sensor question - DPChallenge Forums

02/24/2006 01:42:12 AM · #1

After reading this article I've been getting a little confused. Some questions:

1. So a full frame sensor makes your telephoto shorter?
2. an APS sized sensor needs less pixels to produce the same details as a full frame sensor using the same lens?
3. how does a narrower DOF fits into all of this?

sorry about these newbie questions. Just trying to learn.

Bear_Music

Sony 24-70mm F4 Vario-Tessar T* FE OSS

02/24/2006 02:04:30 AM · #2

1. The lens throws an image circle large enough to cover a rectangle the size of a 35mm film frame. An APS-C sensor (like on the 20D, 200D, etc) is considerably smaller than the s5mm film frame; consequently the sensor is CROPPING to a smaller portion of the lens circle, effectively producing the same result as a longer telephoto would on the full-frame sensor. 1.6 times longer, to be precise; a 200mm lens on a 20D covers the same image area as a 320mm lens on the 5D.

2. That's a very complex subject. Let's leave it for Kirbic. One thing, however; one of the advantages of the larger sensor is that it can use larger photoreceptors and still fit more of them in, and this is a plus. When you cram tiny receptors into very small areas you get a lot of interference.

3. DOF is entirely lens-dependent, and is not affected by the body on which you place the lens. That is to say, a 200mm lens focused at 75 feet at f/8.0 will show the exact same DOF on the APS-C sensor as it does on the full-frame sensor. You can prove this by cropping the FF shot to the coverage od the APS-C shot and then magnifying the image so it is the same size on your screen and comparing the two.

Not only that, but in a very real sense DOF is independent of the focal length of a lens. DOF depends on the actual, physical diameter of the aperture used. At a given reproduction ratio, at a given aperture size, DOF is the same regardless of the lens used. For example, a 180mm macro lens set at 1:1 focusing will show exactly the same DOF as a 60mm macrolens at 1:1, assuming each has the same, physical aperture set.

Note that "physical aperture" and "f/stop" are not the same thing; f/stop is defined as the ratio of the diameter of the aperture to the focal length of the lens. A 25mm aperture on a 50mm lens is f/2.0, while a 25mm aperture on a 200mm lens is f/8.0.

So, in other words, a 50mm lens at f/4.0 and focusing at 1:1 will have the same DOF as a 200mm lens at f/16.0 also focused at 1:1. The only differences will be the working distance from the subject and the shutter speed needed to make the same exposure.

R.

02/24/2006 03:10:30 AM · #3

Thank you Robt for the clarification.

1. very clear answer. Got it!
2. Rephrased question. On an APS-sized sensor, the light is cropped and saved to create a photo. So on a full-frame sensor, the same light is not cropped (comparing to 35mm); so if both sensors in question are 10mp, then wouldn't the output of the APS sensor be more detailed, since more pixels are used for the same subject in the photo?
3. I was trying to understand why some ppl say a larger sensor equals to a narrower DOF? So was trying to fit the puzzle into what I just read.
4. Just as a learning tutorial, lets say there is a camera body with a tiny 1/3" sized sensor, and it is using the same "200mm" lens you used in your example above. Would that give the tiny sensor the equivalent of roughly 440mm? But that leaves the question of, how much detail of light can the lens "optically" produce before it touches the sensor? Wouldn't that create an end result which resembles using digital zoom (interpolation)?

Thanks

Message edited by author 2006-02-24 03:21:10.

02/24/2006 03:44:11 AM · #4

for question #2 and #4, here is a diagram to aid my example. Thanks

Message edited by author 2006-02-24 04:39:46.

Canon EF 24-70mm f/2.8L USM

02/24/2006 05:32:26 AM · #5

anyone?

dwterry

Canon EOS-1D Mark III

02/24/2006 07:38:25 AM · #6

Originally posted by crayon:

3. I was trying to understand why some ppl say a larger sensor equals to a narrower DOF? So was trying to fit the puzzle into what I just read.

This one keeps going the rounds...

It's not that the sensor changes the DOF in any way. The issue is ... if you decide to fill the frame the exact same amount (let's say you want to photograph a person from head to foot) using either a full frame or a 1.6x cropped frame, you're obviously going to need EITHER: 1) a wider angle lens on the 1.6x cropped frame, OR 2) you'll have to step back a ways. Either way, you've changed the equation and thus the DOF has grown.

Now, if you stand in the same place, use the same lens, and live with the cropped image... the DOF will be identical regardless of sensor size.

02/24/2006 09:01:20 AM · #7

Originally posted by crayon:

2.) Yes, you'll put more pixels on a particular subject with a cropped sensor, assuming the cropped sensor has a higher pixel density (they usually do)
3.) Let's say you're shooting a portrait. With a FF sensor, you need a 80mm lens to frame it properly. With na APS-C sized (1.6 crop) sensor, you'll need a 50mm lens to get the same framing at the same distance. If you shoot these two different systems at the same f/stop (not the same physical aperture size), the FF camera will have smaller DoF.
4.) You've hit on a key point. There *is* a limit to the detail that can be projected by the lens. It's usually measured in line pairs per millimeter. The best 35mm lenses can maintain just over 100 line pairs per mm of resolution. The implication of this fact is that it's fruitles to take the pixel pitch of a DSLR using these lenses below 5 microns. We're at about 6 microns today, so we're close to the limit unless lenses improve greatly. For physically smaller optical systems with very short focal lengths, the lenses are able to put a higher resolution image at the sensor plane, but of course the image circle is *much* smaller. The pixel pitch for small cameras can be as small as 2 microns. The ultimate physical limit is around 0.5 micron, which is the point at which the pixel size is the same as the wavelength of the light being imaged.

02/24/2006 09:56:35 AM · #8

Just another thought on FF vs. cropped sensors. It's often said that FF cameras require the best lenses. While it is true that a FF sensor will show if a lens performs poorly at the outside of the image circle (many do), a FF cam is usually easier on a lens than a cropped sensor over much of the image circle (about half the total area). This is due to the larger pixel pitch. I've been amazed that the combination of the 70-200/2.8 IS and the Canon 2.0x II teleconverter, which was barely usable on the 10D, actually looks much better on the 5D. I'll be looking at this combination, as well as with a 1.4x converter, a little more seriously over the summer.

02/26/2006 08:11:16 PM · #9

Originally posted by kirbic:

4.) You've hit on a key point. There *is* a limit to the detail that can be projected by the lens. It's usually measured in line pairs per millimeter. The best 35mm lenses can maintain just over 100 line pairs per mm of resolution. The implication of this fact is that it's fruitles to take the pixel pitch of a DSLR using these lenses below 5 microns. We're at about 6 microns today, so we're close to the limit unless lenses improve greatly. For physically smaller optical systems with very short focal lengths, the lenses are able to put a higher resolution image at the sensor plane, but of course the image circle is *much* smaller. The pixel pitch for small cameras can be as small as 2 microns. The ultimate physical limit is around 0.5 micron, which is the point at which the pixel size is the same as the wavelength of the light being imaged.

Thank you for this info. Where can I read more about this? I'm simply interested.

CarpeNoctem

Canon EOS-20D

Canon EF 50mm f/1.8 II

02/27/2006 10:53:22 AM · #10

I've been fighting myself on this one for quite sometime too.

The only thing to change in the 1.6x cameras is the format, therefore the only things affected by that change are properties of the format.

What I've come to understand is a couple of points that can be referenced by texts such as The Camera by Ansel Adams.

1. Any given lens will always 'give' the same image, regardless of the format.

2. Image angle of view changes the same way cropping in Photoshop will change it. (This is not LENS angle of view because lens properties are unchangable by format, see #1)

3. Properties of the lens stay the same, regardless of format. (See #1)

Mathmatically speaking, putting the same lens on two digital formats (FF and 1.6x) of EQUAL resolution (say 8MP) AND crop the FF to reflect the same image view, the 1.6x should be sharper because in contain a larger NET amount of MP. The 1.6x will have 8MP in the end image because of not post processing crop whereas the FF image was cropped by 1.6 and NETS a 5MP image.

In the end, the lens properties have not changed it's impossible to change them in this manner, you need a teleconvert to change the properties of the lens.

What has changed is the format of the frame and that affects image angle of view. A 50mm 1.6x crop image angle of view is very similar to an 80mm lens angle of view. Compared side by side the images should be the same.

Best regards,

CN

PS. DOF can not change due to format, this is a function of the lens. Just because they use 'default' CoC values depending on format does not mean that taking that lens and placing it on an 8x10 view camera, you will have a different CoC. They just give you different CoC for sake of confusion because these are general values for the format (a 150mm 35mm lens throws a MUCH smaller image than a 8x10 150mm lens, even though they have the same focal length.)

Sigma 18-200mm f/3.5-6.3 DC OS for Canon

02/27/2006 11:15:38 AM · #11

Originally posted by CarpeNoctem:

PS. DOF can not change due to format, this is a function of the lens. Just because they use 'default' CoC values depending on format does not mean that taking that lens and placing it on an 8x10 view camera, you will have a different CoC. They just give you different CoC for sake of confusion because these are general values for the format (a 150mm 35mm lens throws a MUCH smaller image than a 8x10 150mm lens, even though they have the same focal length.)

For digital formats, this is only true if the pixel pitch remains the same. If you pack more pixels into the same space, your linear resolution becomes higher, and your CoC smaller. The change in CoC WILL impact the observed DoF, defined as the depth that appears in focus at 100% magnification. The observed DoF on a print will also change, if the print size is held constant, since the magnification from sensor to print is different between formats.

ElGordo

Canon EOS-400D Rebel XTi

02/27/2006 11:54:47 AM · #12

Originally posted by kirbic:

For digital formats, this is only true if the pixel pitch remains the same. If you pack more pixels into the same space, your linear resolution becomes higher, and your CoC smaller.

I was under the impression that CoC was a function of the lens aperture used for imaging, typically .030mm for a digital camera lens. Obviously, if the pixel size is greater than the lens CoC this has an impact on resolution (lp/mm). But if the pixel size is smaller than CoC of the lens, is not the resolution limited by the lens system rather than the sensor? Please elucidate, perhaps my impression is all wet!

edit: for clarification

Message edited by author 2006-02-27 12:07:26.