Ruminations on the XL2 CCD Block
 

First, let me say that after a week of playing/shooting with my XL2, I'm totally awed by the quality of the output.

Looking at the data available on the CCD block, I read that the total area of the block is 960x720. In the 4:3 mode, it uses only a fraction of the CCD block with an area of 720x480. In the 16:9 mode, it still uses a fraction of the CCD block with an area of 960x480. I'm left wondering why Canon would make a CCD block oversize like this. Wouldn't it make more sense to make the block 960x480? Maybe it has something to do with the format of the CCD block during manufacture?

What would result if the entire block was used? The frame size would be 960x720, which is getting suspiciously close to the HD 720 format, albeit not widescreened. Would it be possible for a new format of 960x720? Not knowing the maximum frame size consistent with the clear aperture of the XL lenses, I'm guessing that the XL lenses won't handle a frame size much bigger than this without vignetting.

Bottom line....what's Canon's intent with a 960x720 CCD block? It seems non-sensical to me. Am I not understanding something?

the interesting thing is that SONY's new HDV camera uses CCD's that are 960 x 1080. We get more V resolution but the chips seem to have as many pixels as the XL2. The SONY shoots those 960 as a 1.5 ratio anamorphic and scales them to get the correct anamorphic 1440 pixels in the camera. If we had a filter for the XL2 to give us a 1.3333 anamorphic onto a 960 CCD we could get 1280 pixels. Then if we could get the full 720 from the chips we would have 1280 x 720. Sadly however this wouldn't be perfect because those 960 pixels from the XL2 are non square so in terms of square pixels we only get 853. If we had a 1.5x anamorphic filter for the camera we could then get 1280 square pixels with the same uprezzing that the SONY HDV camera does. Actually in order to get the full 16x9 image from the SONY HDV camera the H rez gets upsampled twice. It first goes from 960 pixels from the chips to 1440x1080 to fit in the HDV specs. 1440 x 1080 however is anamorphic since it is 4:3. To get 16x9 1920 x 1080 the H rez gets scaled up again by 1.3333. This adds up to mean the 960 pixel chips get scaled up 2 times their size to get the full HD rez.

Sadly I doubt any of this would ever work with the XL2 so everything I just said is pretty much useless.

The probable reason is cost. 16:9 shaped CCD's are for some
reason more expensive than 4:3 shaped once. So it looks like
Canon (some other manufacturers are doing the same thing)
simply choose the 4:3 chip size and only use the middle portion.
Saves money.

See the following thread on what would be needed to turn a
960 x 720 chip into full blown 1280 x 720 HD resolution (if it
where possible to use the full chip):

http://www.dvinfo.net/conf/showthrea...threadid=30740

Thanx for the reply and the link Rob.

The cost reason is manufacturing. It might seem like it's cheaper to manufacture 16:9 CCDs because they would use less silicon, but it's more complicated than that.

Say it costs you$20,000 to setup your process for manufacturing chips, and then each chip costs $1 to manufacture. If you manufacture 100,000 4:3 CCDs, it costs $120,000 total, so each chip comes costs $1.20 total to produce. But if, instead, you manufacture 80,000 4:3 CCDs and 20,000 16:9 CCDs, it costs $100,000 for the chips, plus <em>$40,000</em> in setup fees (you have to setup twice for the two different kinds of chips), thus costing $140,000 total, or $1.40 per chip.

But they're not going to be sold this way. If the factory paid an extra $20k to setup for 16:9 CCDs, they're going to have to justify that in the 16:9 CCD price. Since it costs $20k in setup and $1 per chip, their cost per 16:9 CCD is actually going to be $2 per chip, whereas the 4:3 chips will only be $1.25 per chip ($1 per chip * 80,000 chips + $20,000 setup fee all divided by 80,000 chips). So it's cheaper just to use an "oversized" 4:3 CCD.

This is an extremely simplified example, but the point is the same. Hope it helps clarify.