Canon XL2 CCDs are 4:3. - Page 5 at DVinfo.net

Go Back   DV Info Net > Canon EOS / MXF / AVCHD / HDV / DV Camera Systems > Canon HDV and DV Camera Systems > Canon XL and GL Series DV Camcorders

Canon XL and GL Series DV Camcorders
Canon XL2 / XL1S / XL1 and GL2 / XM2 / GL1 / XM1.


Reply
 
Thread Tools Search this Thread
Old December 26th, 2004, 07:12 PM   #61
Trustee
 
Join Date: Dec 2004
Location: Lipa City Batangas, Philippines
Posts: 1,110
Thought there would be some replies to this by now. Anyway, I'd just like to say thanks to Barry and A.J. deLange for carrying out these tests and posting the results. This is the best way to get closer to the truth.

Richard
Richard Hunter is offline   Reply With Quote
Old December 27th, 2004, 03:22 PM   #62
Inner Circle
 
Join Date: Sep 2002
Location: San Mateo, CA
Posts: 3,840
Great article in this months DV magazine by Adam Wilt on widescreen and CCD sizes. His comment regarding the XL2's use of 4:3 chips, is that the results are "Just as good" as if the company had used 16:9 chips.
Richard Alvarez is offline   Reply With Quote
Old December 28th, 2004, 09:14 AM   #63
Wrangler
 
Join Date: May 2003
Location: Houston, TX
Posts: 4,093
AJ and Barry,

Although I don't understand all of the technical things you are discussing, I do understand what you're saying about the resolution chart results.

I was momentarily taken aback by this apparently proven reduced resolution of 16:9, until I realized that though the result may be technically correct, it is sort of comparing "apples to 1.33 apples" and is inherent to the fact that 4:3 and 16:9 are -- obviously! -- different aspect ratios.

Just so folks don't misunderstand and start saying, "Aha, the XL2's 16:9 sucks after all" please allow me to expand on what I know you already know and, I'm sure, would agree with:

The same subtended arc sees less percentage of the widescreen image in the XL2 than it does of the 4:3 image. Your results compare the horizontal resolution of the full width of a 720x480, 4:3 image against 75% of the width of a 720x480, 16:9 image. So it isn't at all surprising that the 16:9 image looks 25% smaller and resolved fewer lines in your test jpg...less of its width and fewer of its pixels were used in the test.

Without having done any such tests myself, I'm quite confident that if we compare images of equal horizontal or areal percentage, the 16:9 will resolve just as much detail as 4:3, if not MORE...again, since the image aspect ratios and PARs are different, one really needs to compare the whole images, or at least equal percentages of the images, to do a practical comparison.

Please don't get me wrong. We all greatly appreciate the work you are doing and sharing with us. I just didn't want that point to be confused...it briefly surprised me, but I was willing to stop and think it through. However, not everyone might be so inclined. Didn't want anyone going on to misinterpret your interesting findings.
__________________
Pete Bauer
The most beautiful thing we can experience is the mysterious. It is the source of all true art and science. Albert Einstein
Trying to solve a DV mystery? You may find the answer behind the SEARCH function ... or be able to join a discussion already in progress!
Pete Bauer is offline   Reply With Quote
Old December 28th, 2004, 04:35 PM   #64
Major Player
 
Join Date: Jun 2004
Location: McLean, VA United States
Posts: 749
Pete,

See if you agree with this. You have a perfect camera i.e. one in which all that engineering stuff you referred to works at its theoretical limit. It's got lots and lots of pixels in both the horizontal and vertical directions and all its resampling algortithms work really, really well. The camera puts out DV. It's scheme for doing 16:9 is the same as the XL2's. It just takes samples from a wider area in 16:9 than it does in 4:3 and the height from which it takes samples is the same.

You frame a picture of a picket fence in 4:3 such that the vertical extent of the frame just fits the vertical length of the pickets. The pickets are spaced an amount equal to their width and there are 360 of them in the frame. When this image is converted to DV there will be 360 vertical rows of white pixels, one for each picket and 360 vertical rows of dark pixels, one row for each space between pickets.

Now you switch to 16:9 mode. The image captured by the camera now covers 480 pickets and 480 spaces. The picket height is still the same number of pixels. The camera resoves these new pickets as well as it did before because all that has changed is that you are switching more pixels on. But, and here's the heart of the matter, the camera has to down sample in the horizontal direction to the same total of 720 in order to put out DV format. You no longer have just 1 picket or space per column of pixels but rather 1 picket and 1/3 space. Thus the pickets are not as well resolved. In return for having to look at aliased pickets you get to see more of them. To get rid of the aliasing you would have to move in closer so that you only have 360 pickets in the view finder again. This would no longer permit you to capture the tops and bottoms of the pickets.

No, the message is not that the XL2 sucks in 16:9 mode. It is just that it's ability to resolve in the horizontal direction is a little less than in 4:3 for the same level of vertical resolution.
A. J. deLange is offline   Reply With Quote
Old December 28th, 2004, 08:04 PM   #65
Wrangler
 
Join Date: May 2003
Location: Houston, TX
Posts: 4,093
AJ, that's an outstanding "thought experiment!" As before, what you've said is entirely logical and I don't disagree with a word of the experiment. I would just phrase the conclusion a little differently. Let me try again, taking a slightly different approach. See if you guys agree with this, then:

It isn't the XL2's "limitation" per se. What I think you guys are driving towards is that DV 16:9 images are inherently different than DV 4:3 images. They're different geometric shapes composed of differently shaped pixels and you can't really make a generalized comparison of them in one dimension only...16:9 has a 1.2 PAR and 4:3 has a 0.9 PAR. You could conversely say that for a given width, the vertical resolution of 4:3 is only 75% the resolution of 16:9...hence the 360 pixels of "squeeze mode" wide screen in the previous generation of cameras. Thus, one could do your awesome picket fence experiment in the vertical and say the same thing about the XL2's 4:3 that Barry's horizontal experiment causes you to say about 16:9.

Since what we really look at onscreen is a two dimensional image, let's compare 16:9 and 4:3 images of equal area: a 144 square cm screen. The 16:9 image is, simply enough, 16cm wide by 9cm high; the 4:3 image is 13.8564cm wide by 10.3923cm high. Both of them will have (at theoretical limit) 720x480 pixels; the pixels are just shaped differently and thus collectively occupy a differently shaped area. But viewed from the same distance, the images will overall, on average appear equally as sharp because they contain the same amount of image information. The two different geometries will have slightly different strengths and weaknesses, and lead to artifacts under somewhat different circumstances, though...such as horizontal and vertical picket fence tests!

Again, I am not at all saying Barry's experiment is faulty. It is showing specific (valid) results that I think speak to the geometric limits of DV rather than the quality of this or any other particular camera. But, once we can compare 1 apple to 1 apple, then the quality of the CCD block, etc, etc becomes fair game. I wonder...has anyone devised a whole-image "resolved pixels" test? If not, maybe it is time?
__________________
Pete Bauer
The most beautiful thing we can experience is the mysterious. It is the source of all true art and science. Albert Einstein
Trying to solve a DV mystery? You may find the answer behind the SEARCH function ... or be able to join a discussion already in progress!
Pete Bauer is offline   Reply With Quote
Old December 28th, 2004, 09:02 PM   #66
Trustee
 
Join Date: Dec 2004
Location: Lipa City Batangas, Philippines
Posts: 1,110
Hi Pete. I would agree that shrinking down the 16:9 image will make it a bit sharper, but the main point is that for the same vertical image size and resolution the 4:3 image will be sharper than 16:9 since the 720 pixels represent a smaller width.

I'm not so sure of your point about the artifacts, however. The 16:9 will have downsampling artifacts such as those in the picket fence example. The 4:3 has not been downsampled at all, so it will not. Are you saying that the 4:3 image will suffer from some other form of artifacts that the 16:9 image will not?

Richard
Richard Hunter is offline   Reply With Quote
Old December 29th, 2004, 08:51 AM   #67
RED Code Chef
 
Join Date: Oct 2001
Location: Holland
Posts: 12,514
<<<-- Originally posted by Richard Hunter : Hi Rob. As I mentioned, I am not comparing stretched and native 16:9, I am trying to compare native 16:9 and native 4:3. The vertical resolution is therefore fixed at 480 (for NTSC) so is not a variable in this case.

16:9 can be described as higher resolution than native 4:3. -->>>

I don't think that has ever been said. What everyone was saying
was that native 16:9 has a resolution increase over fake 16:9.

You either want 4:3 or you want 16:9. If you want 4:3 then shoot
in 4:3 since it will indeed have more resolution (but not more
sampling) than 16:9 mode.

If you need 16:9 then it is better to shoot in a true 16:9 form than
it is to do a stretch or letterbox since you increase (vertical)
resolution and you loose some horizontal resolution, but you gain
a wider field of view and a higher horizontal sampling.

So yes, you are correct in that the horizonatal "overal" resolution
will be lower in true 16:9 compared to 4:3 on the same camera.
However you get an increased sampling (which might appear as
a higher resolution) and a wider view in return, ofcourse.
__________________

Rob Lohman, visuar@iname.com
DV Info Wrangler & RED Code Chef

Join the DV Challenge | Lady X

Search DVinfo.net for quick answers | Buy from the best: DVinfo.net sponsors
Rob Lohman is offline   Reply With Quote
Old December 29th, 2004, 08:57 AM   #68
Wrangler
 
Join Date: May 2003
Location: Houston, TX
Posts: 4,093
Richard,

My point is that both images end up as 720x480. Both will theoretically show up to 360 pickets horizontally. That's it; no more. If you turn the camera 90 degrees, both will show 240 pickets. That's it; no more. It has nothing to do with downsampling, CCD blocks, or glass; it is simply the limit of DV.

So Barry's test was entirely predictable: it compared the full 720 pixel width of a 4:3 image with 75% (540 pixels worth) of the width of a 16:9 image. So the 16:9 image looked smaller and less sharp in that test. If you turned the camera 90 degrees, you could show the same result against 4:3. You could oversample by a billion pixels and use a Cray computer for downsampling...and still not get better results because that's all that DV can give you.

It doesn't matter that it is 16:9 or 4:3. They just happen to be different shapes derived from a grid of 720x480 pixels that are shaped differently...0.9PAR vs 1.2PAR. For anti-aliasing algorithms, maybe DCT (don't know about that), and such, the edges of pixels may be located differently and that may make a small difference in the final image.

Now, if we look at the full width (or height) of an image and our camera is giving us less than 360 (or 240 for height) pickets, then we are starting to talk about the capabilities of our cameras.

I hope that helps clarify.
__________________
Pete Bauer
The most beautiful thing we can experience is the mysterious. It is the source of all true art and science. Albert Einstein
Trying to solve a DV mystery? You may find the answer behind the SEARCH function ... or be able to join a discussion already in progress!
Pete Bauer is offline   Reply With Quote
Old December 29th, 2004, 02:59 PM   #69
Barry Wan Kenobi
 
Join Date: Jul 2003
Location: North Carolina
Posts: 3,863
I agree with what Pete is saying, completely.

It is important to keep in mind that the 4:3 picture is showing more detail in the picture, but is showing less overall picture. The 16:9 shot was showing more, overall.

There's nothing wrong with the XL2's 16:9 mode, I hope people aren't drawing the conclusion that there is! It's a different shape, and as such it's... well, things are going to be different when you're shooting 16:9 vs. 4:3!

About the only thing that could be conclusively drawn is that you shouldn't shoot 16:9 with the idea that you could pan 'n' scan crop down to get a 4:3 extraction, unless you're prepared to have a lower-res 4:3 version. If you want a fullscreen 4:3 image you'll get better results by shooting in 4:3, than you would by extracting a 4:3-shaped patch out of the 16:9 image. (Not that anyone was discussing doing so, but it's really the only conclusive decision I can draw from the test results!)
Barry Green is offline   Reply With Quote
Old December 29th, 2004, 05:43 PM   #70
Wrangler
 
Join Date: May 2003
Location: Houston, TX
Posts: 4,093
Yup. Pan-n-scan would suffer, just as "stretch mode" 16:9 suffers. Is it possible we have reached consensus on a topic at DVinfo? How often does THAT happen! Mark your calendars! ;-)

Barry, thanks again for doing testing on real rez charts. If I've missed some links along the way, point me to them, but I haven't noticed too much objective testing posted, like the test you placed in this thread.

Are those who have actually done resolution testing on the XL2 satisfied that in either mode it shoots fairly near theoretical limit? It SEEMS to based on the beautiful videos, but it would be interesting to see more rez chart results like Barry's.
__________________
Pete Bauer
The most beautiful thing we can experience is the mysterious. It is the source of all true art and science. Albert Einstein
Trying to solve a DV mystery? You may find the answer behind the SEARCH function ... or be able to join a discussion already in progress!
Pete Bauer is offline   Reply With Quote
Old December 29th, 2004, 06:35 PM   #71
Trustee
 
Join Date: Dec 2004
Location: Lipa City Batangas, Philippines
Posts: 1,110
<<<-- Originally posted by Rob Lohman
2. true 16:9 (whatever method)
- resolution and FoV increase in 16:9 compared to 4:3
- 4:3 is lower resolution than 16:9 (that is the whole idea)

-->>>

Hi Rob. Don't want to belabour the point, but what I was questioning was the statement above that has nothing to do with fake 16:9. If this is not what you were saying, then we (probably) don't disagree after all.

Pete, I agree with what you say. I just don't agree that 720x480 16:9 can be higher resolution than 720x480 4:3, which is what I interpreted the quote from the post above as saying.

I also agree that the reduction in horizontal resolution is not any limitation of the XL2 in 16:9 mode. It is a direct result of the DV spec that only allocates 720 pixels even when the width has increased by 1/3.

And by the way, I am very happy with my XL2. :)

Happy New Year?

Richard
Richard Hunter is offline   Reply
Reply

DV Info Net refers all where-to-buy and where-to-rent questions exclusively to these trusted full line dealers and rental houses...

Professional Video
(800) 833-4801
Portland, OR

B&H Photo Video
(866) 521-7381
New York, NY

Z.G.C.
(973) 335-4460
Mountain Lakes, NJ

Abel Cine Tech
(888) 700-4416
N.Y. NY & L.A. CA

Precision Camera
(800) 677-1023
Austin, TX

DV Info Net also encourages you to support local businesses and buy from an authorized dealer in your neighborhood.
  You are here: DV Info Net > Canon EOS / MXF / AVCHD / HDV / DV Camera Systems > Canon HDV and DV Camera Systems > Canon XL and GL Series DV Camcorders

Thread Tools Search this Thread
Search this Thread:

Advanced Search

 



Google
 

All times are GMT -6. The time now is 07:15 AM.


DV Info Net -- Real Names, Real People, Real Info!
1998-2017 The Digital Video Information Network