Wednesday, 22 May 2013

The hunt for ultimate sharpness, a never-ending story...

Today as I was reading a couple of Ken Rockwell’s documents about the sharpest lens and image sharpening, an old forgotten discussion I had some years ago came to my mind.  I remember that my opinion was not accepted as plausible from my fellow photographers (I was never sure why because it seemed very reasonable) and that’s why I remembered it now.  My sentence back then sounded something like: “My camera has a 10 megapixel sensor, so my photographs should be sharp enough at 6-7 megapixels...” and before long hell unleashed :-P. 

The discussion after this sentence was about image sharpness, lens and sensor detail and I had the ridiculously unheard idea of disconnecting the image detail in megapixels from the lens type (Canon L lenses) and sensor resolution for daily prints.  I also claimed that true 1:1 resolution of images straight out of the camera was next to impossible to be clear and crisp and that the only way to make a crisp image was to resize it to about 70-90 % of its native size and to apply a moderate sharpness like unsharp mask, smart sharpen or even a simple sharpness pass. 
Taken with a 2005 Canon EOS 300D with Sigma 18-55 lens, my first DSLR camera with only 6 megapixels of resolution!
While my opinion is not entirely true, when it comes to today’s equipment, the practical difference among hardware in terms of quality or sharpness, comes down to a very “small” variation.  In addition, I can almost swear again, even after all this time has passed and equipment has evolved, that there is a simple way (I will show you later the guidelines) that can turn any photo taken with a modern digital camera to an astounding sharp image picture, no matter what brand, resolution and lens you used.  Note here that by the term “image” I mean a post processed photo and not the original photo itself.  There is no way to turn a photo magically to its absolute best without manipulating it afterwards in Post Processing and you can only blame physics for that.

Taken with my Canon EOS 60D and Canon 15-85 lens. A 18 megapixels camera with a relatively expensive lens.
First of all, to start hunting for the sharpest lens you must understand how lens work and to start hunting for the sharpest camera sensor you must understand how camera sensors work, too.  Making a long story short, for the acquired image it is true that a fixed lens is almost every time sharper than a zoom, more expensive lens most of the times mean better quality than cheap ones and branded lens are possibly best suited for the brand they were built for.  In addition to that, a more expensive and better in terms of resolution camera sensor will produce more detail in conjunction with the lens used.  Keep in mind that there is always a physical limit, but you shouldn’t be so digit nerd about it. With that out of the way, we can start compromising for the best solution that is aimed for people not hunting for the dream of ultimate perfection, which by nature can never be achieved.

Are film cameras better? Well the film resolution can reach it's limits depending on the paper you are printing, but it was still analog thus not easily measurable!


There are various reasons why a photo can never be ultimately sharp but just relatively and even why one has to use digital and post processing manipulation to “increase” sharpness and clarity.  If one can understand the physics under the hood, it should be just a matter of minutes before one creates the sharpening procedure that fits one’s needs.  But first let’s explore the obstacles of ultimate image sharpness in camera without any post processing:

Air is full of dust particles, even air IS particles.  Many particles mean less light, scattered light, fog, mist and lack of clarity.  Especially when using a telephoto lens, heat shimmer is magnified and refracts what is viewed. You can witness heat shimmer with naked eyes in the summer.  It is usually visible very close to earth when lying on a beach at noon.  So atmosphere is an everlasting “fog” of light and it is our first obstacle (unless you will be shooting at Moon!)

Focal planes
Our 3D world is not flat. If it was, everything would be in focus having the same distance from the camera (and provided our camera lens were flat on both sides).  This means that only one focal plane is in focus and every other focal plane is more or less out of focus.  Depending on the lens and aperture settings, we can only reduce or increase the effect, but we can never have everything in perfect focus.  Wider lenses compress the angle of view reducing the distance between different focal planes (or in other words forgive slight out of focus errors) while narrower lenses (tele) inflate the angle of view, increase the difference between focal planes making focus errors more obvious.

Manual Minolta 55mm f2.8 lens. Is this thing sharper used on a modern DSLR? Well...no! It's the same.

Diffraction in lens
Glass elements of our lenses, depending on their thickness, are always refracting some light (scatter, focus, spread, you name it!). Increasing the distance between the many glass elements inside the lens for zooming or focusing, means more or less refraction is produced.  Clearer and thinner lenses, mean that greater amount of light (and better in quality) is leaving the lens towards the camera sensor. So, light scattering is another obstacle against achieving sharp images.

Diffraction in sensor
What you read above is also true for the sensor receiving the light that comes out of the lens, because the protection surface on the sensor is also (most of the times) an infra-red filter that also scatters light.  In addition, having denser pixels (more megapixels) in two identical sized sensors does not mean that the amount of scattered light can be changed, but that it IS recorded in more pixels. This happens because light is already scattered when leaving the lens and there is no way to make it clearer, so it can only be recorded as it was when it left the lens.  In this sense, camera sensor is secondary to optimal image resolution, but resolution CAN be improved by clearer lens that leave higher quality light pass through them (also less diffracted) and largest camera sensors to receive that high quality light.  Again, it can never be possible to transmit the light without any loss in quality from the lenses to the sensor.

Subject motion, camera motion and accuracy of focus is critical for sharp photos.

Human errors, moving subjects, moving camera
These are the most common errors. The world does not (and should not) stay still. Everything is moving.  You are moving with each breath you take and you can make mistakes in focusing.  Heck, the camera can make a focusing mistake!

The simplest in-camera weapon
Aperture is your only weapon for best achieved sharpness, but not for the entire frame.  Lenses are not flat in any of their surfaces and the entire focal plane (if there was just one) is recorded by definition unevenly in the camera sensor.  To make things more complicated, not every object in the frame has the same distance from the sensor, thus allowing for more focal planes to exist within the same picture.

Compression of focal plane depth. Canon EOS 450D (12 megapixels) with Sigma 70-300 APO @70mm and f16 with 1/500 speed using ISO 800.

You can only “force” a certain amount of focal planes fitting perfectly focused in the frame, but almost always you cannot fit the entire range you can see (if this is what you want to aim for).  Major diffraction is mostly visible at smaller apertures to become the best combination somewhere in the middle and then it starts increasing again.  That’s why in f1.4 we have a very small focal plane depth and scattered blurred light.  This is corrected as we tend to reach f8 and it starts increasing again until we reach the minimum of the lens (f22 to f32).  Even though we should have almost everything in focus in f32, the diffraction amount is so big that it makes the photo seem blurred.

The best method to approximate this combination is by experimenting.  This way you will also improve your knowledge of your equipment without needing to read books about optics and physics.  There’s no sense in me saying that you should use f8 at 1/320 and ISO 100 at 5-10% of your lens shortest end for best results without any justification, is it? And I did not even explain to you about sensor sensitivity setting (ISO) did I? I won’t make things more complicated that they already are, so let’s SUPPOSE you have no choice but to use ISO 100 only.

Taking all of the above in account (and many more of life’s lessons): “There is no way to achieve perfection”, stop trying.  But you can cheat!  Oh yes…this is where this all comes in performance.  This is where you can balance and decide for yourself if you need a more expensive camera and lens or if you just need to re-think the way you approach sharpness in photography.  I will not go in lengths about the fact that when you post in internet posts, blogs etc., you HAVE to resize your photo in order to load faster.  If you don’t, keep in mind that almost no one has the time or desire to wait until a 25 megabyte file loads in their screens.  I already expect you to know this.  But even if you don’t know it, you can easily figure out that this simple way I mentioned can be easily adjusted for whatever size you need.
How much detail is enough? Well it depends on the use.

Compromising, how much is needed
The key word for this simple method is “compromise”.  Because in order to achieve sharper images with the equipment you have already bought, you have to leave something behind.  And this “something” is resolution.  How much, really depends on how you want to use the resulting image.  What is the role of image resolution?  More resolution means more visible imperfections, how about less resolution then?  How much of visible imperfection can go away?

Well, let’s say you have a 12 megapixel photo.  This is 12 megapixels at 1:1 so a dust spot (or sharpened artifact) is 100% its size, right? Let’s make our image 6 megapixels and see the dust spot becoming half the size it was. This is the original principle and this is why it works.

Try this:
Let's use a raw image you have processed it as you saw fit in Lightroom or Photoshop with every option you could think of. Heck let’s say you only have a relatively sharp image (meaning proper focused and using a good combination of aperture and shutter speed).  Then open it inside Photoshop and resize its dimensions by 10%-30% (this is really depending on the megapixels and clarity of the combination of camera and lens).  Then go to the menu and select Filters, sharpen (just plain old sharpen).  If you compare the two images you will see that clarity is greatly improved at full size.  But I guess someone has to tell you how much reduction is needed right?  Well, as I told you before, it depends on the use of the resulting image.

Expensive tele lens? Not yet, but I used a film left-over from my father.  Canon EOS 450D with Tamron 500mm f8 (fixed) mirror @1/100 and ISO 400.

When to care for expensive equipment
To make you understand what I mean, please ask yourself what you usually do with your photos.  If you are always posting images at, let’s say, 1920 pixels wide why rush to buy a 36 megapixel camera? Ok, I went way too far… Why rush to buy an 18 megapixel camera?  Still too far…  Why rush to buy a 12 megapixel camera? Ok, last time: Why rush to buy a 6 megapixel camera?  You are not getting it, right?  Most High Definition monitors can display images up to 1920x1080 pixels at 72dpi (sorry no 300dpi here!), that’s 2.1 Megapixels while a 6 megapixel image is roughly 3000x2000 pixels at 300dpi!!! Not convinced? At what size do YOU post your images on-line for daily use? RIGHT!
Oh, you said you wanted to print sometimes…I got it! Well with this 6 megapixel image you can print copies at 105cm by 70cm at 72dpi or at 50cm by 34cm at 150 dpi (which is the resolution that a normal photo is printed in photo labs) or 25cm by 17 cm at 300 dpi which is among the highest resolution in print you can find.  Do you need more? Well then you should already know why you must pay more for better equipment and you really didn't need to have read all this.  But if this is alright with you, then consider yourself robbed now you own an 18+ megapixel DSLR… 

If you want to sell your photos or print at larger sizes there’s always the way I told you in combination with a "nice" set of equipment that will improve sharpness and clarity of your images, if you want more than you already have.  Of course, there are more complex ways of fooling the eye than resizing and sharpening a photo, but try not to get lost in details and just improve your view.  And don’t worry about sharpening so much for daily photo posting, even an 8 year old, 6 megapixels camera with kit lens is enough for super sharp images.

And don't forget to have fun!!!

My latest images for sale at Shutterstock:

My most popular images for sale at Shutterstock: