Why CMOS over 3CCD ?

[Bastien Tribalat]

Hi,
I'm just curious to learn more about the different technical stuff behind those types of sensors since in my cinema school the only CMOS camera we have is the EX3, the others being DSR's from Sony and Arri super-16 cameras.
Our teachers don't talk about CMOS sensors for cinema and video. They're stucked with CCD technology. But, I've seen a couple of features shot on RED, and the quality is fucking there ! (and to be honest, I prefere features shot on RED than HDCAM or the Viper)

So my question is why RED decided to use CMOS (I know it can be for keeping costs lower.... I suppose) instead of 3CCD and all the technical differences I might learn.

 

[Michael McLaughlin]

Do search first

Do search first

no disrespect, but if you had looked at the thread below

http://reduser.net/forum/showthread.php?t=41929

there are answers (and a good start) right there.

Then I would look up other post.


Then when you have deeper questions about the design areas you don't
understand; you can ask more specific questions and you will learn
a lot more.

Good Luck

 

[Michael McLaughlin]

Also

Also

read anything Graeme Nattress had posted.

The man is a walking encyclopedia.

 

[Juhan Malmstein]

Red was/is meant as a digital cinema camera, thus the sensor must be of the same size as 35mm film to use the same glass / accessories : putting 3x 35mm sized sensors and a prism in a digital cine camera would mean that it would be huge AND that none of the existing cine glass would work. The same for 2/3" sensor sizes: it's really close to the 16mm film frame size and you can use all the 16mm /35mm glass you want. if it had a prism and 3 sensors(be them cmos or ccd) it'd again be larger and it could only take ENG/EFP glass, leaving your with very expensive lens options(not to mention that I've never heard of quality primes for 3ccd cameras, just zooms that are really really really expensive)

 

[Charles Angus]

So my question is why RED decided to use CMOS (I know it can be for keeping costs lower.... I suppose) instead of 3CCD and all the technical differences I might learn.

I think you're confused about some basic stuff. I take it when you say CMOS you mean "35mm sized bayer pattern sensor", because the fact that it's a CMOS and not a CCD doesn't really make much difference to the end user.

Given that interpretation, the simplest answer is because it will work with existing cine lenses and accessories, and give that "35mm feel" that is difficult to achieve with smaller sensors.

(not to mention that I've never heard of quality primes for 3ccd cameras, just zooms that are really really really expensive)

There are excellent 2/3" primes from Fuji and Zeiss. Zeiss calls them "DigiPrimes", I forget what Fuji's are called.

 

[Pawel Achtel]

not to mention that I've never heard of quality primes for 3ccd cameras, just zooms that are really really really expensive

Canon, Fuji and Zeiss make very good prime lenses designed for 3CCD cameras. I used to have a full set of Canon f/1.4 HD primes for 3CCD. Just as expensive as the zooms.

 

[Juhan Malmstein]

My mistake then, but the point remains that having a single chip is the way forward(and backward, because now I can stick my grandfathers photo glass on a red and have awesome footage)

 

[Pawel Achtel]

My mistake then, but the point remains that having a single chip is the way forward(and backward, because now I can stick my grandfathers photo glass on a red and have awesome footage)

It's not quite that simple. You do not have 35mm equivalent to, say 42x HD zoom. There are still more lens choices for 2/3" 3CCD cameras. There are no image stabilised Cine lenses. And, no, you can't just slap on your grandpa's glass on a RED One and expect to replace a professional cinema or ENG lens.

A single chip design has many advantages over 3CCD, but glass selection is not one of them :001_smile:

 

[Petr Dvorak]

As Graeme said nothing is 100% perfect so all is about compromises, and choosing CCD vs CMOS is just example of it.

 

[Graeme Nattress]

With CCD v CMOS we're talking basically two different chip manufacture processes. That's it. It may effect certain aspects of the end image, but for the most part it's not something an end user should overly concern themselves about.

Graeme

 

[David Rasberry]

With CCD v CMOS we're talking basically two different chip manufacture processes. That's it. It may effect certain aspects of the end image, but for the most part it's not something an end user should overly concern themselves about.

Graeme

The end result speaks for itself in Red's case.

 

[Graeme Nattress]

Indeed it does.

Graeme

 

[Eisen Feuer]

CMOS in a RED? Don't worry.

CMOS in anything else? Worry. Make sure to take it for a test drive before buying. I have a Canon HF10 (knock it if you will, but I dare any of you to tape your RED ONEs to a mountain bike) and I also use a buddy's 7D, the jellocam/rolling shutter issues in either could be compared to that of your average cell phone camera.

The only other way I'd differentiate between CMOS vs 3CCD is that 3CCD has higher color accuracy, but it only becomes an issue in EXTREME light conditions, like with lasers, pure colored stage lights, etc.

For info on what rolling shutter/jellocam is, Gavin recently wrote a great description here: http://reduser.net/forum/showpost.php?p=563806&postcount=12

 

[Lee Saxon]

I have a Canon HF10 (knock it if you will, but I dare any of you to tape your RED ONEs to a mountain bike) and I also use a buddy's 7D, the jellocam/rolling shutter issues in either could be compared to that of your average cell phone camera.

I can't speak for the cell phone camera but in the case of the other two I think that's mostly because of line-skipping and not a CMOS v CCD issue.

 

[Dan Hudgins]

Compromise

Compromise

Engineering is about Compromise.

There is no single film stock that is "universal" for all tasks, people pick the one that fits the job best.

Because the results from a 4K Bayer sensor are close to or better than a 2K RGB film scan they are good enough.

If you are talking a 1920x1080 Bayer filter vs. a three chip 1920x1080 camera both using 2/3" sensors, then you should see some difference because the three chip camera has more sensor surface area and a better (more saturated) filter system. The color matrix used for Bayer sensors is not needed with three chips if the filters are saturated enough, and that matrix can add noise to the image in the way it subtracts the low resolution and noisy blue and red images from the green, requiring what are in effect low pass filters to be used.

RED made the right choice going with a 4K sensor to start off with.

Making a large sensor 4K camera with three chips would be nice, three REDs could be used to build one, but the size and weight would be higher and the cost would be greater, so one has to think about what the market might be for a "better" image.

There is no rush to build a "super-camera" right now because film still is being shipped, but when film is gone someone may make one, or maybe not, with movies going 100% CG will there be any need for cameras other than motion capture?

 

[Eisen Feuer]

I can't speak for the cell phone camera but in the case of the other two I think that's mostly because of line-skipping and not a CMOS v CCD issue.

Hmm I can see that, at least on the 7D and the cell phone cam, as their still modes shoot at least 3x the resolution (on the Y axis) as their motion modes. My HF10's still mode only does 25% higher res than the 1080p it shoots, so it's definitely from the CMOS's rolling shutter.

Oh well we're talking about 2 year old consumer technology there—I would think as far as CMOS sensors go in the professional realm; terrible, noticeable jellocam via rolling shutter is a thing of the past.

 

[Thomas Wright]

CMOS is cooler and smaller.

 

[Thomas Wright]

cooler as is not as hot, runs cooler. not as in colour balance (the chip is though, compared to the usual 3CCD's) nor in a sort of attitude way

 

[Todd Anderson]

I read an article on Nikon DSLR's a while back (I think it was an old interview from a Nikon tech from when the D2x was being released). Anyhow, from what I gathered (if I'm remembering this correctly) is that it was too costly and slow to make CCD's to meet the demand. So Nikon switched to CMOS. I saw it as a time and cost issue as to why the DSLR manufactures switched. But maybe someone else can correct me.

-T

 

[Rick Burnett]

I can't speak for the cell phone camera but in the case of the other two I think that's mostly because of line-skipping and not a CMOS v CCD issue.

No, this is actually not true.

First, as Graeme stated, the real difference between these is that CMOS uses the same technology as standard CMOS logic, therefore, the digital circuitry (and analog) can be on the same chip. With CCD it cannot because it's a different process. This is why CMOS is cheaper in the long run.

While they are both based on silicon, CCD uses a special process and is therefore limited in it's application to general circuitry use.

But beyond that, CCD devices shift out all the frames. They do not employ a rolling shutter. This is why they don't get jello.

http://www.internetvideomag.com/Articles_2008/032708_CMOSvsCCD.htm

A CMOS sensor reads out each line and goes on, it round robins through the rows. Since the bottom row is read later, and nothing is preventing that line from capturing more image data, that's why you get skew. Has NOTHING to do with line skipping. Even if you did not skip, they would still be there.