circle of confusion and resolving edges

[brian_carey_photography]

<p >I was introduced to the “circle of confusion” about 20 years ago when I read Harold Merklinger’s book “The In’s and Out’s of Focus”. I was intrigued by the science of focus and dept of field and was beginning to understand it; at least I began to appreciate out of focus objects (bokeh) in stills and in movies.</p>

<p > </p>

<p >I do understand that “if we want an object to be resolved make sure that the disk-of-confusion is smaller than the object”. But what I never did understand was, what about the edges of objects. Even when photographing objects larger than the circle of confusion the edges of these objects may appear soft even when the circle of confusion (in the same plane) is smaller then these objects.</p>

<p > </p>

<p >I always wondered how this all relates to edges of objects. I could be looking at this all wrong and I was wondering if someone out there in Photo.net land could help me out? I’ve been amazed for years about the wealth of photographic knowledge on Photo.net and look forward to creating some dialogue on this topic. I’m sure many of us will learn something.</p>

<p > </p>

<p >For anyone interested Harold has been gracious enough to put his book online and it is available as a free download, a small donation is requested. This book and his book “FOCUSING the VIEW CAMERA” is online. (sorry I am unable to post the url on here on this form).</p>

<p > </p>

<p >One thing I love about photography is that there’s always something to learn!</p>

<p > </p>

<p >Thanks</p>

<p >Brian Carey</p>

[Alan Marcus]

<p >The lens acts as a waveguide, it funnels light rays emanating from the subject, causing them to converge to a tiny point where they kiss off on the surface of the chip or film. This is best visualized by thinking about two ice-cream cones placed pointy end to pointy end. Now if the subject is in sharp focus, the surface of the chip or film will be exactly at the point where the two ice-cream cones touch. If the lens is perfect, this cone of light, where it hits, would be a tiny immeasurable dot of light. Sorry to report we never achieve such a point. What we get is a small circle. Worst, the small circle appears under the microscope, to have scalloped edges. This irregularity is due to uncorrected lens aberrations. We call this disk the circle of confusion.</p>

<p >Like the dots of ink or pixels that make an image, we want the circles of confusion to be tiny. How tiny? Think of a coin in a friend's hand. He/she recedes from you and the coin appears smaller and smaller with distance. At some point, the coin appears as a tiny immeasurable dot. That distance varies depending on eyesight 20/20 or 20/30 etc. However, a point will be seen when the coin is 2000 diameters away. As an example a 1 inch (25mm) coin when viewed at 2000 inches (50 meters) appears as a point to a person with normal vision. This is what we need to get an in-focus image. If we are going to make an enlargement, we need the circles tiny indeed. To make an 8x10 we need 8x more magnification from full frame chip or film. Thus we need the lens to form circles that are 8x smaller. </p>

<p >If the object is out-of-focus it appears fuzzy because the apex of the cone touches the chip/film, not at the apex but somewhere higher up or lower down on the cones. The result is a circle of confusion that is too big; it will not appear as a tiny point. If we are looking at a print or a computer screen, the remedy may be just step back and view the image from a distance. That is because the further we are away from the image, the smaller the circles appear. Consider that a mural mounted on the wall and viewed from a distance does not require such tiny circles of confusion as a print viewed at standard reading distance. </p>

<p >Most depth-of-field tables use 1/170 inches or 0.006 inches (that’s 0.0002mm) for the size of the circle. This takes into considering enlarging the image to make an 8x10. Consider that as you stop down a lens, the cone of light from a tiny aperture is skinner forming a more acute angle at the apex. Thus the film or chip can miss the apex and still sustain a tiny circle. This is why stopping down increases the zone of acceptable sharpness (depth-of-field). </p>

<p >As to the edges, this is a different phenomenon. Sharpness is defined as the steepness of the edge gradient, in other words the abruptness of change from light to shadow. The fact that light has a wave nature prevents a perfectly sharp shadow edge. In plain English, we always get some bleed that degrades the edges. On film this is worsened by the turbidity of emulsions and on digital, some adjacent interaction pixel-to-pixel.</p>

<p >Nobody said this stuff is easy, I call it gobbledygook. </p>

[brian_carey_photography]

<p>Thank You very much Alan I really appreciate you taking the time to explain this. Very much appreciated! Light does bleed somewhat eh and it even bends or curves somewhat! I guess I'm like many a photographer, we are visual people and in my case need a visual description in order to be able to see this!<br>

Thank You very much!</p>

[benbangerter]

<p>Alan - interesting explanation, but you need to check your math ;) 0.006 inches is about 0.15 mm. My impression is that, for 35mm film, a COC value of 30 microns, or 0.030 mm, is often cited. This corresponds to 0.01 inches when the image is enlarged to a (minimally cropped) 8"x10" print. When such an 8x10 print is viewed from a distance of 12 inches, this about 0.05 deg (or 3 minutes) of arc. Visual acuity is a complicated subject, which depends on many facors. Call it what you will...</p>

[charles_stobbs3]

IMHO, edge sharpnes is also affected by camera motion and subject motion and a fast shutter speed is the best solution to both.

[tom_mann1]

<p>Excellent article here:<br /> http://en.wikipedia.org/wiki/Circle_of_confusion</p>

<p>Excellent on-line DOF calculator that allows you to set the diameter of the acceptable COF here:<br /> http://www.dofmaster.com/dofjs.html</p>

<p>Note: In many circles (pun intended), the diameter of the largest acceptable COF is taken as the sensor / film diagonal divided by a number in the range 1500 to 1750. (Zeiss formula - see the Wikipedia article)</p>

<p> </p>

[Alan Marcus]

<p >Sorry I made a clumsy mistake converting inches to millimeters. Thanks for the help. </p>

<p >The desirable diameter of the circle of confusion must be based on the resolving power of the human eye. This is a variable of eye capability, mainly of age, and the brightness of the viewing area. Assuming the image will be viewed in good light, and the eye is well-corrected, the most widely used standard revolves around 1/1000 diameter. This distance/diameter combination sustains an angle of 3.4 minutes of arc. This is the equivalent of viewing a circle 1/100 inch in diameter from a distance of 10 inches. If the image is to be viewed at 20 inches, the permissible size of the circle works out to 2/100 of an inch. The dots of ink in the funny papers do not achieve this target. </p>

<p >Now we are talking about finished images. If we talk about the negative, we must factor in the magnification needed to arrive at the display size. For a full frame 35mm negative to produce an 8x10, we enlarge 8x. Now the permissible circle size must not exceed 1/800 inch.<br>

It is common practice to state the size of the circle of confusion as a fraction of the focal length. This method has an advantage. It takes into consideration the degree of magnification, assuming the print will be viewed from the "correct viewing distance". This is a distance about equal to the diagonal measure. We tend to gravitate to this viewing length. Kodak strived for some of its lenses, a circle of about 1/1750 of the focal length</p>