Our frequent contributor Steve Mullen takes a crack at Canon's 24F. Interesting reading!

http://www.videosystems.com/e-newsletters/hdvwork_10102005/

In his article Steve states, that "Row-Pair Summation acts as a filter that reduces effective vertical resolution by about 25-percent". Why is that so?

Very nice article. I had always felt as though they might use the same or close to it method of frame movie mode which isn't a bad thing really.

1080i=810 lines(in theory)
1080f=810 lines(in theory)

This is how Canon can say "with no loss of resolution". When compared to itself in 1080i mode there is no loss of resolution.

810 lines isn't bad either. I'm sure any more detail than that would be lost due to other 1/3" issues.

In his article Steve states, that "Row-Pair Summation acts as a filter that reduces effective vertical resolution by about 25-percent". Why is that so?

I take that ("acts as a filter") to mean that some kind of blending, or blurring, occurs between pairs of rows - which is like a filter. With the loss of sharpness comes a loss of effective resolution, which is being calculated at 25% or thereabouts. I could be wrong, though.

Si

All interlaced cameras use this type of filtering to reduce flickering that can happen with interlaced CCD's. Even SD cameras have this so as far as I know a SD interlaced camera can only really give 360 lines. (Somebody confirm if this is true I'm taking a guess on the number here)

This is why progressive SD camera are so nice because you actually are getting a lot more detail than before. It doesn't just look better because it is progressive scanned but there is also 25% more detail.

With 1080 HD this isn't much of a big deal because it will be very hard or next to impossible to ever make a 1/3" camera that could resolve more than 810 lines of resolution anyways. Not without buying a lens well over $10,000.00.

In terms of detail you will have a hard time finding a 1/3" camera that could give you any more real detail compared to 1080 24F/30F. There are so many other issues that soften a HDV image at this level that the pixel count over 720p is almost useless anyways.

David Newman, CineForm's chief tech. guy (or CTO), states, that "[t]he 24F image doesn't quite resolve the same image resolution as the 60i, but it seems higher than one interpolated field". Sounds like 608 lines of effective vertical resolution.

Source: http://cineform.blogspot.com/2005/10/canon-xl-h1-early-results.html.

With 1080 HD this isn't much of a big deal because it will be very hard or next to impossible to ever make a 1/3" camera that could resolve more than 810 lines of resolution anyways. Not without buying a lens well over $10,000.00.

Which is why I suppose that optional lens for the new JVC camera costs $13K. More than twice the price of the camera alone.

-gb-

Very interesting info Robert. Thanks.

It may still be doing the mothod to get to 810 but the other factors are keeping the image from reaching the full 810 lines. If this is the case than the HD100 may still have a little bit more vertical detail at 700 lines.

I have done a few tests though and when I down convert an image from the XLH1 to 1280x720 and then back up to 1440x1080 it does get a little bit less detail. Only a tiny bit and you can only see it if you zoom into the image more than 200%. Of course this softening could just be due to the bi-cubic filtering during scaling twice.

Great, so yet again we still don't know the true story and won't for some time. Am I the only one who is frustrated by this? All of the camera companies seem to do this game playing with resolution loss and the terminology (Sony being the worst offender with CF24) but for once I would just like a straight answer.

The straight answer? That's easy:

For the no compromise-solution, get a camera with 3 1920x1080 2/3" progressive scan CCDs with dual-link HD-SDI out.

There are plenty of options in the mean-time:
- 3 1440x1080i 1/3" CCDs with vertical good progressive fudging and HD-SDI out
- 3 1280x720p 1/3" CCDs with split screen and a few other bugs
- 3 960x1080i 1/3" CCDs with resolution gaining pixel shift and field dropping
- 1 1920x1440(i/p?) 1/3" CMOS (with rolling shutter and bayer pattern?)

And then...
- 3 (not 1280x720)p 1/3" CCDs with perhaps some resolution gaining pixel shift

Given how good the first generation of 1/3" HD cams are - imagine how good round 2 will be?

-Steve

Next Tuesday Video Systems HDV@Work Newletter will post an updated version of my story based upon my recent discussion with Canon.

I'll post the most relevant paragraphs here after it appears.