Color quality: 3CCD vs CMOS

[El_Mariachi94]

This is a total newb question...

I'm an young filmmaker and I currently use a Panasonic 3CCD camera. I understand that 3 CCDs result in vastly superior color quality in the image compared to a single CDD. This is because each CCD can take an accurate measure of red, green, and blue in the image, respectively. As I understand, the Red One will be using a giant CMOS sensor. Does this mean my little Pana 3 CCD camera will have better color quality than the Red One? Stop laughing, I know it's a silly question. I' m just trying to learn. Thanks.

[Jason Ing]

Ok, I'm done laughing... no wait...

..ok... I'm done.

CMOS is better. The other guy's and gal's here can give you the technical details... but if you know still cameras... you've probably seen the gorgeous still photos from the high-end Canon cameras. That's CMOS.

And as Alfred Hitchcock has said... "A movie is just a series of still photographs." Or something like that.

[Brook Willard]

Now there's no need to be rude, jing... ;)

This (http://www.reduser.net/forum/showthread.php?t=1487) is a good place to start. Pay particular attention to the "Mysterium™" section and follow the links to learn about Bayer-pattern sensors and whatnot. Also pay attention to the bit-depth section.

[Jason Ing]

Judging from you avatar, Brook... you're still laughing too.

By the way, El... Welcome to Red Revolution! :)

[C.H.Haskell]

I second that. Welcome. Bienvenidos.

[David Mullen ASC]

I don't think the final word has come down that CMOS is always better than CCD -- DSLR Nikons, for example, use CCD's.

But I think the real question here was whether you get "better" color information with a 3-CCD camera versus a single-sensor Bayer-filtered camera. There is no exact answer. If you were comparing a 1920 x 1080 single-sensor CCD with a Bayer filter to three 1920 x 1080 CCD's filtered for RGB, then yes, you might say that the 3-CCD version would have better resolution in the red and blue channels. Does that mean better color or just better color resolution though?

But once you get into Bayer-filtered sensors with high pixel counts, like the RED sensor, it is less of an issue. Color resolution in RGB can be reconstructed fairly accurately with a good de-Bayer algorithm. It's not like you have a 3-CCD camera that has 4K resolution per CCD to make a comparison to anyway (I mean, there are some experimental cameras, like the Olympus "4K" , really 3.8K, camera with two offset 1920 x 1080 green CCD's plus one red and one blue CCD.)

It may be interesting to compare the color and resolution of the Olympus camera to the RED, but the whole 3-CCD prism-block design has certain artifacts and limitations, plus the Olympus prototype has problems like sensor site rows getting overloaded and smearing vertically.

[Jason Ing]

Yeah... like David said. ;)

[El_Mariachi94]

Awesome, thanks guys. I can only hope my film school (NYU) decides to pick up the Red One.

[Jason Rodriguez]

Bayer sensors usually have more color-channnel cross-talk than a 3-chip system . . . so it's not necessarily an issue of color reconstruction accuracy during demosaicing, but also an issue of getting the color registered on the chip in the first place without a lot of optical mixing from the surrounding color pixels. The nature of a 3-chip system with it's dichroic color filters can allow for some very pure decoupling of the R, G, and B channels, where-as you typically need a matrix transform of some sort to decouple R, G, and B in a Bayer sensor. And of course a saturation matrix adds noise to the image, with more noise added depending on the amount of cross-talk and the required gains in the matrix to decouple the color channels.

[Jonas Nyström]

Olympus Octavision - what is the size of the chip. Is it a 2/3 as it is 1920x1080? Thinking about DoF. It doesn't look so experimental anyway...

http://octavision.olympus-global.com/en/

[Poi Boy]

in the high end still market ccd has always had the edge on quality. Having said that cmos has come a long way and I think will take the lead going forward. As far as cinema goes single chip cmos is going to rule.
Aloha
-A

[Evin Grant]

I don't think the final word has come down that CMOS is always better than CCD -- DSLR Nikons, for example, use CCD's.

The two exceptions being Nikon's D2 series cameras, the D2X has a CMOS and the D2H has an LBcast which is a very similar tech to CMOS.

[Michael Brennan]

Perhaps it was the man who invented bayer who said in the 60s that a prism is the only way to get full colour information from a particular image.

But by increasing a bayer sensor size to well beyond audience perception of any inherent losses makes the discussion between single and 3 chip a non event.

Compared to CCD CMOS still suffers from what I call murky shadow detail.

Much mooted control of individual pixels that CMOS has on offer has yet to be implimented to realise stella improvements in dynamic range

For the moment the main advantage is that they are comparatively cheap to make.

Mike

[GlennChan]

Does this mean my little Pana 3 CCD camera will have better color quality than the Red One?
I have a Panasonic GS70 (3CCD consumer camera) and it does not have very good color resolution. You can measure this to some degree with a zone plate test chart.
http://glennchan.info/Proofs/dvinfo/green-magenta-zone-plate.png

My quick and dirty GS70 test:
http://glennchan.info/Proofs/dvinfo/pixel-shift-aliasing.jpg
Note that it wasn't a very good test (no tripod, glare in the screen, camera not back far enough / not enough oversampling/zoom in the test pattern). But if you add a grey fill layer in Photoshop and toggle between luminosity and color, you see that the luma resolution is MUCH higher than color resolution (vertically).

It's because most 3CCD cameras use pixel shifting.

[Graeme Nattress]

Both CCD and CMOS have very different qualities. To say one is better or worse is not correct. That said, I do believe that for high resolution, medium speed digital cinema, you can't beat a single CMOS.

Graeme

[Bachman]

for a very good explanation go to http://www.dalsa.com/markets/ccd_vs_cmos.asp

I think some ppl should think before they laugh

[REDHKSC]

I have read the attached " The Revolution of Digital Imaging from CCDs to CMOS - White Paper for my own reference.

Stewart

[REDHKSC]

I have read the attached " The Revolution of Digital Imaging from CCDs to CMOS - White Paper for my own reference.

Stewart

Sorry, exceed the size can be upload here, please email me so I can send to those interested parties to review the paper.

Stewart