The following is excerpted from a discussion on sharpness in the flickr.com Canon EOS 40D discussion group. As such, certain passages pertain to the 40D, or Canon cameras in general. However, the concepts represented here apply equally to all camera models and brands. The entire thread may be found here.

Thanks to “BoldPuppy” for allowing the use of his thoughts in this article.

(By the way, RAW has two sharpening settings.. one for RAW, and the other for “RGB”… the RGB is a multiplier on the RAW sharpness. You will need to do some amount of sharpening, but the amount you do depends on many, many things - including personal preference, and how it will be used.)

When photographers come to a forum like this to ask for help on why their images don’t appear sharp, the usual advice hits on the basics. To start with, you need a high contrast area to focus on, as the camera needs to find *something* (an edge). The higher the contrast between one color and the next, the better the AF will be.

Non-sharp images are caused by:

·         missed focus

·         motion blur (subject)

·         camera shake / mirror slap (you)

·         too shallow of a depth of field

·         using a lens outside of its ’sweet spot’

Note that all of these are things that are used for creative purposes to get different effects the photographer wants. Sometimes, using a lens way out of its sweet spot is what you WANT to do. Other times, it’s much better to stick with the narrow range of ’sweet’.

Post processing. Hmm… Post has little to do with ’sharpness’. You can do some ’sharpening’ on an image using unsharpmask, but… there’s only so much you can do. If an image isn’t sharp to begin with, you can’t fix that. Post processing comes more into play when you start to pick up RAW, where you HAVE to sharpen the image some, or it’s soft.

To expand a little on what I said above:

·         missed focus

As Andreas noted, you need to be able to give the camera a fighting chance. Something with good contrast to help you attain focus. But, the focus MODE that you’re in can affect things, as well. One Shot is best used for static objects (landscapes, still lifes, animals/people sleeping). It locks focus, and keeps it there. AI Focus is where most people should keep it. This mode will adapt the focus somewhat if your subject moves between when you lock focus and when you take the photo. AI Servo is best used for moving subjects. It will continuously change the focal point until you take the photo. Note that in AI Servo mode, if you loose focus and press the shutter, it will take the photo anyway. In AI Focus, it won’t take the photo until focus is regained.

Also, learn how to move the focus point around. It really is sub-optimal these days to lock focus with one composition, and re-compose and then take the photo. This is especially true with very narrow DoF photos, where you can easily throw off the focal point with slight movements of the camera. Select the AF point that is closest to the eyes or main subject, and go from there. BTW, in AI Servo, the point will briefly light up, and then it won’t light again. Don’t worry, it’s tracking while you hold the button part way down, but it just doesn’t stay lit up.

·         motion blur (subject)

This is all about shutter speed. You really don’t want every photo at 1/1000. Honestly. Some motion blur makes some photos look real, and give it dimension. Too fast of a shutter, and you can have images look ‘fake’/static. This is all about light. To get faster shutter speeds, you need more light. (You can also do this by increasing ISO, but that only works to a point. On the XT, I was comfortable shooting up to ISO 800, but 1600 looked too grainy for my tastes.) Pay attention to the lighting that you’ve got to work with, or bring in additional lighting. If you want to learn about strobes, please visit http://www.strobist.com for how to light up your subjects with remote lights. That’s not beginner stuff, so I’d wait for a while until you’re comfortable with what you’re doing now, and you feel that you want to explore creative lighting.

You can get more shutter speed by using a bigger aperture (smaller number). However, ‘kit’ lenses tend to be very slow, which means that you are limited in what shutter speed you can get to. This is why the 50/1.8 gets exciting. It has a HUGE aperture, and as a result, you can get very fast shutter speeds in good light, or even decent shutter in poor light. This is something you’ll have to play with.

·         camera shake / mirror slap (you)

Too slow of a shutter, and you can start to see camera shake. The old rule of thumb here was get a speed faster than (1/length of lens in mm) and you’d be fine. Now, with the digital crop sensors, you need to modify this to (1/length * 1.6), and Andreas will say that a better multiplier is 2-3x. What this means is that if you have the 50mm, then you need a speed of 1/80 or faster. Andreas would tell you that 1/150 is really required. The reality is that this is a rule of thumb, and that it’s up to how you hold the camera, how much caffeine you’ve had, and other factors that will influence that. Just keep the shutter speed up a bit, and you’ll be fine.

Mirror slap is something that happens with longer exposures (around 1/10 or so). It’s something that you see on tripod exposures. If you need to use speeds around there, use the function called “mirror up” to prevent this. You will need to press the shutter twice - once to raise the mirror, and again to take the photo. Remember to turn this off when you’re done shooting like that.

·         too shallow of a depth of field

With the kit lens, it’s unlikely this is an issue. However, with the 50/1.8, it can be. Very, very narrow depth of field can make one eye in focus, the other out. Or even just a *part* of one eye. You might miss focus (the first issue), and get the ears in focus… or a nose…or the background. All of which has happened to me. Depth of field is related to: length of lens, size of aperture used, distance to the subject, and subject distance to the background. There are many resources on this subject. I like http://www.dofmaster.com This is something that will take some ‘playing with’ for you to get used to how this works. It’s fun, and fun to play with…

By the way, the weird numbering for apertures doesn’t make much sense, until you realize that they’re square roots of doubling the radius….

The first stop is f/1 (square root of 1). The next stop is f/1.4 (sq rt of 2). Then it’s f/2 (rt of 4). Then 2.8 (rt of 8), then 4 (rt of 16), then 5.6 (rt of 32), then 8 (rt of 64)… and so on. Each full stop is a doubling or halving the amount of light that comes in. In between numbers are the square roots of 1/2 or 1/3 the midway points. to figure out just where, you square the value. Take 1.8. Square it. You get 3.24. (Of course, these numbers are all rounded…). It’s closest to 3.33. When you look at the full stops squared, you have 2 and 4. You’ll see that 3.33 is 2/3 of a stop from 2 to 4, so you can say that a lens at f/1.8 is 2/3rds slower than a f/1.4 lens, or 1/3rd faster than a f/2 lens.

Don’t worry if that doesn’t make sense…

·         using a lens outside of its ’sweet spot’

I’ll get a lot of flack on this one. The rule of thumb here is that lenses all have configurations where they give the best images they can. The usual thought is that a stop or two from the maximum aperture and in the middle of the zoom range (if it’s a zoom) is where a lens is sharpest. Some lenses never truly get sharp.

I use http://www.photozone.de to help me determine how lenses perform, relative to one another. On the charts (page 2 of each lens review), I consider ’sharp’ to be anything over 1900, and very sharp to be anything over 2000. They give numbers for center and edge. Some lenses are able to maintain sharpness from edge to edge, and those lenses cost a lot of money. Most consumer grade lenses get soft at the edges, and for some, it’s truly awful. But, if you never shoot photos that you would care about the edges, you may not care.

You will notice that the bars for sharpness get taller as you get to smaller apertures, and then there’s a drop off as you pass a midpoint. Using the lens with too small of an aperture (usually to get very deep depth of field) will result in softness coming in from ‘diffraction’ around the blades of the aperture. The rule of thumb I use is .. use f/8, for as much as possible, and the majority of your photos will be in focus.

If you need to get longer exposures, consider buying *good* ND (neutral density) filters.

I didn’t mention this in the above post… but filters are also a source of issues. They can soften images, cause the AF to miss, introduce glare and ghosting, and a host of other issues. Unless you shoot in a hostile environment (and if that’s truly the case, you should invest in a case, or a really good protective device for the entire camera/lens, not just the front element), the majority of photographer don’t need most filters any more, especially when lens hoods are used.

As part of the same discussion, Robert Seber linked this posting from his flickr photostream:

Optimal Aperture for Foreground Sharpness with Infinity Focus (feet)

Optimal Aperture For Foreground Sharpness With Infinity Focus (feet)

 What this graph shows

To get greater depth of field you need to use smaller apertures. But lenses also become less sharp at small apertures due to diffraction. There is therefore an optimal aperture to use to get the foreground as sharp as possible, which is what this graph shows.

·         Infinity focus

It is usually best to focus on the object that you want to be the main subject of your shot. When that’s a mountain, hill etc it is often a long way off. An approximation to this is “infinity”, marked on most lenses with the infinity symbol which looks like a sideways 8. If you have an autofocus lens it is probably best to focus directly on the far point than to set it manually, but in low light autofocus becomes unreliable so it is probably better to use infinity focus.

·         Choosing an aperture

When the whole of the scene is a long way off choosing an aperture is easy. Most lenses are sharpest at about f/8, so that’s the aperture to use. When you want something closer to be sharp too it gets more tricky.

·         How to read the graph

Suppose you are using a 20mm lens focused at infinity. Something in the foreground is 12ft away. The point at which 12ft and 20mm intersect is closest to the f/13 line, so f/13 is the aperture to use.

Suppose you are using a 50mm lens focused at infinity. Something in the foreground is 6ft away. No line on the graph is close to this. At this point you need to re-think your shot and move further from the foreground, otherwise it will be intolerably soft. 26ft to the foreground will allow you to use an aperture of f/22.

Suppose you are using a 10mm lens focused at infinity. Something in the foreground is 12ft away. You’re above the f/8 line so f/8 will work just fine.

·         Calculations

The lines are drawn though points where the Circle of Confusion is the same size as the Airy Disc. The relationship holds whatever sensor size your camera has - no conversion factor is necessary (you should use actual focal lengths to read this graph, not 35mm equivalents).

·         Useful conclusions

The longer the focal length you use the smaller the aperture you will need to use to get the foreground sharp, or the further you need to move back. If you want to get really close to the foreground, and get sharp shots, a wide angle lens is a must.

Read original blog post