D60's ( APS-C )sensor is too small to stop down below f/11?

I remember the thread. I still use f/16 and even f/22 on my 1D2, just as I did on my 10D, my Elan, and my FT.

Hi Mark!

The real issue here is whether or not the degradation caused by diffraction will be visible in the final print. A wee bit of math will give you the precise aperture at which this will occur.

This depends on four factors:

- How large the spread function is at the sensor (the size of the Airy disks before enlargement)

- Enlargement factor (the bigger the print the more visible will be the Airy disks)

- Viewing distance (a print that looks sharp at three feet can look soft at 10 inches)

- The resolving power of the human eye (somewhere between 5 and 8 lp/mm for adults with good vision at a viewing distance of 10 inches).

Here's a simple, but accurate formula, for taking all of this into account:

At what aperture will the effects of diffraction become visible in the final print?

Maximum N =
1 / desired print resolution / enlargement factor / 0.00135383

Let's imagine we're using the 8.2 MP Canon EOS 1D Mark II. It's sensor has a diagonal of 34.47mm. If you intend to make an 8x10 print that has a diagonal of 348.27 mm, your enlargement factor would end up being 348.27 / 34.47 = 10.1x.

Now what do you want for a desired print resolution? If you want the print to be as sharp as any human being is capable of appreciating at a viewing distance of 10 inches, you'll have to go for a print resolution of 7 or 8 line pairs / mm. That may be excessive in some people's opinion, but this is entirely subjective. It's your choice. A print that delivers only 4 lp/mm will look just as sharp at a viewing distance of 20 inches as an 8 lp/mm print will look at 10 inches. You have to decide what viewing distance you want to satisfy and how sharp you want it to look at that viewing distance.

For this exercise, let's assume a viewing distance of 10 inches, but let's also go with only 4 lp/mm at the print. That's about as "soft" as I personally am willing to go. If the viewer is farther away than 10 inches and/or simply can't focus that closely, it will look sharper. An eagle-eyed human capable of resolving 8 lp/mm at 10 inches will be disappointed, but that's the choice I'm making. Selecting 4 lp/mm, I'm hoping most people won't stand that close to the print.

So here's the formula:

Maximum N =
1 / desired print resolution / enlargement factor / 0.00135383

Maximum N = 1 / 4 / 10.1 / 0.00135383 = 18.3 (that's at about f/16 + 1/3 stop)

So, don't stop down below f/18.3 if you want to prevent diffraction's Airy disks from inhibiting your goal of 4 lp/mm in the final print to be viewed at a distance of 10 inches. If the print will be viewed at 20 inches instead of 10 inches, f/36.6 (twice 18.3) would deliver the same apparent sharpness (in terms of the effects of diffraction).

The question goes back to: At what viewing distance do you want your prints to survive scrutiny? Your desired resolution should fall somewhere between 4 and 8 lp/mm for a viewing distance of 10 inches. It can fall between 2 and 4 lp/mm for viewing distances no closer than 20 inches - but how are you going to stop people from standing closer than 20 inches?

For more information see: http://home.globalcros​sing.net/~zilch0/essay​s.htm

Mike Davis
http://www.accessz.com

More for Mark...

The "nice" thing about diffraction is that unless you do a side-by-side comparison, most people can not tell that diffraction is softening a print. I attribute this to two things:

1) Joe Public spends so much of his life being satisfied with the resolution offered by most television sets and 2) diffraction affects the entire image uniformly. It's easy to detect when a foreground or background element isn't as sharp as that portion of the subject space nearest the plane of sharpest focus, but it's very difficult to detect when the whole image is softer than it might be otherwise. You can choose to exploit this by lowering your desired resolution to something like 3, 2 or even 1 lp/mm, or remain dedicated to producing a high quality product that, like fine wine, might be fully appreciated by only a handful of discerning consumers.

To put this all into perspective, consider that the typical depth of field scale engraved onto the barrels of lenses for 35mm format film cameras, is set to deliver a Circle of Confusion diameter no greater than 0.03mm on-film. After 8x enlargement, this translates to a 0.24mm diameter CoC in an 8x10 print. Taking the reciprocal of 0.24mm, we get 4.167 circles/mm. This equates to a resolution of 2.08 lp/mm. So... If you're happy with the Depth of Field you see in 8x10 prints using lens barrel DoF scales set to deliver 0.03mm circles of confusion with 35mm format cameras, then you'll probably be happy with a desired print resolution of only 2 lp/mm when calculating the aperture at which diffraction's Ariy disks would impede your desired resolution.

I personally believe one should never go below 4 lp/mm in a print to be viewed at 10 inches. This can be adjusted to 2 lp/mm for prints veiwed at 20 inches, etc. I've also found that I have no ability personally to appreciate more than 7 lp/mm at a viewing distance of 10 inches.

From all of this, you should see that in addition to the formula I gave in my last post, for calculating the aperture at which diffraction will become visible, you should also work backwards from your desired resolution and anticipated enlargement factor to determine the maximum permissible diameter for Circles of Confusion. This diameter should be used in the DoF calculations you use to determine the aperture necessary to deliver sufficient DoF for a given subject space (Near and Far distances). Here's the formula you need for that (to calculate the CoC to use in DoF calculations):

Max. Permissible Diameter for Circles of Confusion (mm) =
1 / desired print resolution / enlargement factor

So, going back to the example I used in my last post, for a Canon EOs 1D Mark II, if we were to choose a desired print resolution of 4 lp/mm (for viewing at a distance of 10 inches), the CoC diameter we should use when calculating Depth of Field would be:

1 / 4 / 10.1 = 0.0248 mm

Additional information about selecting a CoC diameter can be found at:

http://home.globalcros​sing.net …Diameter%20for%​20CoCs.pdf

I vigorously recommend you download and use the freeware package called DoFMaster as a solution for calculating Depth of Field. This software allows you to print and create customized spinning-disk calculators for each combination of format size and focal length, but I encourage you to take it a bit farther and come up with DoF calculators for each combination of format (sensor) size, focal length AND print size. When making small prints, you don't need as much DoF as you do when making large prints.

http://dfleming.ameran​et.com/custom.html

Don't bother mounting them on your lens caps the way the author recommends - I prefer to make larger ones that I keep in my camera bag.

Once you have DoF scales that are customized for each combination of sensor size, focal length and print size, the only "homework" left to do is calculating the aperture at which diffraction will kick in for each combination. I write this diffraction-limit aperture on each of my DoFMaster spinning disk calculators.

In the field, things go very quickly. Just whip out the correct calculator for the focal length you want to shoot with AND the print size you intend to produce. If the DoF calculator tells you to shoot at f/16 for the Near and Far Sharp distances of your subject space, but your diffraction limit is f/14, for example, there's only a few things you can do to avoid compromising image quality: 1) Back away from the nearest subject until f/14 provides enough DoF. 2) Select a shorter focal length and maintain your current camera position, or 3) Make a smaller print! Options 2 and 3 will require pulling out a different DoF calculator disk from your camera bag.

If you discipline yourself to follow this regimen, you will never again stop down farther than you have to for sufficient Depth of Field and you will never again suffer image degradation due to visible diffraction. The benefits will include not only sharper images, but faster shutter speeds (thanks to never stopping down more than you have to.)

Now if you really want to go for it... Get yourself a laser rangefinder for measuring the distances to your Near and Far sharps in the subject space, and for finding a subject on which to focus your camera that is precisely at the hyperfocal distance given by the DoF calculator. Not just any laser rangefinder will do because most of them don't work at distances less than about 30 feet, but this one is "affordable" and works down to a distance of 12 feet (below which the distance scales on your lens barrels have sufficient granularity to do the job):

http://store.yahoo.com​/cspoutdoors/op40lasra​n.html

Mike Davis
http://www.accessz.com

I still haven't answered Mark's question - I'll get it done in this third post...

I should point out that there is a a third factor that should be considered before you can the formulas that allow you to limit Airy disk and Circle of Confusion diameter: The combined resolving power of lens and sensor, together as a system (or lens and film).

To simplify this discussion, let's make the assumption that most digital cameras are equipped with lenses that can resolve sufficient detail to satisfy the number of pixels with which the sensor is equipped and the density of those pixels (the higher the pixel density, the greater must be the lens resolution.) That means we can calculate the maximum print size achievable by simply dividing the number of pixels along one side of the sensor by the number of number of pixels per inch we intend to print.

Use these formulas:

Required image desnsity of the file to be printed =
50 * desired print resolution in lp/mm

Maximum Print Width in inches =
Width of Sensor in Pixels / image density

Maximum Print Height in inches =
Height of Sensor in Pixels / image density

For example: Let's say you've chosen a desired print resolution of 6 lp/mm to satisfy the majority of adults with healthy vision at a viewing distance of 10 inches. What image density must your files be saved at before sending them to the printer?

Required image density = 50 * 6 = 300 dpi

How large print can you make with a data density of 300 dpi from the Canon EOS 1D Mark II's 3504 x 2336 effective pixels?

Maximum Priint Width = 3504 / 300 = 11.68 inches
Maximum Print Height = 2336 / 300 = 7.79 inches

So, armed with this information, you can work out the maximum enlargement factor you can use when calculating the aperture at which diffraction becomes visible or when calculating the Circle of Confusion diameter you should use for DoF calculations (see my last two posts).

For film cameras, I recommend you assume you can achieve a total system resolution (lens + film) no greater than 45 lp/mm on-film with the best color films and lenses and no more than 60 lp/mm on-film with the best black and white films and lenses. Thus, if your desired print resolution is 5 lp/mm, for example, you should limit your color print dimensions to an enlargement factor of 9x (45 / 5 = 9) and your black and white print dimensions to an enlargment factor no greater than 12x (60 / 5 = 12).

To answer your question Mark, just do the math using the formulas I've provided, going for a resolution at the low end of what can be resolved by the human eye (4 lp/mm ?) so that you can be sure you'll see the "visible" diffraction in your comparison prints. (We don't know how good your eyes are.) Then shoot a planar subject - a subject that has very shallow depth, so that depth of field will not be affecting your results. Shoot a series of shots, from a tripod, focused as carefully as possible on the planar subject, using f-stops on either side of the predicted diffraction-limit (adjusting shutter speed for correct exposure, of course). Then print them, without sharpening, at the enlargement factor you specified in your calculations. Compare them side-by-side in good light at the viewing distance you specified in your calculations. The prints taken at apertures smaller than the calculated diffraction-limit aperture, will look visibly softer across the entire print than those taken at larger apertures.

Enjoy!

Mike Davis
http://www.accessz.com

Yeah, I guess... I bet that Belmondo will then prefer to shoot rails instead of trains. They sit quiet, not stupidly moving, and - even better - when you shoot straight away there will always be a section to be seen sharp.
Of ourse, no pun intended, it's just to smile a bit about this very (no kidding) interesting thread.

The only solution I see is that we're all going to have to shoot at various f/stops, and then enlarge our images to 12X18 inch prints and look for diffraction.

It is interesting, but I'd like to see the results for myself. Too many double-page spreads come from these cameras that we aren't supposed to stop down.

Just noticed this has had over 27.500 viewings,is this a record ?

In the real world, we use f/16 & 22 (or smaller!) to increase depth of field, not to increase resolution (this has been known for decades in film photography). If your subject is not conveniently all in one plane, and part of your subject that you need to be sharp is so because it is within the greater depth of field due to the small f-stop, then you have achieved your goal. If it is outside the depth of field, and so "out of focus", then the size of the Airy disk and theoretical resolution is totally irrelevant for that object, as it is not "in focus". Airy disks are applicable only to in-focus point sources of light (such as stars).

Having a pixel density greater than the theoretical resolution set by the Rayleigh formulae is not necessarily a bad thing, either, as diffraction disks can then be recorded as round(ish) extended objects rather than rectangular pixels. This near-myth of pixel-to-resolution matching was common in CCD astrophotography over 5 years ago, while only small chips were available, now everyone has megapixel-plus chips and more pixels per point of light, the issue is less of a concern somehow and the images are ... better.

Of course, a bigger sensor is still better in this regard.

I notice that this topic has had over 28,000 views,just interested what is the record ?

Why did you dig up a 5 month old response of a 3 year old post which the topic is not relatative today.

Why did you reply to a month-old post, then?

OBTW, the D60 may be out of production, but there are still 300Ds, with the same sensor parameters, on the shelves, so it's as relevant now as it ever was.

Even worse, there are 20D's on production.

I had to really laugh at this post (I came across it because I was doing a forum search for the D60). If the D60 fails, then the 20D must fail miserably (since it has a greater number of pixels for the same sensor size). I suspect that most 35mm film cameras will also fail, although on average, most modern films have around 60 million crystals on each frame. If we consider crystals = pixels...then the current range of cameras have a long way to go to match, but also, 35mm film cameras are worse off with the Airy Disk phenomenon. Sure, the 35mm neg is larger than a D60 cmos chip, but you're putting 60 million crystals onto it, rather than 6 million pixels. Given the size difference, you'd be looking at approximately 10 million pixels for a full frame cmos/ccd sensor (as an equivalent to the D60 and the smaller sensor size). Compare 10 million vs 60 million...

In all, I think most people would never even notice Airy disks, let along care about them

Dave

PS I'm also an amateur astronomer, so I have a reasonable good idea of what they are

Pixels do not equal crystals, except when photographing a target that has only pure black or pure white. In colour photography, each pixel presents 256 increments of tone in 8 bit pictures. Each crystal is either on or off so it only represents 2 possible increments. As a result, each pixel measures the equivalent light density of 128 crystals. An 8 megapixel camera has 8,000,000 pixels, each equivalent to 128 crystals for colour photography. It would take 1,024,000,000 crystals to measure the same increments of colour and shade as an 8 mp camera. That assumes crystals could be calibrated as accurately as pixels in a digital sensor.

The sensor of most digital cameras actually measures in 12 bit, that is extrapolated in camera to 8 bit for JPEG photos. If you use RAW you can extraplate to 16 bit TIF files. Using 12 bit instead of 8 bit increases the increments that can be measured by 16 fold.

Ah, the Franken-thread arises again.....