Guide: About exposure (1/3)

[zoossh]

Please do not post in this thread in order to maintain a clean and efficient platform.
Status: Updated till 2009 Jan 03



About Exposure

1/3: Nature and distribution
2/3: Inputs and Dependent features
3/3: Judgement and Control

The very first step in understanding creative photography with alteration of settings, is to understand the different components with the reception of light.




Summary: Nature and distribution

1. Exposure and tones
1.1 Overview & foreword regarding exposure
1.2 Getting down to the tonal factor in light

2. Tones and its component
2.1 Tonal range and dynamic range
2.2 Shadows, midtones, highlight
2.3 Exposure latitude and details
2.4 Ability to push back details
2.5 Adaptive Dynamic Range (ADR)
2.6 Exposure distribution (histogram)
2.7 Exposure and its zonal significance

.

 

[zoossh]

1. Exposure and tones

Basic reading, e.g. ClubSNAP's Photography 101 for Newbies, and book: understanding exposure from Bryan Peterson available at Riceball bookshop
Interesting reading, e.g. Ken Rockwell's always straightforward comments, with a suspected Nikon inclination.


1.1 Overview & foreword regarding exposure

Exposure is often the first thing people will get you to understand first. For sure it is not always the most important thing, but the settings involved in manipulation of exposure have vast and complex relationships affecting various other things, which at this point of time of reading, is too early to summarise as it may confuse you. Exposure is in short, whether you have enough light to form a picture bright enough for you to visualise comfortably.

One important aspect of photography in terms of light reception, is
1. how to set that correct exposure for that situation, and
2. how to alter the settings to get that exposure
3. while getting other effects that is derived from that setting.

For example to simulate a real life thinking process, I wanted to get a silhouette of the tree on a mountain with the blue sky. I wanted the blue to be blue, and be correctly exposed, and hence I set the metering to spot metering and metered at the sky. I got the metering midpoint for midtones for the camera, and have a close-to-blue but a little too bright sky and a dark tree and dark mountain outline. However, I feel that still does not give enough impact and drama, and i did not have a polariser, hence I did an exposure compensation going down to EV -1.3, and that gives me the desired blueness of the sky without appearing too bright and pale-out, and the tree and mountain in pitch dark giving me a clear-cut silouette devoid of unwanted details. This portion is the part on getting the exposure determined.

The alteration of the settings would then be using of the various modes, such as PASM modes to achieve the above exposure. And if I used a aperture priority mode, I would take note that for every f/stop I increased, I would be getting a longer shutter duration. And if I'm using a manual mode, I would have to manipulate both the aperture size and shutter duration, but it gives me the option of altering the choice of resultant exposure while changing the two settings, rather than limiting my settings to achieve the pre-set exposure.

And part 3, I need to watch some certain features would be derived and changed with the aperture size and shutter duration setting. In the same example, I may have to determine a small aperture size at f/22 to increase depth of field of the distant silouette and a house that is much nearer but lighted, while having the reciprocal changes of the shutter duration. This can be achieved with an aperture priority mode, but I realise that the shutter duration is too long and slow to be maintained without a tripod, and i may get handshake, and hence i corrected to f/11 and sacrifice some depth of field but it is still not steady enough. I hence need to change my expectation of that exposure to either doing preset exposure compensation to achieve a lower exposure, or switch to manual mode for easier and faster control in difficult situations where presets need to be changed again and again with time. and by this time, i told myself next time i'm going to bring a tripod.

Why do i need to know so long a chapter about exposure?

This process of trying to get the exposure correct with the various settings and having to keep in mind other factors such as the sharpness due to distance and handshakes, have been a common and complicated process that almost every of us would have to go through, if we decided that we wanted to be in control to achieve some pictures that we visualise rather than leaving to the mercy of the full automatic exposure modes including scene modes.

Some of us are getting the grip, whereas some others are well ahead of us, and some are still struggling. And for those still struggling and those who is totally new, if you do not understand what is being described above, then you will probably need to read up step by step below.

.

 

[zoossh]

1. Exposure and tones


1.2 Getting down to the tonal factor in light.

I realised that before understanding exposure, we got to get to tones, and it is not just in black and white photography is that important, although it is more evident in black and white cos no colors is there to affect our perception of tones.

There is a few properties of light but this is the factor that derives the idea of exposure - Tones.

Tones is about the intensity of light and how much of it get passed through. Every spot in your picture reflects or emits light in different strength and intensity, and they fall onto an area of your sensor. You may have the sun in your frame, and it is very bright, and you may have a silhouette shadow, and it is very dark, both in the same picture. Tones is about that intensity which can refer to a single spot to a small area of similar intensity, whereas exposure is about the overall intensity which can refer to a larger area in the picture and the whole picture.

When it comes to capture, exposure is an overall value that affects the framing, ISO, aperture size and shutter duration choices. Any changes of exposure brings the same magnitude of change in tones in all areas of the picture.

Tones can be confused with hues and saturation. The issue is actually more apparent when we start to talk in depth about how you perceptualise exposure which can be often affected by colors. Also they become important when you are doing photo enhancement. Get this kicked out of the way, and now we can start reading about exposure.

.

 

[zoossh]

2. Tones and its component


2.1 Tonal range & dynamic range

Refer here for more.

Tonal range or tonal gradations refers to the ability of the sensor to have a higher number of tones in the range, with closer values next to each other. The practical portion is that a wide (high) tonal range represents how smooth each tone transit to each other without becoming too distinctly different. It is comparative to a smooth slope from dark to bright for a wide tonal range, while a stepped staircase that show jagged tonal intervals for a narrow tonal range. Insufficient tonal range will lead to increase problems of banding, especially in those older compact camera models.

Dynamic range refers to the ability of the sensor to record very dark to very bright details without losing them to large areas of absolute black or absolute white. It is currently still an unresolved problem in most sensors today. Read clockunder's post to understand it better. For details on achieving high dynamic range can be read under Handling frames of wide tonal discrepancy (Page 12: 5.1).

Speaking in terms of arbitrary numbers of tonal values, for example,
a wide tonal range: 5.1, 5.2, 5.3, 5.4 ..... 10 (50 tones from 5 to 10)
a wide dynamic range: 1, 2, 3, 4, 5 ....20 (20 tones from 1 to 20)

Or making it easier to understand, think of it as a ruler.
Tonal range or tonal gradations can be compared to a ruler with intervals. A picture with wide tonal range is like a ruler that have 1mm intervals while a picture with narrow tonal range is like a ruler that have 5mm intervals. Fine differences in tonal intensity can be recorded.
Dynamic range can be compared to a ruler with a longer length. A picture with a high dynamic range is like a ruler that is 1m long while a picture with a low dynamic range is like a ruler that is only 15cm long. A frame with wide discrepancy of tonal intensity can be recorded.

As quoted by gooseberry and rephrased, analogy is used by comparing tonal range to the number of steps in the staircase and by comparing dynamic range to the height of the staircase. Increased bit depth increased the tonal range and more gradations of tonal levels within the dynamic range, but not necessarily increasing the dynamic range. Having more bits does not increase the height of the staircase (dynamic range), but it increases the number of steps (tonal gradations). This is of course at base ISO. At high ISOs, depending on the implementation in the sensor, having more bit depth can improve the dynamic range due to less highlight clipping. Dynamic range is determined by the electron well depth above the noise floor of a sensor. A sensor will have a fixed number of electrons it can count before it becomes full. This number (and the noise floor) will determine the dynamic range of the sensor. Let's say for example, it can count 60,000 electrons. Using a 12-bit or 14-bit ADC doesn't change the fact that the sensor can only count up to 60,000. But using 14 bits does allow you to have more values in between 0 and 60000 (16384 vs 4096) - thus "smoother" transitions or gradations for your colours etc. A kilometre is a kilometre, no matter whether you measure it in metres or centimetres, but measuring it in centimetres gives you more "accuracy". Fuji's sensor has more dynamic range because it uses two pixels at each location (one to obtain the overflow of electrons).

A higher dynamic range can gives very good color reproduction and is very suited for giving good skin tones. However the file size may get very big when saved as RAW (uncompressed or minimally compressed lossless) format.

As quoted by deadpixel and rephrased, the new Fujifilm S5pro DSLR captures images with a sensor that has 12 million paired photodiodes and outputs them as 6MP images. However, as it captures images in 14bit color instead of the usual 12bit, its dynamic range is much greater, resulting in RAW files that are huge. When set to 'Wide' dynamic range, its RAW files come to about 25MB each. It does improve color reproduction as you get a much greater range per color. This will give you much improved color/shade transitions but is more shades than a human eye can differentiate at a glance. Consider the following calculations (Red Shades X Green Shades X Blue Shades):

JPEG 8Bit color: 256 X 256 X 256 ~ 16.7 million colors
RAW 12Bit color: 4096 X 4096 X 4096 ~ 68.7 billion colors
RAW 14Bit color: 16384 X 16384 X 16384 ~ 4.4 trillion colors

After some thoughts, i think the current terminologies are too confusing. Tonal range should be better phrased probably as tonal smoothness whereas dynamic range should be better phrased probably as tonal latitude or even termed as tonal range itself. At the same time, perhaps there should be terms to differentiate actual exposure on sensor (actual light amount, without ISO factors), overall exposure (with ISO setting, actual effect on data collection) and zonal exposure (by specified areas in frame for actual visual impact).



2.2 Shadows, midtones, highlight

This concept is important. The above described the degree of tonal intensity and breaks them into 3 parts - shadows, midtones and highlight. There is no definite division as to what is what, as the whole range of tones is in a continuous spectrum. The dark parts are the shadows, the bright parts are the highlights and everything in between are the midtones.

This is used as descriptive terms so that appropriate measures can be done to control the proportion of the 3 components. The ability to assign appropriate amount of highlight, midtone and shadow contrasting against each other is an important way of introducing impact and a spatial relationship.

An example of significant proportions of all 3 components,

It is also important to note that the current digital systems have a natural tendency to neutralise everything to midtone (midgrey 18%) and averages highlights and shadows around that value. However the world is not seen as such, there are things that are predominantly darker than usual or brighter than usual, but the sensor is not able to tell that a rounded object is a light bulb or a dark blob, and since exposure is a variable, the digital system will meter to neutralise it.

.

 

[zoossh]

2. Tones and its component


2.3 Exposure latitude and details

Wikipedia's definition is in the link, but i think that is insufficient in understanding. For me, latitude refers to the range of the degree of different tonal intensity (from very dark to very bright) that still allows details to be deciphered, i.e. not becoming total shadow or blown highlights. The total shadow or blown highlights marks the ends of the latitude. In a wide latitude, the variability of exposure values range from very dark to very bright, with details still all visible, no matter how subtle they are. In a narrow latitude, the variability of exposure values range from only reasonably dark to reasonably bright, with details still visible in a proportion of the picture and the rest of the picture falling into total shadow or blown highlights at the ends of the latitude.

The easier way to imagine it is to compare a wide tonal contrast situation, such as bright morning sky against a backlit subject like the trees in the forest, to a very fat person, and the latitude, to a pillar. If the pillar is narrow, i.e. narrow latitude, the fat person will not be able to covered by the pillar. He either has his right body or left body showing outside of the pillar, which corresponds to total shadow and the blown highlights beyond the ends of the latitude.

The ability to see details depending on various factors in similar tones, such as the ability of the sensor to differentiate fine intervals, such as arbitrary values of 1.0, 1.2, 1.4, 1.6, 1.8, 2.0 as compared to the ability to do so, i.e. 1.0, 1.5, 2.0. Of cos, the medium that displays your picture is also important.

For example, this is a typical silhouette picture but it is not an absolute silhouette. There are details of some dark tones which is not so dark and should be able to be contrasted against another darker tones as a different tone, thereby giving a perception of details. If your monitor have proper calibration and performance, you should be able to see the details in the cliff shadows. I've noted on a non-calibrated monitor that casts a reddish tinge to dark colors and would not decipher the details in the shadows. In that case, if you also cannot see the details in the shadows in the following picture, I would suggest that you upgrade to a better monitor, or consider getting it calibrated (either buying a calibrator or borrowing from friends)

Depending on your aesthetic judgement, you can choose to retain or lose details. In order to retain details in the highlights, one may have to choose an overall decreased exposure, such that the highlights are not too bright for the sensor, but that means the midtones and shadows will get darker in the same magnitude of shift. In order to retain details in the shadows, one may have to choose an overall increased exposure, such that the shadows are not too dark for the sensor, but that means the midtones and highlights will get brighter in the same magnitude of shift.



2.4 Ability to push back details

Usually towards the end of the latitude, there is actually different tones that is still recorded as different values but is difficult to visualise well. Upon post processing, the exposure range can be altered with the gaps of the different values opening apart, hence allowing better visualisation. Both film and digital sensor has better details perceptibility in the shadow region. For digital, to optimise the ability to push back details, the data input must be faithfully preserved and hence RAW format should be used. Jpeg compression may cause the data input to be lost and the values at the end of the latitude merged into a single value, and pushing back of details become impossible.

Quoted from Godfrey DiGiorgi in photo.net,

"positive and negative films respond differently, and likewise digital sensors respond differently depending upon how you have the camera set to store exposures. And the characteristics of the two media are different at the limits besides.

Film latitude depends upon the type of film and its processing. Falloff at the limits is gradual.
Slides about -1.5 to +0.6 EV latitude.
Color negative about -1 to +3 EV latitude.
B&W negative (dependent upon processing) -1.5 to +4 EV latitude

Digital sensor latitude is dependent upon whether you are storing exposures in JPEG or RAW format. The underexposure limit is a soft limit where detail and noise cannot be distinguished. The overexposure or saturation limit is a hard clip when the sensor runs out of numbers to identify gray levels. Whether you know how to use the RAW conversion software to maximum advantage also makes a difference.

JPEG about -1.5 to +0.3 EV
RAW about -2.5 to +0.6 EV"



2.5 Adaptive Dynamic Range (ADR)

Read Ken Rockwell's article for more details and the illustration of its difference. This is also called by Nikon as Active D-lighting, which essentially refers to in-camera processing that suppresses the strong highlight and bring it subtlely down to slightly lesser exposure to avoid clipping and to retain color vibrancy. This feature is present in the newer models for Nikon.

Another in depth description in photo.net by Anthony Beach dated 2008 Feb, he finds that this setting does not alter centre weighted metering but does decrease exposure value for matrix metering, so he recommended shooting RAW using low ADR with centre weighted metering as the underexposure in matrix metering in conjunction with ADR causes increase in noise in the shadow areas. There is extra processing noted but may not be practically significant.

.

 

[zoossh]

2. Tones and its component


2.6 Exposure distribution (histogram)

Understanding Histograms from luminous landscape

Histogram is a graphical distribution of dark to light tones from left to right. A way of understanding it is to open up one of your file in photoshop, look for the levels function, and play with the sliders to see the effect. It is a tool that helps some people to judge exposure in camera after capture of image, or it appears as a data representation of tones in photoshop which helps in understanding of what you are doing to your picture.

Essentially, it counts the quantity of pixels that have the same tonal intensity, and then put them from the darkest to the brightest. Imagine last time your teacher or school used to put charts of the number of people scoring D, C, B, A grades from left to right.

Alternatively, imagine you have a box full of 1cent, 5cents, 10cents, 20cents, 50cents and 1dollar coins all mixed up and you do not know how much value it has, neither can you judge accurately what is the proportion of the 1 cents and the 1 dollars coins. Histogram basically put the smaller 1 cents coin ont the left and the bigger dollar coins on the right, and stacks the various coins by its values along an axis. This is basically the same as the tones. The darker tones (shadow) with less light is put on the left while the brighter tones (highlight) with more light is put on the right, and the midtones with increasing tonal intensity goes from left to right.

It is to note that histogram is a purely quantitative account of the tonal distribution and not a spatial account of the actual tonal distribution. Whether you want to put the dark tones on the left, on the right, in the middle, 1/3 in top left and 2/3 in bottom right, is all not represented in the histograms. These important but hardly scientifically derivable placement of tones is part of the vague and big word, composition. You can compose the positions of tones, colors, lines, forms, details and textures, and how each element stands next to each other is contrast, and again contrast is another big word meant to refer to many things.

As such, even though some may find histogram useful, there are others who will rely on using naked eye on the LCD display to judge tones and composition together, and skipped the step of using histogram. Histogram does give a good end point if one is concerned about the end points and not crossing it, but not using histograms in between shooting makes things easier and faster too.

.

 

[zoossh]

2. Tones and its component


2.7 Exposure and its zonal significance

This is closely related to the sensor's dynamic range or latitude, but is discussed over her as the various light variables is not to be considered only on the overall exposure itself, but also on the effects on the composition and distribution of the area/zonal exposure.

We made our judgement on the exposure based on the media it is shown on. This experience of whether the exposure is appropriate for a particular situation will then affect the way we measure the exposure required via metering modes, and alter the exposure required for that output via exposure compensation, and finally with the exposure required known, we alter the aperture size, shutter duration and ISO setting to made up that amount of exposure.

The judgement of whether the exposure is appropriate or not, depends on our eyes and how we feel about it - it is subjective but in many cases, is common to most of us, through our common experience in life and how we visualise successful photograph printed in the daily medium, the same as how we see beauty in man and women, with which subjectiveness does exist but consensus can often be made.

If the resultant image appears too dark in overall, it is described as underexposed. And if the resultant image appears too bright in overall, it is described as overexposed. Whether it is too dark or too bright, this depends on what we perceived to work the best for the image and that situation.

On a closer note, we can also go deeper into judging if the exposure is appropriate for a certain area of interest or dis-interest on the photo, i.e. by compartmentalising into different areas and judging each area's exposure individually. For example, we would expect the sun to have an "over"-exposure or likewise a featureless white wall, but we would like to preserve the details on a face, or preferentially like it to be bright enough to distinguish itself from the surrounding. And yet, there are certain areas that we would like to have it darkened, be made into a full silouette or with some dim details.

Instead of judging by overall exposure, exposure can be based on just the perceived important areas. That is for example in the following picture. The peripheries are regarded by the photographer (me) as unimportant areas and the important areas are regarded as the face of the people in the deal, the cloth and the central background. Hence if the exposure of the important areas works right for me, it is right and the less important areas can be left in total blackness, which afterall is a night scene. If i want all the area to have more or less proper exposure, i.e. having more details in the peripheries, i may have to sacrifice the main area by slightly increasing the overall exposure to give the shadows some details and as a result overexposing the main area. It can also be achieved to a lesser extent by post processing and lightening the shadow areas to bring out the details but i wouldn't do that as the effect is limited (details brought out of the shadows are duller) and the pitch black corners contrast well with the bright areas.

And blown highlights are not always undesirable. There are certain occasions where bright areas where they emit light or reflect light, they ought to be blown on where they are. The way these light cast highlights and shadows give a 3D effect to an otherwise dull topic.

On a higher level, i think the zonal system works on fairly similar principles. Those professionals and experts should know them well, but i have not much of an idea except knowing for its existence and it seems commonly mentioned in reference with Ansel Adams and his black and white photography of the classic canyon. I still remember when i first met a group of photographers about a year ago, a girl mentioned about Ansel Adams, and i blurted out unknowingly "Who's Ansel Adams? Is he very famous?". I got no reply and a stare. Always find it funny. Anyway someone have recommended this text, but i have yet to really learn about from it.

As for the newbies, judgement of overall exposure can be further discussed later in subsequent links.

.