Oversized CCD chips

[Barry Goyette]

Ok..we're writing over the top of each other here. I'm not really willing yet to buy that this is a typo (if it is, then fine) but why do we think it is...because the image circle is .240? where is there a listed spec for that...aren't we only assuming the .240 because we're talking about a .236 chip.

I've been shooting photographs for a long time...and I've never known a lens with an image circle that was nearly an exact match for the sensing area...they are typically 1/3 again or more larger. I don't know this to be true for the xl series...but it makes sense to me that it would be more than .004 larger than the xl1s sensor, that is very small tolerance for an image circle.

on the typo...typo's are usually one number not two... you are suggesting that both numbers are incorrect, yes...that wouldn't be a typo...that would be just wrong.

Again, I just have to ask the question as to why we think that the numbers are wrong...they seem pretty precise to me...not some casual mistake.

And this has always bugged me about the supposed smaller 4:3 sensor area....its the WHY....why would canon not use the whole sensor if it had lens coverage to cover all of it...it just doesn't make sense. (what if they didn't have the coverage...oooh...)

Barry

[Boyd Ostroff]

<<<-- Originally posted by Barry Goyette :.why would canon not use the whole sensor if it had lens coverage to cover all of it. -->>>

There are just too many unknowns here, so everything degenerates into mere conjecture. I would just note that there is a precedent for not using the entire CCD for 4:3 video. The PDX-10 does the same thing: http://www.greenmist.com/dv/16x9/10.JPG. It could have to do with a variety of things, such as not being able to read data from that many pixels 30 times per second. But anyway, from the info we already have, it seems like Canon wanted to optimize this camera for 16:9.

[Barry Goyette]

Boyd...

Your example also brings up this...if sony has the ability to process a 667k chip area on the pdx-10 (and an even larger 16:9 pixel count)...I would imagine canon could do the same with the effective area of a 680k chip....but now we are digressing.

I agree that all of this is conjecture. Like Chris said...somebody call the product manager...anyone got his phone #?

Barry

[Boyd Ostroff]

Well since we're having so much fun just wildly speculating... remember that the PDX-10 only shoots interlaced video while the XL-2 is shooting progressive, so one might expect that to require more bandwidth...

[Barry Goyette]

Boyd

ok so we're having fun, huh...then teach me how to make those cool italics in my posts....that's fun.

It seems to me that 60 fields of interlaced footage would require about the same bandwidth as 30 frames of progressive footage...but I'm not even close to informed on such matters.

Noticed that the "typo" made it onto the just released XL2 brochure as well. (they fixed a few other things)

If Chris is still stirring...has anyone actually compared (visually) the angle of view between the xl1s, and xl2 in 4:3 mode to see if they are different. (don't quote numbers here...we're all learning not to trust them). Short of taking a micrometer to the sensor, this would be the easiest way to tell if the sensor size has in fact changed. I did a few calculations that seem to indicate that the sensor is the same size based on the 35mm comparison chart you built, and my lens spec sheets. (since excised from a previous post). But they aren't backed up by what I'm seeing in camera.

Barry

[Chris Hurd]

Barry, Canon's pretty sure the angle of views are different between the XL1S and XL2... see their focal length w/XL2 (scroll down a bit there) and compare that data to their old XL1S lens data... or better yet...

Compare this chart: http://www.dvinfo.net/canonxl2/articles/article04.php

to this chart: http://www.dvinfo.net/canon/articles/article58.php

I took those numbers from Canon's published specs. Also in my CCD article, to quote myself, here's what I came up with:

"The multiplication factor for using an XL lens on the XL2 is 1.105 in the widescreen 16:9 aspect ratio and 1.35 in the standard 4:3 aspect ratio."

Use [ i ] and close with [ / i ] to make your text appear in italics. Hope this helps,

[Barry Green]

<<<-- Originally posted by Chris Hurd : Barry, Canon's pretty sure the angle of views are different between the XL1S and XL2... -->>>

Sorry to rain on people's parade, but yes indeed, in 4:3 mode the Canon XL2 is using the equivalent of a 1/4" CCD.

I just cross-referenced Canon's stated focal-length equivalencies against a chart of equivalent fields of view, and in each case the XL2 4:3 equivalencies matched the 1/4" CCD column.

No need to worry about actual diagonal specs and whether Canon printed 'em wrong, just looking at the focal length equivalencies will show that the camera is looking through a 1/4"-sized window.

Would have been nice to be wrong, but it looks like we're not: in 4:3 mode the XL2 uses a patch of its CCD equivalent to a 1/4"-CCD camera.

[Chris Hurd]

A better way to make the same point I did. Thanks, Barry,

[Luis Caffesse]

"No need to worry about actual diagonal specs and whether Canon printed 'em wrong, just looking at the focal length equivalencies will show that the camera is looking through a 1/4"-sized window."

Well, I think there is a need to worry about it actually, because either the diagonal specs are wrong, or the focal length's are wrong.

But, it seems that everything is pointing at the diagonal specs being wrong.

And I guess now they are wrong both on the website, and on the brochure. That's a shame.


By the way, Barry (Goyette), if you look on the Canon specs website, you'll see the same numbers repeated not only for the chip, but for the LCD. It looks like it was just a bad cut and paste job. Or, like we were saying, a very bad typo.

Either way, it seems to be a mistake.
I won't be surprised if we see a correction in the near future, and the actual diagonals match the numbers I posted earlier.

-Luis

[Barry Green]

<<<-- Originally posted by Chris Hurd : A better way to make the same point I did. Thanks, Barry, -->>>

I was agreeing with you completely -- that's why I quoted your response in there. You pointed towards the charts, so I went and looked at 'em and yes, sure enough, Canon lists different focal length equivalencies for the two cameras using the same lens. Then I went a step further and compared those equivalencies on an angle-of-view chart and yep, it matches what a 1/4" CCD would deliver.

I know people were getting excited about there being a mistake made, and that maybe the XL2 had a bigger chip than we had all originally thought (hence the title of this thread) but Chris Hurd knows his stuff, and his block diagram article spelled it out in the beginning.

In 4:3 mode, the XL2 is basically a 1/4" CCD camera. Might be a heck of a good performer, that remains to be seen -- but one thing we can say now is that in 4:3 mode the DOF will be equivalent to the GL2, rather than the XL1s.

[Dan Vance]

Regarding the image circle though, my guess is that it is significantly larger than the .236 diagonal (this would be indicated by canon's ability to utilize a .28 16:9 sensor area) typically lenses project a circle much larger than their intended imaging size...to minimize distortions and vignetting near the corners.


Hi Barry (Goyette),
I've measured the image circle on my 1/3" Fujinon, my 1/2" Fujinon and my 35mm SLR lenses and they are all just a tiny bit bigger than the stated format diagonal. This makes sense, because any increase in the image circle size means an increase in glass diameters and therefore cost. So from a manufacturing standpoint, they would always try to minimize the size of the image circle to make it as small as practicable for the format with room for manufacturing tolerance errors.
But .289/.236 is a huge increase (22%). I really don't think the circle would be that big. (But, I've been wrong before!)

[Barry Goyette]

Chris, All,

Thanks for clearing this all up. I had tried to make the same comparison's you suggested, but had been unable to find focal length conversions for the xl1s to match up with the xl2's. That left me comparing my xl1s with my canon 35mm ( and the strong impression that I was making an ass of myself).

While I was hoping for all of us that I was right about this, I was, in the background, trying to prove myself wrong. Still, it does seem remarkable that numbers like these could get published.

Sorry for all the hubbub.

Barry

[A. J. deLange]

Given that the XL2 active area is 16 units wide and 9 units high the diagonal is sqrt(16^2 + 9^2) = 18.3576 units and the diameter of the exit pupil must be at least that. If one masks off a vertical stripe 2 units wide at each end the active area is now 12 units by 9 units (4:3 ratio) and the diagonal of this area is sqrt(12^2 + 9^2) = 15 units. As the XL lenses are still useful we can assume that whatever the XL1s chip diagonal may have been, the diagonal in 16:9 mode in the XL2 must be about the same and that the active diagonal in 4:3 must, thus, be 15/18.3576 ths of that. If the XL1s chips are 0.3333" then the active diagonal in the XL2 in 4:3 mode must be about 0.28". Is it really any more complicated than that? Am I missing something?

What I'd really like to know is how the extra 33% (4/12) pixels in 16:9 mode get stuffed into the DV stream and, the related question, what is the effective resolution. I've noticed that discussions of TV lines (and SNR) are glaringly absent from any of the material Canon has released. Anyone have any idea about what these most basic specs might be on this machine?

A.J.

[Luis Caffesse]

"If the XL1s chips are 0.3333" then the active diagonal in the XL2 in 4:3 mode must be about 0.28". Is it really any more complicated than that? Am I missing something?"

Yes, read the entire thread and you'll see a discussion about the actual measurement of CCD diagonals.

A 1/3" chip is not .33 inches in diagonal, but closer to .236 inches, or 6mm.

-Luis

[A. J. deLange]

OK so then the XL2 4:3 active area is .236 * 15/18.3576 = 0.193"

I did go through the whole thread and if someone posted actual measurements I missed it. Sorry.

A.J.