filmguy123
Veteran
What are the main benefits/drawbacks here? Do CCDs dedicated to each color channel produce that much better of an image? Why would someone use one over the other?
Bayer Pattern VS 3x Dedicated CCDs
- Thread starter filmguy123
- Start date
Barry_Green
Moderator
Professional camera manufacturers and customers voted long ago: three chips are better than one. Every professional video camera uses three chips. Better color accuracy, bigger pixels, better sensitivity, everything's better with three chips.
Three chips is a lot more expensive to manufacture than one chip is; not only is the base hardware cost 3x as much, but the manufacturing tolerances for aligning them with the prism are extreme, they have to be aligned exactly right, etc.
For equivalent pixel count you can get much higher resolution and way, way better color resolution out of three chips. You can also employ spatial offset techniques to either a) boost resolution, or b) reduce fixed-pattern noise.
One chip/single-sensor designs used to be relegated to consumer cameras, but it's coming more in vogue recently with cameras like the SI-2K and Red One. One chip offers certain advantages such as the ability to use film lenses. Professional video lenses are designed with the idea of the beam-splitting prism in mind; film lenses weren't, so you can't use film lenses on a video camera (directly; you'd have to use a mini35-kind of thing). But for a camera like the SI-2K or Red, three chips just wouldn't have worked at all, so they had to go with a single-chip design. (and the results are fabulous; just not as fabulous as they could have been with three chips, but that's not possible so they made the best decision for those products).
The compromises in a single chip involve having to use a Bayer pattern, where you get 1/4 the res in red, 1/4 the res in blue, and 1/2 the res in green. Take the SI-2K for example in a 1920x1080 mode: you're getting 960x540 red, 960x540 blue, and 960x1080 green (those numbers sound familiar?) You can get similar style performance from three chips with spatial offset.
The Bayer pattern also prevents you from getting 4:4:4 color; 4:4:4 is only possible from 3 chips (or a Foveon design, which brings its own tradeoffs).
Ultimate resolution is capped on a single-chip system too; the Bayer de-mosaic process costs you about 30% of your total potential res. Graeme of Red says he's getting about 80% efficiency out of the Red's system, which is the best I've ever heard of. But that does mean your 4K system is delivering more of a real-world 3.2K of performance. (and there's nothing wrong with that, that's still the best of any camera on the market).
So, in all ways video performance is better with three chips than with one. But there are some things you just can't do with three chips, a Red wouldn't have been possible with three chips (from a lens standpoint, plus probably from the cost standpoint).
Three chips is a lot more expensive to manufacture than one chip is; not only is the base hardware cost 3x as much, but the manufacturing tolerances for aligning them with the prism are extreme, they have to be aligned exactly right, etc.
For equivalent pixel count you can get much higher resolution and way, way better color resolution out of three chips. You can also employ spatial offset techniques to either a) boost resolution, or b) reduce fixed-pattern noise.
One chip/single-sensor designs used to be relegated to consumer cameras, but it's coming more in vogue recently with cameras like the SI-2K and Red One. One chip offers certain advantages such as the ability to use film lenses. Professional video lenses are designed with the idea of the beam-splitting prism in mind; film lenses weren't, so you can't use film lenses on a video camera (directly; you'd have to use a mini35-kind of thing). But for a camera like the SI-2K or Red, three chips just wouldn't have worked at all, so they had to go with a single-chip design. (and the results are fabulous; just not as fabulous as they could have been with three chips, but that's not possible so they made the best decision for those products).
The compromises in a single chip involve having to use a Bayer pattern, where you get 1/4 the res in red, 1/4 the res in blue, and 1/2 the res in green. Take the SI-2K for example in a 1920x1080 mode: you're getting 960x540 red, 960x540 blue, and 960x1080 green (those numbers sound familiar?) You can get similar style performance from three chips with spatial offset.
The Bayer pattern also prevents you from getting 4:4:4 color; 4:4:4 is only possible from 3 chips (or a Foveon design, which brings its own tradeoffs).
Ultimate resolution is capped on a single-chip system too; the Bayer de-mosaic process costs you about 30% of your total potential res. Graeme of Red says he's getting about 80% efficiency out of the Red's system, which is the best I've ever heard of. But that does mean your 4K system is delivering more of a real-world 3.2K of performance. (and there's nothing wrong with that, that's still the best of any camera on the market).
So, in all ways video performance is better with three chips than with one. But there are some things you just can't do with three chips, a Red wouldn't have been possible with three chips (from a lens standpoint, plus probably from the cost standpoint).
