PDA

View Full Version : Diffraction and Lens Flare



Paul Mongillo
9-Mar-2000, 12:51
I am a little embarrassed to ask this question. I have a fairly broad backgroun d in photography starting as a miltary photographer in 1968. I have been into bl ack and white 4x5 for about a year now and love it. I have read everything I can get my hands on and enthusiastically try new stuff. I have not really come acc ross any explaination of diffraction and flare that I understand enough to look for it on my negatives.

What do I look for on my negative to determine if I have stopped down too much ? Does everything begin to get fuzzy ? Perhaps I need to do another set tests fo r my lenses. I tend to shoot at f32 or 45, but am tempted occassionaly to shoot at f64 (Nikkor 90mm f8 and Fujinon 150 f5.6)

I can readily see flare in the form of strange images on my negatives when light starts bouncing around between the lens and filter. I also see the outlines of the shutter in a backlit shot, but how do I know if I am getting a more subtle general flare(not sure thats the right way to describe it). I had an experience shooting some 3200 film of my son playing hockey a couple of weeks ago. I was very carful to meter the rink and players and set the focus and exposure on my 3 5mm Pentax and 135mm lens. The lens is at least 35 years old with a blue tint t o it (has a lens hood). I assume its not multi coated. All of the photos, even the ones I thought would be overexposed were flat. The several I took elswhere with my 28mm multi coated lens (same roll of film) were great. Is that flatness caused by lens flare?

Any suggestions would be appreciated including reading sources.

Thank you in advance.

Glenn Kroeger
9-Mar-2000, 13:06
Diffraction effects will appear as a gradual softening of edges and reduction in resolution of fine detail as the lens is stopped down. In 4x5, diffraction begins to be detectable (but not necessarily objectionable) at between f/32 and f/45. Whether it is a problem depends greatly on subject matter and degree of enlargement. I try to avoid apertures smaller than f/32 WHEN POSSIBLE. If depth-of-field is essential, f/45 and even f/64 are still usable but will soften edges and reduce fine detail in the image.

Ron Shaw
9-Mar-2000, 13:25
The flatness could indeed be caused by flare. The hocky ring probably has a lot of white (bright) in the shot, which is where uncoated (and possibly single coated) lenses start to have problems with flare, and a reduction of contrast, even though you dont see obvious 'flare' artifacts. In a studio setting, where I can control the lighting, my uncoated lenses give me contrasty, saturated images like my coated lenses, but if its a high key shot(or white background), I cant use the uncoated lenses, as the contrast goes out the tubes.

N Dhananjay
9-Mar-2000, 15:21
Flare (the non-obvious variety) essentially distorts the toe portion of your films characteristic curve. Phil Davis talks about how can test and try to minimize its effects. Cut a small hole in a sealed cardboard box. Mount little cardboard strips to form a window lintel all around the hole so the hole is completely shaded from any direct light. You now the photographic equivalent of a black hole. Take a picture of this box. Process normally along with an unexposed sheet. The unexposed sheet gives you your film base+fog density. The density of the 'black hole' gives you the added density in the shadow areas due to lens flare. Typically, a little extra processing time (the exact amount can be determined by running tests and calculating the gammas of your curves) will give you the equivalent of the same contrast with slightly higher FB+F. I would suggest its not worth doing unless you have a lens that you suspect is really flar-y. I've never run any tests but would suspect that the amount of flare must show some correlation with lighting situation. Most modern coated optics should allow you to get reasonable pictures with proper technique (using lens shades etc). The jump from uncoated to single coated is quite noticeable and the jump from single coated to multicoated less so, at least to my eyes.

Diffraction is pretty much a function of f stop, regardless of format. However, the lower enlargement ratios with larger formats make it less of an issue with them. However, the more you stop down, the closer you're getting to a pinhole i.e., you're using less of that expensive glass. The usual rule is that 2 stops down is where the optics balance out i.e., you correct some aberrations but your diffraction increases as you stop down and a couple of stops down is the optimum. This may not be the case with some older LF lenses where the largest aperture is really to facilitate brightness and focusing on the ground glass. You can test for this by shooting the same test pattern at various f stops but why bother. There is no way out. If your DOF demands dictate small f stops, you will suffer diffraction softening and earlier limits on print sizes. At some point, you might as well have used a pinhole. At the risk of generalizing, somewhere between f/16 and f/32 is probably optimum for LF. However the lower enlargement ratios mean that f/64 might work out fine in some situations, depending on your print size, optics etc. One area where diffraction is a real concern is with macro work because the extension is long enough that your effective f stop is much smaller than marked. I ran a test once and saw definite softening due to diffraction at smaller apertures and this was on film, not even on the print(which means I should have seen it on the ground glass if I had been paying enough attention). I would definitely start to worry about anything beyond f/22 when shooting macro. The intelligent positioning and use of movements might help you get sufficient depth of field at wider apertures. Occasionally, there is no easy way out and then you're out of luck. I suppose this should be possible to see on the ground glass as you vary the aperture (and using a loupe of the appropriate magnification should let you decide on enlargement limits) but the simultaneous variation of brightness makes it quite difficult for me to spot it. So I fall back on these rough heuristics. Hope this helps. DJ

William Marderness
9-Mar-2000, 15:44
As the format size increase, depth of field decreases and diffraction decreases at any given f-stop. Assuming that all formats will be printed to the same size, f16 in 35mm equals f64 in 4x5, which equals f128 in 8x10. So if the optimum aperture for 35mm is f5.6, the optimum aperture for 4x5 is f22, and for 8x10 is f 45. See this lengthy explanation:

Title: Dof is Squelched by Diffraction Author: Michael Davis Date: 1998/08/15 Newsgroup: rec.photo.misc

Diffraction is the only aberration suffered by pinhole cameras. It is an image degrading phenomenon for which there is no means of correction. The smaller the aperture through which light must pass, the greater the effect and the more an image at the film plane must be magnified to create a print of a given size, the more visible the degradation in that print.Small formats are proportionately more vulnerable to the effects of diffraction than larger formats, so much so, that the smallest formats, APS and to a lesser degree, 35mm, can not exploit the Depth of Field advantage they have over large formats at their smallest available apertures. They are diffraction-limited to using wider apertures than those which can be used by the larger formats.

Depth of Field is squelched by diffraction and the point at which this happens moves to smaller f-number values (wider apertures) as the format diagonal decreases. The good news is that at the diffraction-limited apertures for each of the formats, the exact same depth of field can be achieved, with the larger formats having a disadvantage of longer exposures. Let's look at why this is true.

John B. Williams authored a book called *Image Clarity, High- Resolution Photography* that has a discussion of diffraction. He gives the following formula that calculates the radius (r) of an Airy disk. (G.B. Airy is the astronmer who discovered diffraction in 1890 and diffraction's disks are named after him, Airy, not airy.) I can't type a lamda, so I have substitued a "w" in the formula below, for wavelength. "f" is for focal length and "a" is for diameter of the aperture.

r = 1.22w(f/a)

OK, f/a can be renamed N where N is the familiar f-number that describes the ratio of focal length to aperture diameter. That gives us this:

r = 1.22wN

In his book, Williams selects 0.0005 mm as an average wavelength of light, but I prefer 0.000555 mm, or 555 nanometers as being the wavelength that is dead center in the spectrum of sensitivity. It happens to be a nice yellow-green, not far from William's choice anyway.

OK, moving on, that gives us this formula, using 555 nanometers for all future calculations:

r = 1.22 * 0.000555 * N

r = 0.0006771 N

To convert this formula to diameter (d) instead of radius (r):

2 * r = 2 * 0.0006771 N

d = 0.0013542 N

This is the diameter in millimeters.

It was at about this point that I decided I didn't like the fact that William's formula had so few significant digits in the constant 1.22, so I went searching and found a longer, more accurate version of it and here it is -- infinitely accurate: __ 1.21966 (66 repeating)

Not much different from 1.22, but it does change my formula for diameter of an Airy disk to this:

d = 0.00135383 N

Previously, I introduced a bit of myself, so to speak, when choosing 555 nanometers as the average wavelength for calculating the diameter of Airy disks and now I would like to state that I believe 1/175-inch is a good, aggressive choice for maximum permissible circles of confusion when doing depth of field calculations and thus, it is also my choice for the maximum permissible diameter of Airy disks. The reason this is expressed in fractions of an inch instead of millimeters is because that is the convention for discussions of circles of confusion and that convention also adheres to stating such diameters not at the film plane, but rather at the print, after magnification from a negative, and that print size is a print with a 10-inch diagonal that is expected to be viewed at a distance of ten inches.

More specifically, the viewing distance is measured from the eye to any equidistant corner, while centered over the print, not from the eye straight down a line perpendicular to the plane of the print. If the two of us are discussing circles of confusion and are both adhering to this convention, we will be comparing apples to apples even if you use 8x10 and I use 35mm and better still, if you decide to make a 16x20 print and I decide to make a 32x40 print, as long as viewing distances are equal to the print diagonals in each case, they will both have the same perceived sharpness (in so far as depth of field can effect sharpness) if we have both chosen the same value (i.e. 1/175-inch) as the maximum permissible diameter for circles of confusion for a 10-inch diagonal print to be viewed at 10 inches. Our mutual decision to limit circles of confusion to 1/175-inch in a 10- inch diagonal print would limit CoC's at the film plane to 0.024724 mm for my 35mm, but your 8x10 could permit CoC's at the film plane that are much larger, 0.185874 mm. Both formats would however, deliver the same illusion of depth of field after magnification to a given print size, at a given viewing distance.

Quoting page 131 of "Basic Photographic Materials and Processes" by Stroebel, Compton, Current, and Zakia (c1990 Focal Press): "Permissible circles of confusion are generally specified for a viewing distance of 10 inches, and 1/100 inch is commonly sited as an appropriate value for the diameter. A study involving a small sample of cameras designed for advanced amateurs and professional photographers revealed that values ranging from 1/70 to 1/200 inch were used -- approximately a 3:1 ratio."

It's somewhat subjective, but I like 1/175 inch, toward the more critical end of the range used by manufacturers. The larger the value you specify for the denominator, the more conservative your calculated depths of field will be. The rotating-disk Depth of Field calculators published by Kodak in their Photoguides use a generous, less critical value of 1/100 inch.

Using 1/175 inch, the maximum tolerable diameter of circles of confusion for a given format can be calculated as the format diagonal divided by 1750. (There would be 1750 circles set end to end along a print diagonal that is 10 inches in length.) As discussed above, the diameter of Airy disks is calculated as 0.00135383 * f-number.

If we can calculate the aperture at which Airy disks become 1/175 inch diameter when the format is enlarged or reduced to a 10-inch diagonal print, we will know the aperture at which it is pointless to make circles of confusion any smaller than 1/175 inch. This will be the aperture at which a quest for more Depth of Field should be conducted using techniques other than going to a smaller aperture (increasing the subject distance, using tilts and swings, etc.)

Here we go: To set the size of the Airy disks equal to (and no larger than) the tolerable diameter for circles of confusion for any format after magnification or reduction to a 10-inch diagonal print to be viewed at 10 inches, I just have to equate to 1 the quotient had when circles of confusion diameter is divided by Airy disk diameter, then reduce.

1 = (Format Diagonal mm / 1750) / (0.00135383 * f-number)

or

f-number = Format Diagonal mm / 2.36920501777

Tah-dah! My formula makes two assumptions. This constant is specific for yellow-green light at 555 nanometers and we don't want our Airy disks to exceed 1/175-inch on a 10-inch print viewed at 10 inches. You may modify the constant proportionately if you want to change the values 0.000555 for wavelength or 1750 for the number of Airy disks set end-to-end along a 10-inch diagonal print.

Here's the formula modified to allow specification of a diameter other than 1/175-inch:

f-number = (Format Diagonal,mm / 0.01353831438675) * Max. Disk Diam.,in.

But for the remainder of this disuccsion, let's stick with 1/175-inch as our maximum permissible diameter in a a 10-inch diagonal print to be viewed at a distance of 10 inches.

So, this calculated f-number is the aperture at which diffraction's Airy disks would have a diameter of 1/175th inch in a 10-inch diagonal print. As long as the viewing distance is equal to or greater than the print diagonal, there would be no visible evidence of diffraction, no matter how large the print is. If however, the viewing distance were to be cut to one half the print diagonal -- say a viewing distance of 12.5 inches for a 16x20 print, then the aperture number would have to be cut in half.

For example, using the formula above, the f-number at which the 8x10 format would begin to show evidence of diffraction in a print viewed at a distance equal to its diagonal is:

312.51 mm / 2.36920501777 = 132

So, according to the math, we can use f/90, but not f/128 (very near f/132).

If we know in advance that we'll be viewing the final print at half the print diagonal, we have to cut the f/number in half -- in this case, from f/128 to f/64 -- a two-stop difference. In this case, f/45 would be acceptable, but not f/64.

Another easily overlooked point is that since this the formula uses format diagonal, if you know in advance that you will be cropping to use only a portion of the full image area, you should use the resulting cropped diagonal to calculate the f-number at which diffraction effects become visible! The smaller the diagonal, the greater the effects of diffraction because of the increase in magnification necessary to yield a given print size.

Format diagonal and maximum permissible diameter for the Airy disks are the only variables for determining the f-number at which the effects of diffraction become visible. Using the formula given above, where Airy disks will be limited to a diameter of 1/175 inch in a 10- inch diagonal print, and which will be found equally acceptable in any size print as long as the viewing distance is equal to or geater than the print diagonal and where the full format diagonal is used, without cropping, I get the following values for these formats:

Format Full Diagonal Diffraction No Diffraction Visible at Visible at

APS 34.51 mm f/14.56 f/11 35 mm 43.27 mm f/18.26 f/16 4.5x6 cm 69.70 mm f/29.42 f/22 6x6 cm 77.78 mm f/32.83 f/22 + 1/2 stop 6x7 cm 87.46 mm f/36.92 f/32 6x9 cm 102.08 mm f/43.09 f/32 + 1/2 stop 4x5 in 153.67 mm f/64.86 f/45 + 1/2 stop 5x7 in 208.66 mm f/88.07 f/64 + 1/2 stop 8x10 in 312.51 mm f/131.9 f/90 + 1/2 stop 10x12 in 383.47 mm f/161.9 f/128 11x14 in 447.78 mm f/189.0 f/128 + 1/2 stop

Now let's generate the same table, but this time using the cropped image diagonals that would be used to produce 4:5 aspect ratio prints, (8x10, 11x14, 16x20, etc.) instead of the full format diagonals. Since the diagonals are smaller in some cases, where the full format diagonal is not already a 4:5 aspect ratio, the resulting diffraction limits occur sooner, at wider apertures!

Format 4:5 Cropped Diffraction No Diffraction Diagonal Visible at Visible at

APS 26.73 mm f/11.28 f/8 + 1/2 stop 35 mm 38.42 mm f/16.22 f/11 + 1/2 stop 4.5x6 cm 66.43 mm f/28.04 f/22 6x6 cm 70.43 mm f/29.73 f/22 + 1/2 stop 6x7 cm 87.08 mm f/36.75 f/32 6x9 cm 88.04 mm f/37.16 f/32 + 1/2 stop 4x5 in 153.67 mm f/64.86 f/45 + 1/2 stop 5x7 in 193.69 mm f/81.75 f/64 8x10 in 310.55 mm f/131.1 f/90 + 1/2 stop 10x12 in 377.78 mm f/159.5 f/128 11x14 in 442.18 mm f/186.6 f/128 + 1/2 stop

If viewing distance is one half of print diagonal, open up two more stops. An acceptable aperture of f/16 for uncropped 35mm must be opened to f/8 if the print will be viewed at a distance equal to half its diagonal!

At the top of this article I stated that at the diffraction-limited apertures for each of the formats, the exact same depth of field can be achieved, with the larger formats having a disadvantage of longer exposures. We've got the foundation to look at that, now.

Everybody laments that an 8x10 has less Depth of Field than a 4x5, than a 6x7, etc., assuming they are using equivalent focal lengths and people argue that tiny formats like APS offer more depth of field, but thanks to diffraction, the small format DoF advantage is squelched.

The achievable diffraction-limited Near Sharp distances are IDENTICAL for all formats, given that the ratio of focal length to image diagonal is the same from one format to the next. In other words, if several formats are using focal lengths that are equivalent in their ratio to the format diagonals, the effects of diffraction will limit each format to a unique minimum aperture at which diffraction becomes visible AND it turns out that if you calculate the Depth of Field for each focal length/image diagonal pair AT THOSE UNIQUE APERTURES, you'll find that ALL the formats can achieve the SAME Near Sharp (without movements and at different f/stops, course). The only disadvantage had by the larger formats is the longer exposure times necessary to reach their diffraction-limited apertures. The smallest formats, can not use their smallest apertures, but they too can achieve the same Near sharps had by the larger formats using the apertures that aren't diffraction limited for them. They have no depth of field advantage, only the advantage of shorter exposures to get the same depth of field larger formats have with longer exposures. (I'm compelled to mention here, that aside from the issues of depth of field and diffraction, the larger formats benefit by all that comes with having less magnification to get to a given print size and until someone makes a 35mm with full movements, the larger formats also benefit by using movements to control the position of the focus plane and perspective.)

In the table below, the third column (Near Sharp at f/22) was calculated using a maximum permissible diameter for Circles of Confusion of 1/175th of an inch. The fourth column (Largest Aperture with Visible Diffraction) was calculated with aerial disk diameters of 1/175th of an inch, also. This reduces to the equation:

Format Diagonal in mm / 2.36920501777 = Aperture where diffraction becomes visible. Focal length is not a variable for this calculation.

Format, Focal Near Sharp Largest Near Sharp at Using 4:5 Length Distance Aperture Largest Aperture Aspect Ratio (mm) at f/22 With Visible With Visible Diagonal (feet) Diffraction Diffraction (f/stop) (feet)

APS 24.6 3.0 11.3 5.7 35mm 35.3 4.2 16.2 5.7 6x7cm 80.0 9.6 36.8 5.7 4x5in 141.2 16.9 64.9 5.7 8x10in 285.3 34.2 131.1 5.7 11x14in 406.2 48.7 186.6 5.7

First, notice that at f/22, the Near Sharps are much closer for the smaller formats. (The Far Sharps are all nominally at infinity.) So you can see that 8x10 has half the Depth of Field enjoyed by 4x5, with a resulting Near Sharp that's twice as far from the camera.

But, also notice that the last column calculates the Near Sharp distances for each format at each format's largest aperture with visible diffraction. The Near Sharps at THESE apertures all work out to be exactly the same! For these focal lengths, they are all 5.7 feet. And notice that thanks to diminished diffraction, the larger formats can bring their Near Sharps to that had by APS and 35mm, just by stopping down to apertures where small formats should not follow.

Diffraction limits them all to the SAME Near Sharp. Large Format can get just as much Depth of Field as small format because diffraction is getting out of the way exactly in proportion to the loss of Depth of Field! Guess what? At the diffraction-limited apertures for each format, the size of the hole the light passes through is identical in proportion to the format diagonals! That's why this whole discussion is true.

The above chart illustrates the fact that the impact of diffraction is diminished linearly just as the Depth of Field diminishes with increase in format diagonal.

So, to avoid diffraction, the astute 35mm photographer stops down no further than f/16, the APS photographer no further than f/11, etc. and the depth of field promised at apertures smaller than f/16 can never actually be enjoyed. With a focal length of 35.3mm, the 35mm format can not enjoy a Near Sharp closer than 5.7 feet. BUT, this figure holds true for EVERY format using equivalent focal lengths, at the diffraction-limited minimum apertures for each format. They can all achieve a Near Sharp of 5.7 feet when stopped down to their respective diffraction-limited minimum apertures. Obviously, the larger formats will need longer exposures to achive this Depth of Field (independent of movements) and the 11x14 format would be hard pressed to find a lens with f/180.

So, in summary, I contend that even without their movements, large- format cameras CAN achieve the exact SAME Depth of Field had by the smaller formats (with longer exposures.)

Ron Shaw
9-Mar-2000, 19:10
Now arent you sorry you ask?

Glenn Kroeger
10-Mar-2000, 03:35
Williams repost of Mike Davis analysis is, of course correct, but hidden amongst the equations and tables is the one "gotcha" and that is: "Obviously, the larger formats will need longer exposures to achive this Depth of Field..." So if the leaves are moving, its much harder to get that DOF with LF.

At the optimum balance between DOF increase, and diffraction losses, all formats do achieve the same depth of field, but since fstop goes as focal length of lens... so if your 4x5 lenses are three times as long as your 35mm lenses, you need 3 stops smaller opening therefore 3 stops longer exposure, or 8 times the film speed.

Fortunately, grain doesn't increase proportionally to film speed in some cases. For example, Fuji Provia 100F pushed to EI 200 is not perceptably grainier than Velvia at ISO 50. So instead of shooting Velvia in 35mm, you can gain back 2 of those stops with the film, then only contend with double the exposure. I assume there are similar choices in B&W where some faster films are nearly as fine grained as slower choices.

Chris Partti
10-Mar-2000, 13:25
The Michael Davis analysis is intersting in a theoretical way, but not all that useful practically. What I understand it to say is that assuming a constant print size, for a given level of defraction degradation, DOF across formats is a constant. I see a couple practical problems with this: (1) Defraction may not be the only relevant limiting factor on achieving DOF. As Glen Kroeger points out, achieving the same DOF with larger formats requires longer exposures. These exposure times often get into the range of serious reciprocity breakdown, requiring exposures for larger formats that might be many minutes versus seconds for smaller formats. Such long exposures may be impractical for some subjects, even stationary ones (for example, when the light is changing). There are also physical limits on the DOF that can be obtained by a given lens. For example, I might be very satisfied shooting at f/22 for enlargement to 8x10 in 35mm format. The equivalent for 8x10 format would be f/180, which none of my LF lenses will do. (2) Actually living by this rule (i.e. shooting LF at very small apertures to acheive DOF equivalent to smaller formats) raises serious questions about the benefits of LF. You CAN shoot at very small apertures but this means (a) you won't be able to make prints larger than you would with 35mm and (b) (as I understand it) your prints won't be any sharper than those made using 35mm. If so, it seems to me the rationale for carrying around all that heavy equipment (and waiting around for those reeeaaaaly long exposures) is seriously undercut.

What I do take from the Davis analysis is that I really don't have to worry much about shooting 8x10 for contact printing at f/64. I guess this is the "defraction equivalent" of f/8 for 35mm, and I, for one, have never noticed defraction effects on 35mm 8x10 enlargements shot at f/8.

Ron Shaw
10-Mar-2000, 14:05
Keep in mind that these diffraction comparisons are for using aperture only to achieve DOF. Using tilts is another powerful tool for achieving DOF with LF (or other formats offering tilt lenses). It can save you several stops. By the way, I shoot chromes exclusively in 4x5 (usually Provia at 100EI), and for most landscape type shots, I am at less than a second, and usually at around 1/5 sec. or faster (at f22-32 typ.), so you arent always relegated to many second exposures.

Bill Glickman
12-Mar-2000, 06:28
Some great input... I would like to add that not every shot is germane to LF. So to compare the examples above are a worst case scenario for LF.. i.e. shooting 35mm at f22, then acheiving the same DOF in 8x10. In that paticular shot, if movements can't help the DOF issue than I most likely would not shoot that paticular shot in 8x10. I shoot all formats, and I find this general rule always applies... any DOF close to the camera off to infinity is not a LF shot unless you benefit from moevements, otherwise diffraction will degrade the image equiv. to what 35mm or MF could produce. However, when you shoot on top of your car (Ansel Adams trick) you remove the near DOF way out there which enables you to open up and get away from diffraction limited f stops. So bottom line, sometimes smaller formats are just as effective as LF when their is extreme DOF, but when this is not the case, like focussing at infinity, the bigger the better!

Sean Billy Bob Boy yates
12-Mar-2000, 13:57
Paul,

Has anyone ever stopped dead in their tracks when looking at your prints and said to you, "My Goodness! You're stopped down entirely too much look at that diffraction!"

If the prints look good, and work for you and other people - the subjects, their friends and family, etc. what are you worried about?

Ron Wisner said it best in reference to his old un-coated Protars "No one has every said, when looking at my prints, 'This must have been taken with one of those 100 year old lenses you use'."