Ask Our Optics Team (Optical Frequently Asked Questions)

Thanks for your many questions! This page is completely devoted to our colleagues in the Machine Vision business who have optical questions. Some of these basic questions may be addressed in the Machine Vision article published in Laser Focus World by Dr. Jonathan Kane see (Machine Vision from an Optical Point of View). However there are always a variety of general purpose questions. We realize that there are many proprietary issues with this particular area so would request that you indicate whether or not we can post your question and our answer. In order to protect the privacy of those postings, we will strip the return e-mail and name and just post the answer. If you wish your name and e-mail to be posted here, please let us know.

What do these optical terms mean?

Well, you can either play "Buzzword Bingo" with our web pages or check our optics definitions page. Let us know if there are other terms you would like for us to define. We'll be happy to expand this list of definitions.

How it works:

Please send us your question. One of our designers will then post the relevant answer to the question. If you feel motivated to respond to another's post, we will also post that information as well wherever applicable. This is a moderated group so only questions deemed relevant will be posted here.

Disclaimer: Although every attempt will be made to provide an accurate answer and or allow free and open discourse in this chat board, COI is not responsible and cannot be held liable for any actions taken as a result of this discussion. By participating in the discussion you consent to the above.

Q: How can my Sony camcorder offer 360 X optical zoom? They also offer "zero lux imaging" how is that possible? Can you service my Sony Camcorder?

A: While it is not our standard question to handle questions about other companies products, we have received enough of these types of questions that we felt it was time to offer an answer here. Sony has introduced a new digital camcorder that offers a 20:1 Optical Zoom and a 18:1 digital zoom. Since the light goes as the square of the f/# it would be very impractical to produce a 360:1 zoom lens optically. Sony achieves the digital zoom by expanding the center of their focal plane array through the use of electronics with gain. Their "zero lux imaging" it achieved by using the long wavelength portion of the spectrum and using infrared LEDs to illuminate objects up to 10 feet away from the camcorder. In essence, they are not at zero lux but instead are using the LEDs to light the scene,, however, the ambient light (light in the environment) can be "zero lux".Since we do not have any association with Sony, please direct all questions regarding their camcorders directly to them or to your place of purchase.

Q : I've been reading a lot about Telecentric lenses. What are they? I was told they would increase the depth of field for our application. Will it? How do I increase the depth of field?

A: Telecentric lenses provide constant magnification for any object distance. They do this by having the chief ray be parallel to the optic axis. The result is that as objects are moved back and forth in front of the lens, they may go in and out of focus, but they will always be the same size. For gauging applications this is an extremely useful property. Deep round holes will look like deep round holes regardless of whether they are at the edge of the field or in the center. In a normal lens, perspective is introduced so near the edge those same holes will look elliptical.

Do they improve the depth of field? No.

We have had customers argue you with us on that point. They say that theory is all well and good but in the field they have observed an increased depth of field. However, to really tell experimentally you would have to start with a standard lens with the same f/number and compare it with a constant metric side by side. We'd be interested in any anecdotal evidence people have one way or another. Theoretically there is no difference.

If you want to increase the depth of field of any lens, the usual method is to stop it down (increase the f/number). Beware, however, the light throughput decreases as the square of the f/number so you will need more light proportionately and at really high f/numbers diffraction starts to take its toll so that resolution starts to suffer.

Q: You recently quoted me a custom lens design that is quite a bit more expensive than your standard lenses. Why?

A : Custom lenses are built in small lot quantities. Stock lenses are built in high volume. It's the same amount of time involved for both, except that in one case you end up producing a much higher amount of end product. This is the well known advantage of volume manufacturing that applies from automobiles to optics. It really is that simple.

Not so simple is determining whether you need a stock or custom lens. If you are in doubt, fill out our lens requirements form so that we can take a look for you. All lenses are compromises, so in many cases it pays to have a lens designed such that your important features are included for your application. These usually involve specialized magnification, distance of operation or resolution.

Q : When we talk about depth of field, I'm used to thinking about hyperfocal distance. Can I use the hyperfocal distance for a machine vision application? Why or why not? Can you explain what hyperfocal distance is and its use etc?

A:Hyperfocal distance came into vogue with the advent of photographic optics. It has this useful property; if a lens is focused on the hyperfocal distance, then the depth of field extends from infinity down to half the hyperfocal distance (see "Photographic Optics" by Arthur Cox). So for example a lens with with a 55 foot hyperfocal distance would mean that when the lens is focused on 55 ft, all objects will be in focus from 27.5 feet to 55 feet. It was useful to calculate this number in order to determine what the lens setting needed to be in order to maximize the depth of field for a given shot.

Mathematically, we can determine the hyperfocal distance as:

H = f + 1000f/N where f is focal length and N is the f-number.

For all intents and purposes most people just use 100f/N and drop the first "f". Now it turns out that the depth of field can be approximated by

Nearest Limit = H u / (H+u)

Farthest Limit = H u/(H - u)

where u is the distance to the object from the nodal point. These are the equations most people use when talking about depth of field, subtract the two above and you get the total depth of field.

Another expression that can be used however is :

Nearest limit = (fu)(f + CN)/(f^2+ uCN)

Farthest limit = (fu)(f - CN)/(f^2-uCN)

where N is the f-number, f is the focal length and C is the diameter of the circle of confusion. Note that as C gets smaller, so does the overall depth of field.

So that's how it WAS done.

Nowadays,

Figure 1 : Depth of field sketch

Machine vision designers don't use Hyperfocal distance too much because they generally are looking at object relatively close to the camera plane. So,the depth of field is given as (see Figure 1):

DOF = 2 x f-number-object x blur-circle

where the blur circle is represented by the arrow above.

In image space this can be rewritten as :

DOF = 2 x f-number-image x blur circle/m^2

If we take the blur circle and f-number as the width of a pixel (P) then this simplifies to :

DOF = 2 X (P/m)(P/m)

where m is magnification, P is pixel width. Thus by knowing the pixel pitch of the camera we can estimate the physical depth of field. Stopping down the lens improves the depth of field.

Q : What is distortion? How do you measure distortion?

A : Distortion is the amount the image shifts closer or further from the optical axis. The two most common are pincushion or barrel distortion where a uniform grid is either bowed outwards or inwards. Generally it is quoted in percent across the whole field. So for example if we have a lens with 1 % distortion and the idea is that the image should be located at exactly 400 pixels, instead it will show up at 404 pixels (barrel distortion) or 396 pixels (pincushion). The amount of distortion that you can tolerate in your software should be specified for a custom lens design.

Q : What is contrast?

Contrast : Contrast is defined as = (light - dark)/ (light + dark) . A contrast of 1 means modulation from full light to full dark. A contrast of 0 means the image is gray (no contrast at all). The modulation transfer function gives you the contrast at a particular resolution. As the resolution increases, the ability of the lens to fully pass a contrast image goes down due in part to the details of the lens design and due to diffraction.

Q : How do I figure out the basics behind what lens model I need?

A: We have tried to answer that in our technical library, click here to go directly to the article. Basically the easiest way is to first figure out your magnification and from a basic understanding of your object distance the focal length can then be derived. You then specify the lens in terms of the desired format (1/3", 1/2" etc) which is determined by the camera you have in mind.