Quite Simply - Depth of Field

[Alex Mack]

Like?

How tiny it is. My dream sensor would be like an inch or so.

2/3s is pretty small. But workable in a way 1/3 is not.
There is enough space for four 1/3 sensors in a 2/3...

 

[John Nguyen]

Wow, the 2/3 sensor looks significantly smaller than Super16 to me, which is what I thought everyone meant when they said "the 2/3 sensor will give you DOF workings almost exactly like 16mm film." Just a guess, but it seems to me the 2/3 sensor is more like a full 3 stops away from Red One sensor when it comes to apparent DOF rather than 2.5 like many have been saying.

 

[Pietro Impagliazzo]

How tiny it is. My dream sensor would be like an inch or so.

2/3s is pretty small. But workable in a way 1/3 is not.
There is enough space for four 1/3 sensors in a 2/3...

Yeah... Still tiny compared to S35.
Sometimes I get into the DOF debate too.

But after I remember great movies shot on 16mm (City of God) or Sony HDCAM cameras (Oldboy), remember the leap from my old camera sensor (1/4") and the crazy fast FPS I gonna get with it, it starts sounding super sweet.

According to the FOV Calc, crop factor related to RED S35 is ~2.5x.

 

[Eki Halkka]

Hey halsu, why don't you give an in depth explanation of focal length and depth of field.
Truth is, you're really good with that... (hv20.com)

I've thought about writing something like that, trying to explain the stuff in layman's terms, from practical standpoint... but it's a lot of work to do really ,-)

And there's good info on the web already...

 

[Charles Angus]

What kind of DOF is the human eye closest to?

My understanding is that optically speaking the human eye is about a 28mm f2.8-3.5.

 

[Harrison Hui]

Well, if I am not wrong, our eyes are a 50mm lens(35mm Still equivalent) F/2.1(in very dark conditions) and around F/8.3(in bright conditions). Each eye has a approximate resolution of 101mp and together they have a resolution of around 576mp and a refresh rate of 60 frames a second(rumours).

 

[Pawel Achtel]

DOF is all to do with the sensor size, focal length of your lens, and the aperture you are using.

That's incorrect. DOF does not depend on the sensor size. It does, however, depend on the CoC (Circle of Confusion).

 

[Gerald Menclik]

yup, now I AM confused ....

 

[Dave Gray]

That's incorrect. DOF does not depend on the sensor size. It does, however, depend on the CoC (Circle of Confusion).

It does in a way. As to cover a large sensor area you need to use an greater focal length. Which equals a larger aperture. Here is some aperture sizes over vairous format's which have the same FoV:

27 Degrees FoV
s35- 50mm/2 = 25mm
65mm- 100mm/2 = 50mm
2/3” – 20mm/2 = 10mm

Sensor size - Focal length/F-stop = Aperture size

The only thing aspects that really affect DoF are Aperture (which is affected by Sensor Size, The focal length, the actual aperture) distance to focus and as you said CoC.

I actually did my thesis on DoF adapters; so I had to do a bit on depth of field. I shall host it or paste relevant bits when find some time.

 

[Pietro Impagliazzo]

Gerald, optically speaking, there's only a single spot that is going to be completely IN FOCUS.



And the perception of the out of focus parts depends exclusively on the ability to determine when it's going out of focus. ie The blur is going to get discernible only when it's larger then the photosite or grain.

Longer focal lengths tends to enlarge the blur and make it seem that things are more out of focus.

Every article posted on this thread should help you have a clear understanding of this.

See this: http://toothwalker.org/optics/dof.html

The author gives an example of the same subject, with the same framing, with different focal lengths and at the same aperture.

When the background of the image taken with the wide angle lens is enlarged it is noted that it has the same amount of blur that the image taken with the telephoto.

However, we perceive the telephoto as having less DOF because the background blur is enlarged.

Hope it's clearer for you now.

 

[Dave Gray]

Here is the best description of CoC I found when doing my research. It's from one of Paul Wheeler's books.

Suppose an image in photographed on 35mm film and show in a large theatre, and we assume our audience is one-third of the way back from the screen and we know that the eye’s maximum resolving power is usually considered to be one minute of arc. We can now work out the size of the smallest dot on the screen that the eye can see in perfect focus. We can then measure the distance of our audience from the screen and, by applying a little trigonometry this distance and our one minute of arc, come up with a physical dimension on the screen for the largest dot the audience will see as a single, sharp point.
- Wheeler, P

 

[Robert Craig]

Ok lets make this simple for it is about the lens no matter the sensor if the lens says 85mm f1.2 it will ALWAYS be that no matter the sensor. If the lens is 14mm f1.2 it will ALWAYS be that. The only thing that changes with the sensor is the amount of image you see. Thats why small cameras the little point and shoots have really tiny numbers like 2mm-14mm and such but they are still the same DOF at 14mm f5.6 on that little cam as on a big camera having a 14mm f5.6 lens its very simple all based on the mm # and the aperture #. :smile:

 

[M.Paul El-Darwish]

What kind of DOF is the human eye closest to?

Wrong question ( I'll explain why later) but it depends on the size of the eye's iris . Like the iris of a camera, the human iris or pupil is largest when ambient illumination is very low and tiny when it's bright. In terms of nearest camera lens cousin. Here's what eyes are like:
Horizontal field of view 170º. Vertical field of view 90º.
Focal length is best simulated by a "normal' lens. Typically 500mm on a 35mm frame. Angle of sharpest focus 7º (remember that the eye's sensory array is spherical not flat like picture images. I don't even know how many datapoints the eye generates in a second. must be in the tens of millions. Human eye is roughly 40megapixels at the 7º point of focus.

Something I'll research some day is the 'image decay' of the visual cortex. That is, how much and how quickly data is forever lost that only takes a millisecond to 'see'.

The reason it's the wrong question is that the human eye, indeed, the visual process is so complicated, there's nothing man made that even comes close to it. I read a book by Colin Ware recently that just blew me away. Two eyes, processing information and tracking both still and moving objects in space, either while the person is standing still or moving are doing an amazing dance of information processing, leveling and interpretation and well as compensation that there are no parallels for in mechanics or optics.
Master cinematographers conjour feats of amazing illusion in film making, so as to both accommodate and fool our eyes and brains. Matching the capability of the visual sense is not one of those feats.:yikes:

 

[Gerald Menclik]

Hi, Pietro, I'm back ... got it .... was confused by the term CoC

 

[Pawel Achtel]

Resolving power of human eye

Resolving power of human eye

The retina contains two types of light sensing elements: rods and cones. While the rods are more sensitive to light, they do not deliver the resolving capability of the cones.

In the centre region of the fovea there are only cones, at a density of approximately 200,000 cones per square millimeter. Cones are spaced at approximately 2.5 μm and are about 1.5 μm in size.

Retina is approximately spherical, not flat as sensors, and the curvature is about 12.5mm.

At pupil size of about 3.5mm, the MTF at the retina yields the maximum resolution around 110 cycle/mm. This is luma MTF. Chroma MTF is significantly lower.

 

[Andy Taplin]

This is based on that article floating around that has the picture of the toy trucks called sky something or other?

Well, it works mentally, but its technically wrong. He doesn't say that for his comparison shots he moved backwards, and that this is what caused that 'space distortion' for the given frame of view, not focal length.

Things in the distance appear flatter, and focal length simply magnifies this optically to give us a field of view our eyes could not achieve. Focal length does not distort space! And slap anyone who would tell you so.:nads:

My comment was not based on an article floating around the web. If you change focal length but keep the subject size the same (by moving back and forward) and the f stop stays the same then the DoF will stay the same. DoF is not the same as Bokeh appearance which is greatly affected by magnification.

 

[Andy Taplin]

My comment was not based on an article floating around the web. If you change focal length but keep the subject size the same (by moving back and forward) and the f stop stays the same then the DoF will stay the same. DoF is not the same as Bokeh appearance which is greatly affected by magnification.

My last post was a bit simplistic, so here's another attempt to explain my understanding of the DoF thing with regard to sensor size as it's a quiet day in the office!

I'm going to ignore circle of confusion and diffraction as their effects are not something you can change when shooting and conceptually they need too much explanation. I'm also really only talking here about the rear DoF as it relates to background softness appearance (bokeh) which seems to be what people are most concerned about. So I'm also not going to mention hyperfocal distance even though it can be an important consideration for the DoF of a shot.

I'm not saying this is a comprehensive explanation (indeed it is quite simplified) but merely my attempt to articulate my own understanding.

Sensor size - affects Field of View but not Depth of Field or magnification directly (magnification not to be confused with crop factor). However, the sensor size is going to determine your FoV and therefore you lens choice and subject distance - which will affect DoF. So you could say sensor size affects DoF but only with regard to these other factors.

Focal length - affects FoV, magnification and perspective. Only affects DoF if the subject distance or point of focus remains the same as focal length changes. The DoF at most practical focal lengths can be maintained (more or less) by changing the point of focus. The distribution of the DoF will change with focal length. Magnification will affect the look of the BG bokeh.

Subject distance (point of focus) - affects DoF in combination with focal length as described above.

Aperture - affects DoF.

If you are shooting with a larger sensor camera such as a Red or a DSLR you can achieve that nice shallow DoF look to the BG much more easily than with say a 1/3 inch sensor. Why is this?

Because of their wider FoV, larger sensors allow you to be closer to your subject with any given lens - this is what decreases the BG sharpness, not the focal length of the lens itself. The closer your point of focus the more out of focus the background will be. Typically you will be using longer lenses for the same FoV with larger sensors which will also magnify the BG and change the perspective. Combined with the closer focus point this is what gives you the soft, magnified bokeh that people mean when they talk about shallow DoF.

The posts here are mostly saying the same thing but it's quite a difficult concept to explain simply. I'm sure someone can improve on my attempt

 

[deepfreezevideo]

I think Mr. Taplin has hit the proverbial nail on the proverbial head.

 

[Hubert Pietrzykowski]

Whenever I read those arguments about whether it's the sensor size that affects the DOF, the focal length / object distance or the CoC I can't help to smile. People argue even though they all are right (usually). It's all maths - equations expressing relationship between these four variables:

focal length
aperture value (f-stop)
circle of confusion (CoC)
object distance

Keep in mind it's the viewer who ultimately decides what's in focus and what's not and he does it by looking at the projection screen (or tv). Assuming the same projection screen dimensions and viewer distance the CoC diameter depends solely on the sensor size.
If you're unsure, think of the CoC as the maximum amount of blur that can be applied to the object to still be perceived as 'in-focus' by viewer (this blur is measured on the sensor or emulsion plane, in mm, and not the screen).
Now if sensor is smaller the image needs to be magnified more to fill the projection screen (compared to bigger sensor). But the CoC (emulsion-measured blur) is magnified more as well, making it more apparent. Thus, for smaller sensor the CoC must also be smaller to perceive the same image element as 'in-focus'.
The conlusion is that the CoC is proportional to the sensor size if the image is to be projected on the same screen size and viewed from the same distance. We can replace the CoC in the formulas by sensor size. Now we have four variables bound by equation:

1. focal length
2. aperture value (f-stop)
3. CoC or sensor size (alternatively, when screen size and viewing distance is constant)
4. object distance

Which of those are constant depends on the photographer's intention:

- if she changes the camera (sensor size) but wants the framing to stay intact, the magnification must stay constant which puts some restraints on the focal length / object distance relationship and (you'd need to do some math to prove this) there's only the f-stop and sensor size that affect the DOF.
- if she puts the same prime lens on both cameras and stands in the same place (same distance from object) it's only the CoC and f-stop that determines the DOF

The only reason why it is more correct to think of DOF as being determined by CoC (and f-stop) is that CoC automatically takes into account the projection screen size and viewer distance (because these determine the CoC).
In contrast, when you say that sensor size affects the depth of field you assume that projection screen / viewer distance are the same (for bigger/smaller sensor cameras), which is not necessarily true (16mm movies are often projected on smaller screen).
You need to take the projection parameters into account as well, and by saying the DOF depends on CoC you do it automatically.
One important thing to keep in mind when using DOF calculators is that you shouldn't simply choose the frame size (sensor size) from the dropdown list and calculate the DOF - you must be aware that by doing this you forgot about the screen size / viewer distance parameters (you are actually choosing the predefined CoC without knowing too much about the projection parameters it was calculated for). It is much more correct to putting the CoC manually knowing how big your screen is gonna be and how far your audience is gonna sit.

 

[Andrew Rieger]

Wow, the 2/3 sensor looks significantly smaller than Super16 to me, which is what I thought everyone meant when they said "the 2/3 sensor will give you DOF workings almost exactly like 16mm film." Just a guess, but it seems to me the 2/3 sensor is more like a full 3 stops away from Red One sensor when it comes to apparent DOF rather than 2.5 like many have been saying.

2/3" is very close to Super16, a tad smaller but not enough to effect DOF shallowness all that much.

This is a good article about the size differences of a host of digital and film formats complete with dimensions.

Link: http://en.wikipedia.org/wiki/Image_sensor_format