Depth of Field

I do not agree compeltely, of course the depth of field blurring it’s not a “trick” because it’s normal to have (for example) far objects blurred and near object in focus using a normal camera, but it’s a trick if you use it to focus attention.

And YES there are cameras able to shoot EVERYTHING in focus, without waiting for long, I mean movies, not pictures.

Anyway, nobody can say the DOF “blurring”(unfocusing?) isn’t used in movies. They use it! And a lot! And they do it to “focus” your attention to a particular object/person in a particular moment (I’m involved in movie making and I’m sure about it).

I just wanna discuss if it’s worth the work coding a depth of field blurring in a software to focus the attention of a user on a particular zone of the scene or not ??

IMHO in some situation it’s really a good effect.


I think I’m getting to the end of my english. The user has its own region of high resolution, which is defined by the depth he is focusing on (which has to be simulated by depth of field) and the point he’s looking at (what I called point of concentration). So it’s basically realistic to simulate the depth of field effect for a human viewer (not a camera!), when you track the current user’s eyes for it. For camera effects, these depth things can be presaved.
My question was, if it is possible to use LCD glasses (no shutter!!!) to simulate an image, that is greater than the fov of normal human eyes. If so, that would be really great.

And, for normal eyes it may be normal to have depth of field blurring, but depth of field is relative! The human eyes can focus on a nearly unlimited number of layers in depth, so you can’t say a certain layer needs to be blurred, since you don’t know which of the layers the user is currently looking at!

By saying its not a trick, i meant it’s not something that a post-production house works on. You don’t see lucas giving some film to ILM and say “add some DOF blur. cheers”

are you sure about motion cameras that can capture completely focused images? I mean, not “i think i’ve seen a movie that loooked pretty focused”, but REALLY sure? I ask, because the DOF effect IS DUE to the fact cameras use lenses. I mean, any book on optics will tell you this, and i do not belive that they use some camera system (if it does, in fact, exist) to film entirely in focus images.

I could probably even draw an ascii diagram of the effect. you have a point source irradiating light in
all directions. more than a single photon beam is going to to find its way through the appature, and you NEED the lens to converge all this light into a single point. But you can ONLY do this for a single plane from the camera. All other planes will have a refractive error. The error for a point is given:

refractive error = | 1/d + i - a |

er, i’ll just cut and paste bits of my thesis here. Although I’m talking about bad eye-sight, its just (!) and exaggerated form of what happens in a lens system:

The majority of cases, poor eyesight can be attributed because the optical propor-tions
of the eyeballs are awry [12]. The eye is much like a camera. It has a pupil
that corresponds to the camera’s aperture; a cornea and lens that corresponds to
the camera’s lens; and a retina that corresponds to the camera’s film. The distance
between the retina and the lens is the focal length of the lens system.
Nature abhors the exact geometry required for a perfect optics system. In
the case of myopia, or short-sightedness, the eye is longer than normal. Distant objects are inevitably out of focus, but near objects can be brought into focus
by ‘pulling out’ the optical system, ie. so the retina is further away from the
lenses [12]. In the case of hypermetropia the eye is shorter than normal. The eye
can compensate for this by utilising some of its internal focusing power usually
reserved for near-vision to reduce the focal length of the system for a clear retinal
image; but this leaves less reserve for focusing on near objects, and thus the eye
is long-sighted [12].
A simplified optical system is illustrated in Figure 4.4. The ability to focus
light from a point source to a sharp point on the retina (or focal plane) by an
optical system is known as accommodation. Those with poor eye-sight have a
reduced range of accommodation. In these circumstances, the eye tries ‘as best it
can’ to project the ‘focused’ 1 plane onto the retina with the least amount of blur.
The smallest disc obtained by the converging light is known as the circle of least
There are two techniques for simulating refractive errors. In the first, an op-tical
system designed to blur is placed directly in front of the observer’s eyes.
Depending on the observer’s accommodation range, the eye may be able to bring
the image into focus. This technique is known as the “observer” method. The
second technique is to defocus an image as it is projected onto a screen by placing
an optical system in front of the user. The eye is unable to bring the image into
focus because of the intervention of the screen. This technique is known as the
“source” method.
A metric for the level of defocus in [1], is the defocused point-spread function.
The level of defocus is measured as the size of the image it forms on the retina,
known as the defocus blur disc (or simply ‘blur disc’). In practice, the light dis-tribution
is affected by aberrations, diffraction, the refractive error and the pupil
size [1].
! = D L (4.1)
If the complexities of factoring aberrations and diffraction into the light distribution function are ignored, then the size of the defocus disc is cited in Ef-fectDefocus
as equation 4.1. Where D is the entrance pupil diameter and L is
the refractive error for any level of accommodation. The result, !, is the angular
diameter of the blur disc, ie the angle subtended from the eye to the blur disc.
L = j
d + i

didn’t come out well, but check out:

Chan C, Smith G, Jacobs RJ 1985: Simulating refractive errors: source and
observer methods. American Journal of Optometry and Physiological Op-tics,
Mar;62(3), 207-216.

yes, you do need to track the plane the user is accomodating (the a in the eqn above). we used a cursor system because we didn’t haev eye-tracking h/w


Well, I think I misexplained myself, but you mentioned very interresting things! No doubt that the eyes has depth of field blurring, and that it would have to be simulated, but as you said you would need to know at which distance the eyes are looking, and they could only be looking realisitcally if you used 3d glasses. So, for eye simulation on a normal monitor, I think the effect is not very good. In cool 3d glasses it is cool and realistic. For both it is realistic if you need to simulate a camer (cut scenes etc.).

Yes, I’m sure that exist a camera that has no DOF blur, dunno how it can be done, but I’ve seen it in action some time ago, it’s a Sony product if I remember.

What I can’t understand is that why you accept a movie (2D) that has a DOF blurring effect to focus the attention, but say that the same effect is not so good in games (again 2D).

With “trick” I do not mean a digital effect, but just a way directors use to focus attention on a particular spot in the scene, even with an analogic camera.

Why it’s the most used effect in movies and it should be wrong for digital graphics ??

Isn’t it always 2d ?


I accept it also for cutscenes in games, because these are like movies. But if you play the game in 1st person (which is for me a human, and no camera), the game can’t bet where you’re looking at, so the user may see important (at least for him) details blurred.

An example would be: YOu’re driving down a street and there’s glass on the road (you would like to avoid driving over it, so you must look directly ahead of the car). But in the far distance there’s going on something really interresting. Now, the good drive would still focus on the glass but the bad driver would look at the interresting thing. If you would be the game, how would you decide to blur the scene (the human driver can only focus on one of them!)?
Hope you get the idea.

If that scene is however a movie (so it’s not interactive), it wouldn’t be bad if the game would blend the scene, since it’s normal for movies to be seen out of the sight of a camera (which has manual dof).

Of course I never ment a first person game, as it simulates a “real” view from eyes, but imagine a real time strategy game like Warcraft (for example), if one of your party is attacked by enemies the engine focuses attention just on that player for some seconds using DOF blurring.
I think it could be a good effect.


Wouldn’t that be a bad example, since you don’t see a great near->far gradient in a world which is displayed from top (or nearly) like strategy games are. And what if the player is attack from two sides?