3 Chip or 1 Chip?
 

Does anyone think there is a chance that Panasonic will make their new HDX-100 a 1 chip camera?

(I'm not sure if this belongs on the HDX board, so feel free to move me around if need be).

In an email discussion with Jeff Kreines recently (maker of the Kinetta) he explained how he is able to get 110 minutes of uncompressed footage on a 480 GB raid. After doing the math, I had figured you could only fit 36 minutes of uncompressed footage in that much space. He explained that because his camera is only using 1 chip, he can capture a 12 bit image uncompressed, whereas with a 3 chip camera that same frame would be 36bits. So, my calculations were correct, 36 minutes for a 3 chip camera. But you can get 3 times that with a 1 chip camera, with no added compression.

This made me think, DVCProHD is 100mbs, as defined by the spec. If Panasonic were to make a 1 chip DVCProHD camera, could they achieve the 100mbs stream with LESS compression than the 6.7:1 that the varicam is using now? (closer to 2.2:1)?

By the same token, would this mean that 1 chip DV cameras actually have lower compression than 3 chip DV cameras, because they are starting out with less information to begin with, but both are being compressed to the same size (3.125 MB/s)

It doesn't seem to make sense, and I'm sure I'm probably wrong in this thought process, so please let me know what I'm missing. I'm no engineer, so I'm just sort of putting things together from what I seem to understand.

It just seems like the human eye can't really see the difference between a modern 1 chip vs a 3 chip image. And, in general, most of the problems I see in image quality (from DV and DVCProHD) are really caused by compression.

So, if they released a camera using only 1 chip, they could not only increase the chip size, but lower the compression. This would give us potentially higher image quality, which would hold up better in post, yet take up the same amount of drive space (12.5 MB/s) making it still practical to work with.

But then again, I'm probably way off.
:)

Hi Luis,

A charge-coupled device (CCD) is an analog device; it's not digital (lots of folks aren't aware of this, or that a CCD is monochromatic and doesn't know what color is). The compressed digital signal going to tape comes from the camera's analog-to-digital converter and digital signal processor. In other words, in very general terms relative to today's camcorders, how many CCD's and the number of pixels on them doesn't really have that much to do with the recording format.

Color accuracy is primarily a question of three CCD's vs. one CCD with an RGB color filter vs. one CCD with a complimentary color filter. The bit depth of the camcorder's A/D converter and DSP will determine how clean the image is and how much digital quantization, or banding, there is in the image. The recording format governs compression. Of course I've put this in highly simplistic terms because that's all I'm capable of, but you get the idea.

"The recording format governs compression"


I was aware that CCDs are analog devices, I guess I got caught up in bitrate and thought that perhaps the compression was only concerned with the final bitrate of a given format.

So, what you are saying is that the compression for any given format is a fixed rate? Meaning, regardless of the number of chips, or the amount of analog information being fed to the A/D converter, the compression will always remain the same?

So, a 1 chip DV camera and a 3 chip DV camera are both compressing the signal at 5:1?

And, in terms of the HDX-100, no matter how many chips it has, DVCProHD will always be 6.7:1?

<< Meaning, regardless of the number of chips, or the amount of analog information being fed to the A/D converter, the compression will always remain the same?
>>

Yep!

<< So, a 1 chip DV camera and a 3 chip DV camera are both compressing the signal at 5:1? >>

Yep!

<< And, in terms of the HDX-100, no matter how many chips it has, DVCProHD will always be 6.7:1? >>

Yep!

Hope this helps,

On a three chip camera, one chip takes RED, one takes GREEN, and one takes BLUE.
When you have a camera with only one chip, the chip usually has this thing called a Bayer pattern. It means that the first pixel records green, the second takes red. And the third takes green and the fourth red.
On the second row this would be B,G,B,G...
So a bayer pattern is then
G,R,G,R,G,R...
B,G,B,G,B,G...
G,R,G,R,G,R...
B,G,B,G,B,G...

This means that you can save a raw bayer pattern image as only having one channel of info. Jeff Kreines is propably doing this. He's propably recording uncompressed bayer pattern images.

The bayer images must be debayered in order to get them to be normal color images. This process will ofcourse make them take just as much data as any other color picture of that format would take. So if you'd debayer your footage to RGB files, then they'd take up three times more data, than just a raw bayer mage. You can think of a bayer image as a kind of a black and white image, that kind of has it's color information compressed onto the individual luminance pixels.

This is why 3CCD:s has a better colour reproduction than a 1 CCD camera. But one could argue, that on HD resolution it's not that big of a deal. Most of the new HD cameras from film camera makers are going to have one big chip (like Kinetta, and Arri's new camera.)

So, DVCPRO-HD, or DV, can never be raw bayer, as it's always in 4:2:2 or 4:2:0 or 4:1:1 color space.

Kinetta is propably going to have CMOS chips instead of CCD chips, but that's another issue. To put it short: CMOS is better, and cheaper.

Hope this explanation doesn't make the issue even more hard to understand. Hope it helps someone.