Why do bigger CCDs = Shallower Depth of Field?

filmguy123

Veteran
Now, I know larger CCDs = larger pixels = better low light sensitivity.

And I know opening the iris all the way when at full zoom = shallower depth of field.

However, why is it that larger CCDs give a shallower depth of field at any zoom position? How does this work?

Finally, the RED camera says it has full 35mm size sensor(s)? What size is this, in fact? 1-inch?
 
However, why is it that larger CCDs give a shallower depth of field at any zoom position?
They don't. (cue the screaming mob with pitchforks calling for my head... but whether it challenges conventional wisdom or not, it's still true.)

Sensors have no effect on DOF. Larger CCDs give you a wider field of view for any given focal length. So, to get comparable framing with a smaller-chip camera, you'd have to use a wider-angle lens on the smaller-chip camera. It is the wider focal length on the smaller-chip cam, or the longer focal length on the bigger-chip camera, that is causing the difference in DOF.

Look at the EX1 and HVX, they have about the same zoom range. The EX1's lens is 5.8 to 81mm. The HVX's lens, with the equivalent field of view, is 4.2 to 55mm. Longer lens = shallower DOF.
 
Longer Lens AND more light?

When you open iris all the way, it lets more light = shallow DOF. So does having bigger sensors and more light at any given F-Stop mean the DOF will also be shallower, based on light alone? And then combined with the longer lens?
 
When you open iris all the way, it lets more light = shallow DOF.
More light does not = shallower DOF, the iris size = shallower DOF (more open iris = shallower, closed-down iris = deeper). The iris doesn't "know" how much light is coming in, and the amount of light is irrelevant to the DOF. It's the size of the iris that's determining the DOF, not the amount of light flowing through that iris.
 
Finally, the RED camera says it has full 35mm size sensor(s)? What size is this, in fact? 1-inch?
1", 1/3", 1/4", 1/2", 2/3" -- these all mean nothing in terms of an actual measurement. For example, there is nothing about a 1/3" chip that measures 1/3" in any way, shape, or form.

It's like a 2x4 board: how big is it? 2" by 4"? Try measuring one, you'll be surprised...

The Red's sensor is the size of 35mm cinema film, about 22mm x 12mm.
 
Oh, interesting. Hmm, so the RED only has one sensor for all three color channels?

So what would 22mm x 12mm equate to in terms of how we are measuring CCDs? What would the CCD size by notated in if it were to be around 22mmx12mm

Do you think we'll see a RED with 3xCCDs that are 35mm anytime soon?
 
Oh, interesting. Hmm, so the RED only has one sensor for all three color channels?
Yes. But, sensors are monochrome, so the Red has to employ a Bayer pattern color filter array over the sensor. So while it is technically 4096 x 2300-ish, that would only be in monochrome; 1/4 the pixels are filtered so that they only see red light, 1/4 are filtered so they only see blue light, and 1/2 are filtered so that they only see green.

This pattern needs to be de-mosaic'd into a color image, which typically results in a loss of about 30% of total resolution. Graeme of Red says he's getting it down to about 20% loss.

So what would 22mm x 12mm equate to in terms of how we are measuring CCDs? What would the CCD size by notated in if it were to be around 22mmx12mm
Well, the CCD size indicators are a reference back to tube sensors; a 2/3" "type" CCD was the size of chip necessary to produce performance about comparable to a 2/3" tube camera (where 2/3" actually meant 2/3"...)

Basically the Red's sensor is about 4x larger than a 2/3" chip.

Do you think we'll see a RED with 3xCCDs that are 35mm anytime soon?
Never. Having three chips raises the cost, makes the engineering a lot more complex and costly, and makes the use of film lenses much more difficult. Red's about inexpensive bang for the buck, and working with film lenses.
 
I see. So the equivalent would be around a 2.66" CCD for film size, but about 3.5" to account for the resolution loss when using a Bayer pattern. Perhaps someday as price decreases, or if they make a more expensive model, we'll see a 3.5" CCD w/Bayer-pattern? Would that be ideal?
 
Dalsa claims to have the finest digital cinematography image sensor on the planet, and it's a full-frame CCD. Sony's also putting out the F35 with a full-frame CCD.
 
(where 2/3" actually meant 2/3"...)

The sizes of imagers, both tube and CCD, have always just been a relative reference. The 2/3" and 1/2" measurement of tubes never represented the image area, that was the diameter of the tube itself. The image area of a 2/3" tube is the same as the image area of a 2/3" CCD (4:3). So even with tubes, 2/3" didn't actually mean 2/3". So the numbers have always been "wrong," but consistent!!

Doncha just love it?
 
On a related note, have CCD prices really come down much? I seem to remember 10 years ago, weren't 3x1/3rd inch CCDs still the standard in prosumer gear? Is it likely that in 5-10 years we'll see 3x1-inch CCDs standard in prosumer gear @ the 4k-6k range?

If not, that's a little disheartening. I feel like HD has simply made achieving the same results with video MUCH more expensive thanks to the decreased light sensitivity. For us live-event folks, that is. 2/3rd-in CCDs are achieving SD light results at 1440x1080 images... when we're working with full raster 1920x1080 its going to need even more light... maybe 1-in CCDs... GRRRR. :(
 
Look at the EX1 and HVX, they have about the same zoom range. The EX1's lens is 5.8 to 81mm. The HVX's lens, with the equivalent field of view, is 4.2 to 55mm. Longer lens = shallower DOF.

So the EX1 has larger CCDs, meaning a wider angle of view @ any given focal length. Does this mean, that the EX1 at 5.8mm, may actually be WIDER or EQUIVALENT in angle of view to the HVX @ 4.2mm?

If so, is there any way to measure objectively the angle of view of a camera (or find specs for this), which takes into consideration both lens and sensor size?
 
Back
Top