me is the emphasis on longer range videography made possible by the 20x telephoto.

I'll tell you, that's one reason I went with the pint sized Sony HVR-A1: because since the video needs some kind of stabilizing device, I wanted to be able to do this as lightly as possible. Once the Steadicam Merlin becomes available, I plan to buy one and use it on every non-tripod shot I do. The way I see it, the HVR-A1 is so light that I'll be able to do this without some kind of a vest or support system.

Tom: where are you finding sample clips from the Canon XL-H1 camera? If there are any available online please provide the URL.

Tom: where are you finding sample clips from the Canon XL-H1 camera? If there are any available online please provide the URL.

Right here at DVInfo.net, home of the best HDV forums on the planet!
http://www.dvinfo.net/conf/showthread.php?t=52060

Thanks DSE, appreciate the tip!

So the Sony is pixel shifted and the Canon is not.Actually yes in fact the Canon XL H1 *is* pixel shifted. It employs electronic pixel shift in the horizontal axis. As Barry states above, most camcorders do use one manner of pixel shift process or another, and this is very much a *good* thing.

Chris: I saw that noted on your comparison between the JVC and Canon cameras, and I'm puzzled by that. If the XL-H1 has as many pixels in the sensor as it records to tape, wouldn't it at least be less pixel-shifted than other HD/HDV cameras which have to interpolate just to generate enough data samples?