How does sensor size produce a shallower DOF?...

[Leon Wittwer]

on viewing the same sized print at the same distance. If you want to define a new concept based in part on equivalent image quality, go ahead but call it something else so others will not be confused.
--
Leon
http://homepage.mac.com/leonwittwer/landscapes.htm

 

[Leon Wittwer]

The sensor size does not change the DOF at all.

DoF is based on the perception of focusl blur when viewing the same sized print at the same distance. Different sensor sizes require different amounts of image enlargement (magnification) to get to the same print size so sensor size does impact DoF. This shows up in the circle of confusion used in DoF calculations which scales as the sensor size. See
http://www.outsight.com/hyperfocal.html#digital
http://graflex.org/lenses/lens-faq.html
http://www.dofmaster.com/dofjs.html
--
Leon
http://homepage.mac.com/leonwittwer/landscapes.htm

 

[PhotoTom]

The depth-of-field equation involves 4 factors - focal length, subject distance, aperture (f/stop), and the circle of confusion.

The circle of confusion is a mathematical attempt to standardize the effects of enlargement from the sensor or film size to a somewhat arbitrary standard print sized. It also takes into account normal viewing distance as this effects percieved sharpness and focus as well. Anyway, smaller sensors get a smaller circle-of-confusion factor because they require a larger increase in size to achieve the proverbial "standard" print size. (note that the standard sized print is debatable as well, some say 6 X 9 inches, others 8 X 10 or 8 X 12 inches.)

Anyway, on with the images (one of which has already been posted):

First image, taken with the 5D vs. 30D from the same location and same f/stop, but with the focal length changed on the 5D to match the framing of the 30D. I feel that this is the most reasonable comparison since it matches the way we tend to use our cameras. We select the focal length that achieves the image we want.

Note that the depth-of-field is more shallow on the 5D than on the 30D. This is a result of the change in focal length and not sensor size. But the change in focal length is a direct result of the change in sensor size. If you want to take an equally-framed image from the same perspective (location), you will use a longer lens on the larger sensor:



In the second comparison, I used the same lens on both the 5D and the 30D and shot again from the same location. I then cropped the 5D image to match the framing of the 30D (with a little slop - sorry). The result is very similar DOF. I do see a slight difference, but I believe that to be the result of sloppy test methods more than anything else. I did use a tripod so there was no significant movement involved. Anyway, here's the results:



In the third example, I again used the same lens on the 5D as on the 30D but I moved closer with the 5D to achieve similar framing (again, pardon my slop - I didn't maintain exact side-to-side equivalence here). The result is that again, the DOF is shallower on the 5D as noted by the blur on the 1.4X teleconverter and the little statuette. Note how moving the 5D closer causes the background to appear smaller and further back in comparison to the original 30D shot. The perspective is changed and so the shot is different. The table with the subjects was not moved:



--
Tom

 

[Hellashot]

I'd pick the full frame image

 

[sugar208304]

5D 140 mm 2.8



D200 85 mm/2.8



5D 85 mm/2.8



And I'm always using the same enlargment, because I guess we are always willing to frame the same shot

sugar

I understand aperture size, subject to camera distance and focal
length and I understand that all of these change compared to APC
size sensor in order to get the same FOV. But how does the size of
the sensor contribute?

Thanks

--
EOSMan (got print button?)
http://www.pbase.com/eosman

--
-----------------------------------------------------------------
I can crop at the long end myself if I want to

some humble pictures : http://www.flickr.com/photos/67259727@N00/

 

[carlk]

Seems you don't understand the difference between aperture size and lens size.

.. will make the lens look like it was shot on an f4 lens. Taking
this to the extreme - this is why there is so very little DOF
control on a P&S camera - because so little of the lens is being
used that bascially everything is in focus (unless your focus
distance is very close to the lens on a P&S)

P&S's are about a 7x crop factor.

 

[Lee Jay]

And if I enlarge a cropper's image to 10 feet by 15 feet then cut
out a 4 x 6 inch print? You can't compare at different quality
levels because you can get any result you want.

Whatever. As I posted above, these are from the exact equations for DOF. Do you not believe this or something?

They smaller CoC is caused by the increased enlargement. And that's why smaller sensors have less DOF at the same focal length - smaller CoC.

5D, 100mm, f2.8, 10 feet, DOF = 0.5 feet
20D, 100mm, f2.8, 10 feet, DOF = 0.32 feet
Pro 1, 100mm, f2.8, 10 feet, DOF = 0.13 feet
S3, 100mm, f2.8, 10 feet, DOF = 0.08 feet

From: http://dofmaster.com/dofjs.html

--
Lee Jay
(see profile for equipment)

 

[malcolm82]

Are you talking about the actual physical size of the aperture, as
opposed to the f number???

Exactly, why would you need to use a bigger aperture just because your sensor can collect more light? Someday we will get sensors that support iso25, does this mean you need to use a bigger aperture? Besides even now people dont always use iso 50 or 100.

When you chose a certain aperture and iso mode on a 1.6 crop camera you make those decisions for depth of field, noise level and shutter speed. Those variables dont change when shooting full frame wich means you will now use a 2.5 times higher iso mode with a 1.6times smaller f-number at the same shutter speed.

These different settings are actually exactly the same, iso and f-number are just irrelevant numbers and have no meaning across different formats, it would be much better to use numbers for lens speed and sensor sensitivity that reflect the total amount of light transmitted and collected. Since we dont have these the best we can do is use equivalent values just like is done with focal length.

 

[Lee Jay]

Are you talking about the actual physical size of the aperture, as
opposed to the f number???

Exactly, why would you need to use a bigger aperture just because
your sensor can collect more light? Someday we will get sensors
that support iso25, does this mean you need to use a bigger
aperture? Besides even now people dont always use iso 50 or 100.
When you chose a certain aperture and iso mode on a 1.6 crop camera
you make those decisions for depth of field, noise level and
shutter speed. Those variables dont change when shooting full frame
wich means you will now use a 2.5 times higher iso mode with a
1.6times smaller f-number at the same shutter speed.
These different settings are actually exactly the same, iso and
f-number are just irrelevant numbers and have no meaning across
different formats, it would be much better to use numbers for lens
speed and sensor sensitivity that reflect the total amount of light
transmitted and collected. Since we dont have these the best we can
do is use equivalent values just like is done with focal length.

I think this is all correct. However, people are used to thinking in f-number instead of aperture because that number normalizes focal length for exposure.

But you are right, if you use the same FOV and aperture (not f-stop), the DOF and SnR will be independent of sensor size.

There are some practical limitations why this is not usually possible, however. For small sensors, you'd end up with very fast f-numbers. Lenses with very fast f-numbers are very hard to produce with good optical qualities because you can't scale one of the needed parameters - the refrative index of the available materials. If we had materials of arbitrary refractive index then ultra-fast, ultra-short lenses would be practical. But, for now, they aren't so scaling the sensor size up to avoid such lenses is more cost effective.

--
Lee Jay
(see profile for equipment)

 

[artur5]

They smaller CoC is caused by the increased enlargement. And
that's why smaller sensors have less DOF at the same focal length -
smaller CoC.

5D, 100mm, f2.8, 10 feet, DOF = 0.5 feet
20D, 100mm, f2.8, 10 feet, DOF = 0.32 feet

Lee Jay
(see profile for equipment)

Yes, DOF with APSC sensors is smaller due to a lesser CoC but, in order to take the same picture with a 20D you have to move away from 10 feet to 16 feet (compared to a 5D with the same lens )

So :
5D, 100mm, f2.8, 10 feet, DOF = 0.5 feet
20D, 100mm, f2.8, 10 feet, DOF = 0.32 feet
20D, 100mm, f2.8, 16 feet, DOF = 0.79 feet

 

[malcolm82]

It means the real reason for bigger sensors, assuming sensor technology is not limiting is optical quality, even at small apertures you get better results with lenses on bigger sensors. And also as you said it means larger sensors are more suitable for larger apertures.

 

[saynomore]

But in any case, if you move, there's no way you can get the same picture (perspective is different) with any lens or any camera combination. The real equivalence would be to take the pic with equivalent focal lengths without moving.

The only DOF change directly affected only by sensor size is the CoC. The one you are pointint out is not just because of sensor size, but both sensor size and distance to the subject.

They smaller CoC is caused by the increased enlargement. And
that's why smaller sensors have less DOF at the same focal length -
smaller CoC.

5D, 100mm, f2.8, 10 feet, DOF = 0.5 feet
20D, 100mm, f2.8, 10 feet, DOF = 0.32 feet

Lee Jay
(see profile for equipment)

Yes, DOF with APSC sensors is smaller due to a lesser CoC but, in
order to take the same picture with a 20D you have to move away
from 10 feet to 16 feet (compared to a 5D with the same lens )

So :
5D, 100mm, f2.8, 10 feet, DOF = 0.5 feet
20D, 100mm, f2.8, 10 feet, DOF = 0.32 feet
20D, 100mm, f2.8, 16 feet, DOF = 0.79 feet

 

[Lee Jay]

And if I enlarge a cropper's image to 10 feet by 15 feet then cut
out a 4 x 6 inch print? You can't compare at different quality
levels because you can get any result you want.

I finally figured out what you were trying to say here (I think). And you are right. One of the basic assumptions of DOF calculations is that you have sufficient pixels to resolve the CoC. If you don't, what you will see will not be just OOF blur, but also pixelation.

However, that is not the issue of the OP or of my examples. Unless you are examining your monitor with a loupe, you are seeing OOF blur, not pixelation. And when that is the case, DOF does get small with smaller sensor size, when all other parameters stay the same, just as calculations and photographic examples illustrate. When that is not the case, you have other problems (pixelation).

--
Lee Jay
(see profile for equipment)

 

[Bach184117]

It makes me 6 moths to really uerstand the concept behind all those.

Very simple when you get it ... but very hard to get it.

IF I am not wrong, at 100% (with same lens etc..) a 18Mp camera have same DOF than a 12MP camera. To have the same print size AND same frame, the DOF is the same if you crop your 18Mp to 12Mp. If you want more details (ie 18mp instead of 12mp in he same frame you have to get a lens (18/12=1.5) 1.5x tele which gives you les DOF.

Bach.

 

[S Hess]

Sensor size (on a dslr) determines how close to the subject you need to be to achieve the fov you want. With ff and a 50mm lens, you need to be closer to the subject but with a 1.5 or 2.0 crop you need to be farther away from the subject to get the same fov. This difference in distance will change the dof.

 

[RapidPix]

I think they show that for practical purposes the photographer achieves a shallower DOF for a given composition with a full-frame camera.

Of course one can create artifical differences by cropping and magnifying but that's sort of like just moving closer to that theoretical "standard print size" rather than viewing from a "standard viewing distance."

--



See equipment list in profile.

 

[Donald Cooper]

Technically a larger sensor gives a larger DOF because the image doesn't have to be enlarged as much. That's how a larger sensor affects the DOF. (This is the Circle of Confusion).

The problem is that in practice in order to frame a shot in a similar way with two cameras, one with the larger and one with the smaller sensor, you have to either move further away with the larger sensor or use a longer FF lens when using the larger sensor camera.

The tricky bit is that the change in the DOF effected by the sensor is linear while the change effected by the distance or lens focal length is expodential. As a result, the effect of the sensor size is lost in the larger change attributable to the focal length or distance.

So as a practical matter you may not notice the sensor size. People think consumer cameras produce very large DOF -- and this is true. But that is due to the very short focal length lenses being used and despite the small sensors, not because of them.

I understand aperture size, subject to camera distance and focal
length and I understand that all of these change compared to APC
size sensor in order to get the same FOV. But how does the size of
the sensor contribute?

Thanks

--
EOSMan (got print button?)
http://www.pbase.com/eosman

 

[RichO]

Phil has a good albeit limited glossary which is like a FAQ.

One has to know what to look for when searching a FAQ or glossary.

The subject "Focal Length Multiplier" is appropriate here:
http://www.dpreview.com/learn/?/Glossary/Optical/Focal_Length_Multiplier_01.htm

Here is another one for Canon EOS - beginner's FAQ
http://photonotes.org/articles/beginner-faq/

--
RichO
http://www.pbase.com/richo
http://www.pbase.com/richo/txwildflowers

 

[RichO]

http://www.dpreview.com/learn/?/Glossary/Optical/Focal_Length_Multiplier_01.htm

--
RichO
http://www.pbase.com/richo
http://www.pbase.com/richo/txwildflowers

 

[artur5]

The only DOF change directly affected only by sensor size is the
CoC. The one you are pointint out is not just because of sensor
size, but both sensor size and distance to the subject.

Of course the perspective changes, but my point is not to use dif.focal lengths to get the same view from the same distance with FFrame and APSC sensors. The fact is people says that lenses have less DOF with FFrame that with APSC. That's not true strictly speaking, rather on the contrary, but the fact that to take a portrait you shoot 3 feet away with a 5D and 5 feet away with a 30D, implies practically that you get less DOF with that lens on the bigger format. Is the way you use the lens what changes DOF.