ISO Control

Bradley Hill Spring, Chesterfield, NH

Getting It Right in the Digital Camera
My father handed down my first 35mm SLR at a time when
everyone was taking snap-shots with Instamatics. I know, I'm older than dirt, but I just got a $10 lifetime National Park Pass, so suck it!

Song in a Dark Czech Bar  : ISO 6400

Anyway, when I got the camera, I had to decide which speed of Kodachrome to use. I made the radical decision to forsake the time-honored ASA 25 and go with the new, lightening fast, ASA 64. My father worried that I was a communist, but that extra stop felt liberating. As film improved faster speeds became available, always with the trade-off of coarser grain. The increased sensitivity was great, but If you wanted to change film speed in the middle of a roll, you had to wind the old film back into the canister while trying to stop before the lead got sucked in all the way. I know what it feels like to open the camera and find that the lead is gone, and with it, the remaining exposures on the roll. The whole process was complicated, time consuming and wasteful. The result was that changing film to adjust speed for varying conditions was not a routine consideration. Exposure was almost always adjusted through the manipulation of just two variables; aperture and shutter speed and these two factors were unbreakably linked. For any specific aperture there is only one shutter speed that will result in the desired exposure, but the ease of adjustment of ISO in digital cameras has broken the chains of that strict inverse relationship.

With digital photography the adjustment of sensor sensitivity is achieved through the turn of a dial or tap of a button, making ISO the important third variable in controlling exposure. Through the adjustment of ISO, there is now a range of shutter speeds that can be used with a fixed aperture while keeping the exposure constant.

Nazca Boobie, Galapagos Ecuador;: ISO 1600

Conversely, when a fast shutter speed is needed to capture action, adequate depth of field can be preserved by stopping down on the aperture and compensating by increasing the ISO. The combination of easily adjustable ISO and the ridiculously high sensitivity of modern digital sensors has remarkably expanded our flexibility in controlling exposure. But this flexibiliy comes at a cost and that cost is noise.

Sensor Noise
As is true with film, higher ISO's lead to increased image noise. Newer sensors have greatly improved low light response making it possible to get quite serviceable images with surprisingly high ISO settings, but different cameras have varying abilities to handle noise. Even as it is essential to know how shutter and aperture affect exposure and image quality, it is also important to know how adjustments of ISO affect image noise and sharpness on your camera. The physics of shutter and aperture are essentially the same for all cameras, but digital sensors vary greatly and the only way to understand how your camera handles high ISO is to test it. I have to admit that, until recently, I have used the" shoot and pray" approach to high ISO, but after testing my Canon 5D Mark 2, I feel much more comfortable pushing the ISO when needed. The result has been better low light images and sharper action shots.

The Test

ISO Test Imag

1) First start with a scene that has a wide range of brightness. Dark areas are especially important since noise is most noticeable in these regions. If you wonder why this is true, check out my recent "Exposing to the Right" article.

2) Shoot a series of images using your full range of ISOs. Aperture preferred mode works best since it keeps the depth of field constant. The scene should be well illuminated to avoid long exposure noise from contaminating the images taken at low ISO. If possible, shoot in RAW for maximum flexibility.

3) In my test I first I brought each image through Adobe Camera Raw with identical settings for optimal exposure, but without any sharpening or noise reduction. Zooming in on a small segment of the image, it is easy to see how noise increased with higher ISO. To my eye, results were quite good up to about ISO 800, but then progressively deteriorated up to the max of 6400. 

Image Detail, No Noise reduction

4) I then went back to the images and used Camera Raw's excellent noise reduction function to clean up the images. The noise reduction function in the new Lightroom 4 has also been improved. There are many other good software solutions for noise reduction, but in all of them the best final result is a compromise between noise reduction and the inevitable softening of detail. In my test, noise reduction expanded the acceptable ISO range by a step or two. 

Noise Reduction in Camera

A test like this can be a helpful guide, but the impact of high ISO on any specific image will vary depending on its brightness, the importance of shaddow detail and the final destination of the image. A large fine art print will have a narrower range of acceptable ISOs than a picture destined to be a small, low resolution jpg on Flickr.

As is always true, photography is defined by the compromises you make. The control and expansion of ISO made possible by digital photography, provides a great new creative variable to the complicated equation that makes up each image. The art is in how you balance the competing factors to achieve your own solution to the equation. It is just one part of getting it right in the digital camera.

Read More

Exposing to the Right

Old Stone Bridge, Ashuelot River, Keene New Hampshire

Getting Exposure Right in the Digital Camera

Last week I began the discussion of exposure control in digital photography and more specifically about the use of the histogram to nail the optimal exposure every time. Much of that discussion focused on the importance of adjusting exposure to avoid the extremes. To avoid overexposure, positioning the histogram against the right side and clipping the highlights or underexposure, pushing the histogram against the left side and loosing detail in the shadows, but there is more to optimal exposure than avoiding the extremes.


Technically if the histogram is safely enclosed within the graph without piling up on the right or left then you might expect that the  Goldilocks Rule would apply, since it is neither too bright or too dark and all tonal levels are recorded. But unlike Goldilocks, every position within the graph is not “just right”. It is here that the “Expose To The Right” (ETTR) rule applies.
 

ETTR
Exposing to the right is an approach designed to get the most from the digital sensor in terms of tonal range, noise reduction and color quality. In practice it means slightly overexposing images, shifting the histogram to the right where digital light sensors work more efficiently.

The Physics
Digital image sensors are linear devices that translate the incident light into an electrical signal which is directly proportional to the intensity of the light. The greater the light, the greater is the portion of the sensor's tonal range that is dedicated to that part of the image.

Low Contrast : Histogram in the Middle

But f stops are not linear, but logarithmic. Every time you drop by one stop you are cutting the light in half and, remarkably, what this means is that the brightest stop on a digital sensor uses one half of the total tonal information. The next stop down gets only one half of what is left and so on until the darkest tones are recorded with a much smaller amount of tonal information. The result is that dark regions are much more susceptible to noise, loss of detail, and posterization in post processing. Digital sensor noise is actually higher in the bright tones, but since there is so much more information (signal) at this end of the range, the high signal to noise ratio (S/N ratio) obscures any noise. Noise can be significantly more intrusive at the dark end of the range, where the smaller signal is susceptible to being overpowered by the noise.

Anytime I start talking about the interaction between linear and

Exposure Right Shifted, ETTR, Not Pretty

logarithmic scales, I know I am in trouble, so let's get to the simple, practical application. Since there is more tonal information available at the bright (right) end of the scale, by increasing the exposure, the darker regions are moved into a portion of the scale that can better record detail with a more favorable S/N ratio. Again the resulting image on the LCD will not be the most attractive, but it will give you the highest quality information to take home to your editing software. This is a real test of your ability to trust the histogram and I have struggled to avoid the natural tendency to "improve" the resulting overbright, washed out image by lowering the exposure to get the prettiest LCD rendering.

An Example

Left Shifted Exposure

Exposed toward Right

 






As an example, here are two exposures taken of the classic old stone bridge in Keene New Hampshire. The first was underexposed, but looked better on the LCD screen. The dark tones are recorded on the far left of the histogram. The second was "exposed to the right", appearing overexposed and washed out on the screen, but the histogram showed that the dark tones were shifted to the right, on a more favorable portion of the scale, and that the highlights were not clipped. The two images looked very different on the LCD screen, but I then processed each in Camera Raw and Photoshop to get the best results. The two images were adjusted to look grossly the same, but examination of the detail, especially in the shadows, shows that the brighter image has smoother tonal range and less noise. ETTR does work!

Detail Comparis

A Few Caveats for ETTR:

First, to get the tonal benefits from ETTR, images must be shot in RAW, with the proper brightness brought back in the RAW converter. In addition to having a much smaller native tonal range (only 256 at 8 bits), JPG images are processed in camera. Their tonal range is "baked in" before the picture ever reaches the computer, leaving little margin for salvaging the over exposed images.

Secondly , as you increase the exposure it is important to avoid going too far, clipping or blowing out the important highlight tones. As I discussed last week, you must trust the histogram and avoid piling up against the right side. Most cameras have an additional visual aid to help avoid clipping. The "blinkys", the flashing lights on the LCD image are there to identify areas of blown highlights. Annoying but very helpful.

Finally, although ETTR has become the conventional wisdom for digital exposure, there is still some controversy about it's value. No surprise, we are photographers, we must debate! Digital sensors are constantly changing and improving. Only experimentation will tell you how valuable this technique will be for you.

So give ETTR a try. The biggest hurtle is to get over the strong desire to make the LCD image look it's best, but here again, in the digital world, getting it right IN the camera means something different than getting it looking it's best directly FROM the camera.

Check out Part One of Getting Exposure Right in the Digital Camera

Read More

Using the Histogram

Getting Exposure Right in the Digital Camera

A while ago, someone asked me how to get rid of the little graph that kept showing up on her camera's LCD screen. I could only assume that she was trying to make it go away so she could see the full size, unobstructed picture of her drunken uncle with the lamp shade on his head? Although I am confident most of my audience understands the value of the histogram, this article is for those who are still trying to figure out how to make it go away - DON'T. In a future article, I will be getting more technical, but this week I would like to show how a simple understanding of the basics can make a world of difference in your control of exposure. Anyway, that little graph is called a histogram and it is the essential key to unlocking optimal exposure, every time.

This week, to continue my series on "Getting It Right In The Digital Camera", I will begin a discussion about getting exposure right in the digital camera. So far I have discussed the control of focus and of perspective, but for me, the improvement in my ability to control exposure has been the most significant benefit of my change from film to digital photography. With film it took great care to assure a properly exposed image and even with the help of light metering and bracketing, I still couldn't be sure that I nailed it until the pictures came back from the lab or were revealed in the development tray. Now it all comes down to the histogram, and with the use of this marvelous tool there is no excuse for a poorly exposed image. If you want to take a bad picture you must simply find another way to do it.

Find the Histogram
The LCD image on the back of essentially all digital camera can be very useful. It is helpful for a rough review of composition, and an even rougher assessment of focus, but it is should not be relied upon for exposure monitoring. The best exposure is often not the one that looks best on the screen. The brightness of the LCD image is affected by the JPG rendering that is done in camera and also on the brightness you have set for the LCD screen itself. The primary reason I look at the LCD is to evaluate the exposure from the histogram. The histogram is simply a visual representation of the range of brightness in your images. Brightness varies from the blackest black on the left to the whitest white on the right. If you are shooting 8 bit jpg images, that brightness range is divided into 256 steps. In RAW at 14 bit the dynamic range is divided up into over 14,000 steps – you should be shooting in RAW. Regardless of the bit depth, at every point on the graph the height of the column is proportional to the number pixels at that level of brightness. 


Study the Histograms

 The best way to understanding what the histogram is showing is to study the graphs for a range of images.   In the wildflower photograph, the histogram shows a full dynamic range with a bias toward darker tones (to the left).  The histogram for the winter picture of the hill-top Oak at Alyson's Orchard, is shifted to the right by the preponderance of bright tones.  The slight tail extending to the far right represents the brightness of the moon whereas the few deep shadows in the tree explain the tail to the left. In a photograph dominated by a bright sunny beach the histogram columns will be shifted far to the right, while a dark, shady forest will be represented by columns bunched over to the left side. An extreme example is seen in the silhouette image of a Frigate on the Galapagos Islands.  Here there is only bright light and deep shadow and the histogram shows the two peaks at opposite ends of the dynamic range.

The histogram of a low contrast image will float in the middle of the graph without extreme whites or blacks. How to best position this truncated graph is a discussion for another time, but think "expose to the right".

 
Control Exposure with the Histogram
Every image will have its own distinct natural pattern, but in digital photography the goal should be to capture the range of brightness that will provide your editing program with the information it needs to show quality detail throughout your desired dynamic range.  There is no "ideal" histogram. The best exposure, and therefore the best histogram, depends on the dynamic range of the subject and the artistic goals of the photographer.

 My routine approach is to lock down my initial settings in aperture or shutter priority, depending on whether depth of field or shutter speed is most important. I don't worry a great deal about other setting until I have taken my first image. That first picture is primarily for metering the scene. Once I have seen where the graph lies I will have a good sense of how the final exposure will need to be approached





Avoid Clipping
 It is most important to check the general position of the graph. To assure that there is no significant loss of detail in the brightest highlights of the image or in the shadows.

If the histogram is shifted too

Right Shifted

far to the right with data piling up against the far side of the graph, the brightest areas of your image will be blown out or "clipped".  There will be no salvageable detail in these regions, and there is nothing that even the most magical editing program can do to bring this back. You will need to reduce exposure to shift the graph to the left.

Highlights Recovered

There will be times when it is ok to blow out the extreme highlights, when it is not necessary to find detail in some highlight areas. The classic example is when specular reflections are recorded, such as bursts of sunlight reflecting off a chrome automobile fender or sparkling light on ocean breakers. In these situations as well, the histogram is essential to controlling the results.  The sunflower picture has a spike at the far right of the histogram, but these "blown" highlights are in the direct sun where no detail is expected or possible.

Sun Clipping

Excessively dark images

Left Shifted Histogram

will be shifted to the opposite side of the histogram, with the columns piling up against the left side. Lost shadow detail can often be retrieved in post processing, but the results are usually excessively noisy with a muddy appearance. Here again the shadow detail can be preserved by increasing exposure, shifting the graph to the right. As with the highlights, there are times when extremely dark areas can help to provide contrast in an image.
 

Shadow Detail Recovered

When dynamic range is high, it may not be possible to protect both highlights and shadows and compromises need to be made. Using the histogram, I will often increase exposure until important highlights begin to be compromised and then back off slightly. This is called "exposing to the right" and although it often does not yield the prettiest picture on the LCD, there are good reasons why,  regardless of the dynamic range, this is usually the optimal approach to positioning the histogram .  But I will reserve that discussion is for another time.

For now, get into the habit of reviewing your histograms. Understand what they represent and then use this to get all the exposure data you will need to fulfill your vision of the scene.  Your drunken uncle probably will appreciate having his indiscretion obscured by the graph and you will get a perfectly exposed record of the event. 

Check out Part Two of Getting Exposure Right in the Digital Camera: Exposing to the Right

Read More