Thursday, October 31, 2013

Tools of the Trade - Some Interesting Properties of Digital Sensors

In the previous blog post, I wrote about optics and various properties that are commonly discussed.  This post is devoted to a discussion of the other end of a camera kit, the image capturing device called a digital sensor.

Much discussion in the camera world is devoted to cameras, their sensors, and who is trouncing whom in the Great Megapixel Race.  Without using an inaccessible scientific or engineering language, I will try to shed useful light (oh, yes, keep those puns coming) on the subject.

To start, there are many sensors made by many many manufacturers in a great many sizes.  Frankly, I was shocked to see the list of all the companies that make sensors for various photographic applications.  Most of us only know the more popular camera brands, such as Canon, Sony, Nikon, Olympus, Panasonic, Sigma, Fuji, Phase One/Mamiya, Hasselblad, Leica, Samsung, and a very long list of cellphone manufacturers.  Their sensors may be manufactured by the parent camera company, but sometimes they are not.

The basic function of a photographic digital sensor is rather simple and obvious.  That is, a sensor receives light rays directed to it by a lens.  Upon receipt of these light rays, a series of very tiny sensors record the intensity and the color of the light.  Each very tiny sensor's record is an electronically generated series of numbers.  Records from millions of tiny sensors are gathered and presented in a way that we, as viewers, can interpret as an image.

Sensor Size Descriptions -
There are primarily two aspects of "size" that are used in describing photographic sensors.

The first is the physical size, or dimensions, of a sensor.  Useful physical photographic sensor sizes span everything from amazingly tiny cell phones through to medium format.  That is, some sensors are truly small and others are seemingly quite large.  When discussed, you will hear cell-phone, or APS-C, or Full Frame used as a description of the physical dimensions of the sensor. 

The second size important in understanding photographic sensors are the number of light recording sights a sensor implements.  In common language, this is the number that we refer to as megapixels, or millions of light recording sites.  You will see everything from 3.1 megapixel (from some of the earliest commercially available sensors) through to 120 megapixel sensors (as of this date) that sit on a lab bench somewhere in a camera manufacturer's Research and Development facility.

Light Sensitivity -
In the real world, lighting conditions are highly variable.  When we experience full mid-day sun, the amount of light reaching us from the sun is quite high.  When we experience light from a single candle set in the middle of a large room, the light reaching us is comparatively low.  For a camera sensor to be useful in as many lighting situations as possible, we need a camera/lens/sensor system that is flexible enough to enable image capture across a broad range of light conditions.

Lenses provide an aperture that is used to provide one of three ways to control the amount of light hitting a sensor.  For instance, the smaller the diameter of the aperture opening, the less light that will hit the sensor.  Aperture control is as old as photographic lenses (from the early 1800's).

A second way to control light reaching a sensor is with a shutter.  This is particularly useful when trying to "stop action" when shooting sports or when capturing the Milky Way on a particularly clear and beautiful night.  Shutters have been used since shortly after sensitive film emulsions required accurate control of exposure (mid to late 1800's).

The third way that is used to help balance shutter speed and aperture against the amount of light hitting a sensor is by varying the sensitivity to light of the sensor itself.  This is accomplished in a camera by controlling electronic signals to the sensor.

Borrowing terminology from the original chemical, atomic silver halide film technologies that describe how reactive a sensor is to light, we have the acronym ISO. The lower the ISO, the less sensitive a sensor is to light.  Conversely, the higher the ISO, the more sensitive a sensor becomes.

Interesting Properties to be Aware of - part one

There is an interesting relationship between image resolution and senor megapixel count.  It is precisely as follows.

As previously described, an image consists of a collection of pixels that describe light intensity and color.  It is safe to assume that, in terms of image resolution, that a sensor can accurately capture a sharp edge and reproduce it by moving from a white pixel to a black pixel.  Using this, we can look at the number of image pixels and use ideas from the USAF Resolution Test Chart and determine the maximum resolution a sensor can return.  The math is quite simple.

Resolution in Line Pairs = [(Number Information Sites) divided by (Physical Dimension of Sensor)] divided by (2 Line Pair per Millimeter)

For example, looking at an 8 megapixel Canon DSLR sensor, the 30D, we see the maximum output file dimensions are 3504 by 2336 image information sites.  The physical dimensions of the APS-C sized sensor are 22.5 by 15 millimeters.  The answer is calculated as follows.

78 Line Pairs per  millimeter = [(3504) / (22.5)] / (2)

Continuing a little...

78 Line Pairs per  millimeter = [(5616) / (36)] / (2) - Canon 5D MkII 22 megapixel full frame sensor
116 Line Pairs per  millimeter = [(5186) / (22.3)] / (2) - Canon 7D 18 megapixel APS-C sensor

102 Line Pairs per  millimeter = [(7360) / (35.9)] / (2) - Nikon D800 36 megapixel full frame sensor

97 Line Pairs per millimeter - [(10380) / ( 53.7)] / (2) - Phase One IQ180 80 megapixel medium format sensor

This information will be useful when we evaluate lens performance against sensor capabilities, and when we look at people's ideas of image quality and the need to buy "better" lenses.

Interesting Properties to be Aware of - part two

Looking at the number of line pairs per millimeter that the best human eyes can resolve (5 line pair per mm), we can calculate the maximum print size we can make while retaining all of the sensor resolution.  The math is, again, quite simple.

Maximum Print Size = [(File Image Information Dimensions) divided by (2 Line Pair per Millimeter)] divided by (Maximum Human Eye Resolution in Line Pair per Millimeter)

Converting for English and Metric centimeters, we see, again using our example cameras, roughly the following.

13 x 9 inches or 35 x 23 centimeters - Canon 30D

22 x 14 inches or 56 x 37 centimeters - Canon 5D MkII

20 x 13 inches or 52 x 34 centimeters - Canon 7D

28 x 19 inches or 73 x 49 centimeters - Nikon D800

41 x 31 inches or 104 x 78 centimeters - Phase One IQ180

This information will be helpful when we evaluate the relative maximum print sizes that each imaging system is capable of, and compare it against the needs of the publishing industry and making fine art images that hang in galleries.

Interesting Properties to be Aware of - part three

The last thing I would like to note here are the effects of changing the light sensitivity of a sensor.  Measured in terms of dynamic range, or the range of light from dark to light that a sensor can capture, something interesting happens.

Take a look at any sensor's ISO chart (here is the Canon 5D MkIII example) that tests for dynamic range and what do you see?  At low ISO, a sensor is capable of capturing a broader range of light than the same sensor set to a high ISO.    The dynamic range delivers 12 EV (or f-stops of light) at ISO 100.  The range of light captured drops to 8 EV at ISO 12800.  The sensor is loosing sensitivity to a broad range of light as the ISO rises.

Now compare a full frame sensor against the smaller physical dimension APS-C sized sensor (in this case, a Canon 70D).  What do we see here?  The 70D's sensor captures a similar 12 EV of light at ISO 100.  But in this case, the sensor's ability to capture light at high ISO drops quicker than the full frame sensor.  The 70D captures less than 7 EV of light at ISO 12800 or 1+ EV less than the full frame sensored Canon 5D MkIII.

This information will be helpful when we evaluate actual sensor development advancements and high ISO performance against marketing hype.

I realize this may be a lot of information to absorb.  Each piece is vital to understanding the current generation of cameras, lenses, and their real world capabilities.  I will try to tie all these sensor numbers together with optical performance, marketing hype, un-enlightened commentary, and Reality in the next blog entry.

Well, perhaps it will take several blog entries...

Tuesday, October 29, 2013

Tools of the Trade ~ Optics, Lenses and their physical properties

If you spend any time over on the message boards and forums around the 'net reading about lenses and cameras while trying to keep up with the volume of "information" regarding the tools of the craft, you might be a little confused.

It appears that people feel they need a "better" lens to make a "better" photo.  There are folks who are convinced that Zeiss or Leica lenses are demonstrably better than, say, something Canon or Nikon or Sony might offer.  Some folks have strong feelings about the selections they have made and will defend them to the ends of the earth.  Many websites offer frankly misleading information about lenses and how to select something that will work well for you.

What I would like to do here is start with a basis for understanding optical properties.  I will begin with a set of definitions and their practical effect.  I would like to do this in a non-scientific language manner so as to keep this accessible to anyone who has an interest in furthering their understanding of what is really going on when we talk about lenses, optics, and imaging systems.  Future blog entries will cover the reality of modern optics and compare them against marketing perception and some people's beliefs.

To begin with, the function of a lens is to take light rays bouncing off or emanating from a subject, pass them through glass elements of various shapes, and to send those rays of light on to a blank surface or light sensitive material.  Traditional materials have include canvas (for artists who worked inside a darkened room - see Vermeer's work as a good example), photographic film (including wet plate collodion and dry plate film - see Kodak's revolutionary work in this area), and the current widely available digital light sensors.

The challenge is how well light rays that pass through glass are "focused" onto the intended medium.  It is this simple, fundamental act of making sure that an image is free from as many "un-desirable" optic artifacts that the entire conversation of "lens quality", product prices, and who makes the "best" optics arises.

Resolution -
When people think of image sharpness, they are thinking of optical resolution.  When a scene has a transition from a light area to a dark one, resolution is how quickly and accurately that transition can be captured by our imaging system.

While I promised not to throw too much science into the discussion, it is important to realize that there is a natural, physics based limit to how sharp a lens can be.  It is called optical diffraction.  Lens designers know these limits and, in some cases, try to build optics that come as close as possible to these limits (given time and cost of materials and manufacturing).

We, the common human, can measure resolution if we so chose.  The classic method is to use a United States Air Force (USAF) Resolution Test chart.  For years I used this one from Edmund Optics in the US.

You may notice from the theoretic limits that it is expected that a lens will more accurately preserve light to dark transitions for lines that radiate away from the center of the field of view than such transitions made perpendicular to those rays of lines.  This is an interesting property of optics and one that is good to keep in mind as we look at the following optical effects.

This approach to measuring optical qualities has fallen from favor as a stand-alone test method.  However, it is used as the basis for the most commonly used current method of describing optical qualities, and this is as follows.

Modulation Transfer Function (MTF) curves -
When you look for lenses from current manufacturers you many times see MTF curves offered as a proof of demonstration of quality.  This test method is useful because the human eye sees "sharpness" in terms of contrast, not resolution.

The MTF test method expresses how much contrast is preserved by an optic as a scene transitions from light to dark areas at various levels of resolution.  These levels of resolution are taken from the resolution test method for radial and tangential lines (see prior section).  The thickness of these lines as well as the focal length of the lens under test and the distance between the lens and test chart are what predetermine those levels of resolution.

Typically you will see two different levels of resolution used in published test results.  If comparing lenses from different manufacturers, it is important to know what resolutions were used.  Different results will be reported for different levels of resolution.  Further, it might make a difference to you to learn which manufacturers publish expected design results (nearly everyone) and which offer actual test samples (Zeiss in some cases and Sigma).

If you would like to better understand MTF, how it works, and how to read MTF curves, Cambridge in Color's tutorial might be a good place to start. They provide a nice overview of the relationship between resolution and contrast, as well as providing a good understanding of various test methods and physics involved.

Chromatic Aberrations -
An optical effect that you might read about in lens tests is Chromatic Aberration (CA).  This is where a lens fails to successfully align colors across the visible spectrum.  This effect is typically more difficult to control near the edges of a scene, which is why that is where testers look for the effect.

Additionally, you will read test reports that measure the amount of CA various lenses have at different apertures.  In practice, you will see CA as color "fringing" of portions of a scene rendered near the edges of a field of view.  The amount of CA can vary with the size of a lens aperture.  Optical designers work to control, if not outright eliminate CA.

Field Flatness -
An important design element in creating a lens is to come close as possible to ensuring that elements in a scene arranged on a flat plane are accurately reproduced.  In other words, objects arranged along a line in a scene are in focus across the scene after passing through a lens.

In many cases, lenses are sharp along some kind of curve.  The practical effect is that if you were to take a photograph of a painting, for instance, the edges of the painting may not be sharp in your photograph, but subjects slightly in front of or slightly behind the painting would be sharp.

Lens Distortions -
Another design element in lens creation is controlling spatial distortions.  Said another way, lens designers work to ensure that straight lines along the edge of a scene are accurately reproduced.  This is why you will read in lens test reports the amount of distortion they were able to measure.

If lines near the edge of a field of view bow out and away from the center of an image, it is called barrel distortion.  If these lines are reproduced in a curve shape leaning toward the center of the field of view, it is called pincushion distortion.  If there is no distortion, the lens must a gift from the gods.

This pretty much sets the stage for future blog entries where I will rant and rave about how people perceive their lenses, the prices they are willing to pay for them, and try to compare these subjective "feelings" about lenses and lens "quality" against physical reality.

If you find this kind of information fascinating and if you would like to delve further into common photographic optical properties, take a look at Zeiss' primer on the subject.

Wednesday, October 16, 2013

Peaks and Valleys (2)

No.  The Muse has not yet returned.  She must be away on extended holiday somewhere.  Not that having spent the last five weeks with family guiding them through their Europe Vacations (including time in Spain) had any impact on... wait... that must be it...

Shapes and Light

I have not had any time to work my art.  Of course the Muse couldn't find me.  I've been up to my ears in distractions!

One thing I've had time to think about is that I can proceed in at least one of two ways.  I can strive to create images that make other people happy by studying and then making images that are culturally "current".  Or I can simply create the art I want, for better of worse.  It was then that I had another, stronger realization; I need to know what I want to create if I take the second approach.  That's the tough part, now isn't it?

Visiting museums in Spain, I had the opportunity to see just how great artists of the 19th and 20th centuries could be.  The real surprise was Picasso.  Up until this visit, I'd viewed him from the position of my own ignorance.  Then, after visiting the Prado in Madrid, I became convinced that Europe's greatest artists deserved their places in history.  Fabulous works all around.

If I studied beautiful works of art, would the Muse would return?

Shapes and Light

Once back in Paris to home and hearth, I couldn't help but notice that the World of Photographic Tools continues to grind out new and interesting toys to ogle and drool over.

Joining Canon's WIFI-only PowerShot N comes Sony's hybrid offerings in two lenses with sensors in their QX series.  These can be strapped on to and controlled by a cell phone or tablet.

If you remember, I wrote a fair bit about how nearly instantaneous art creation could become when combining a WIFI enabled lens/sensor system to social and image sharing networks.  Canon's and Sony's product offerings have yet to take the fully integrated step of combining a lens/senor system with and Android or iPhone operating system that Samsung has.  Still, progress is being made, even if it is in Baby Steps.

Then, yesterday, like a meteor hitting Terra Firma, came Sony's full frame E-mount (NEX-like) product announcements.

I've been thinking about down sizing my image capture systems.  The older I get, the harder it is to hold and manage a full frame Canon DSLR.  Would Sony's new products attract me enough to encourage me to sell my old gear and move into a new system?  The costs would be high and living on a fixed income would force me to seriously study any potential wholesale move such as this.

Shapes and Light

Quick as a bunny, I took a look at the specifications of the new Sony 7R and Vario-Tessar products.

It seems like Sony has done a nice job in creating a new family of products that are WIFI connected while offering the kind of image quality that large sensors can help an artist achieve.  The weight of the 7R body is 407grams without battery.  The weight of the 24-70 f4 Zeiss is reportedly 430grams.  While the lens is a little short on the long end of things, I would minimally need combination to shoot in the studio.

The weights compare with the Canon 5D MkII's 810grams and the 24-105L f4's 670grams.  That's Sony's 837grams, not including battery versus Canon's all up kit weight of 1480grams.  This seems a useful improvement.  It would be really great if a full frame Sony could also replace my current "walk around" NEX5 kits too.  The all up weight of the NEX5 with battery and kit lens is 502grams.

The old NEX5 would be 60 percent of a 7R/Zeiss kit weight.  The new Sony full frame would be 56 percent of the weight of the 5D/240-105L setup.  Hmmm... this is squarely in the middle between my "walk around" and "studio" setups.

Shapes and Light

Obviously, weight is only one dimension to be considered when evaluating imaging systems.  The breadth and depth of optical solutions, 12 versus 14 versus 16 bit A to D's used in the sensors, as well as support by third party suppliers, and long term engineering investment are important too.  In this way, Sony's new products do not contribute anything new nor compelling.

I don't yet see a clear way through this.  At the bottom of it, Sony's newest full frame mirrorless offerings are not really any more capable than my current images makers.  In fact, if I consider long lenses for bird and automobile photography, as well as ultra short optics used in tight situations, my Canon DSLRs remain the Cock of the Roost.

I will continue to watch the industry to see if they can strike the kind of balance between size, weight, and capability that I've been waiting for.  This might change completely should Canon buy a medium format sensor company and start engineering very large sensor solutions.  In which case I might head in a completely different direction with my art, enabled by a radical chance in tools.

Shapes and Light

What I'm really waiting for is the Return of my Muse.  Then all these mental machinations over new toys will subside and I can once again get down to the business of image creation.