Focus%20On

1
Note to all HPS members viewing this
presentation This material was prepared as
visual aids to a lecture-type of program, and
may not always be very meaningful without the
accompanying narration. Questions may be
addressed to the author at stan_at_sprevost.net
. The color filters for flashes mentioned in
some slides were Rosco filters (or Roscolux) .
A sample swatchbook may be obtained through BH
Photo for one cent plus shipping. I have heard
that they will sell you up to five of the
swatchbooks. The samples are perfect for many
flash units, like the Canon 430EX or 550EX.
Just use a bit of tape to hold the filter over
the flash aperture.
2
Focus On Exposure

• Stan PrevostHuntsville Photographic Society

3
Image Quality

• The technical aspects of image quality include
• Exposure (including white balance)
• Sharpness (including focus, depth of field,
  camera and subject motion, diffraction, and lens
  quality)
• Noise
• Dust
• Lens/camera Flare, Color fringing
• This presentation Focuses on. Exposure.

4
Other Planned Programs in the Focus On Series

• Focus On.
• Sharpness
• Shutters
• Flash
• Perspective
• The Inner and Outer Workings of Digital
  Photography

5
What Is Exposure?

• We talk a lot about it, so we ought to be able
  to rather precisely define it.
• Generally speaking, it is a measure of the total
  amount of light the film or digital imaging
  sensor is exposed to.
• More precisely, it is the product of the
  intensity, or brightness, of the light that
  reaches the film or sensor and the amount of time
  the sensor is exposed to that light.
• Relative to a given intensity/time combination,
  if we double the intensity of the light reaching
  the sensor, then we can half the amount of time,
  and have the same effect on the sensor. Or, vice
  versa, we can half the intensity, and double the
  time.
• In general, if we change the intensity by some
  factor, then we should change the time by the
  reciprocal of that factor to maintain the same
  exposure.

6
What Is Exposure? (contd)

• One may consider the Exposure Index, or ISO,
  setting to be included in exposure, but that is
  actually a film characteristic or signal
  processing issue rather than how much light the
  film or sensor is exposed to.
• Nevertheless, the ISO setting provides a context
  for the exposure, and affects the exposure
  setting. In this presentation, we will
  interleave discussions of ISO with strict
  exposure issues, hopefully without confusion.
• Likewise, we will include white balance in this
  presentation, since it involves the relative
  exposures of the primary colors.

7
Examples of Exposure
f/8
f/5.6
f/4
1/125
1/60
1/30
8
Overexposure Example Loss of Highlight Detail
JPEG As Shot
JPEG, Attempted Tonal Adjustment and White Balance
Raw, Tonal Adjustment and White Balance
2 Stops
707
703
9
How Do We Control Exposure?

• Camera Controls
• Lens aperture (partially block the light path)
• Shutter speed (control the time that light is
  allowed to reach the film or sensor)
• External Lens Attachments
• Neutral Density Filters
• Special Effects Filters (graded filters,
  polarizers)

10
How Do We Determine What Exposure To Set?

• We have to measure the light
• Sometimes, light measurement can be bypassed by
  application of a rule of thumb, such as Sunny
  16, at the expense of accuracy.
• We have to know the film ISO rating, or decide
  what ISO to set on the digital camera.
• We have to evaluate the scene to determine which
  scene brightness values should be placed at which
  image brightness values.
• Or, just set the camera on full Automatic and
  shoot away!
• How do we measure the light?
• Light metering system built into the camera
• External hand-held light meter
• Reflected
• Spot
• Averaging
• Incident
• Take an exposure, chimp the shot, evaluate the
  histogram, adjust the exposure, expose again.

11
Light Meter Characteristics

• All photographic exposure meters are calibrated
  to a middle gray, approx. 18 reflectance.
• Consider metering an 18 gray card
• Reflected light meters, pointed directly at the
  card, measure the light reflected from the card
  and determine the exposure required to render the
  card as a middle gray in the image.
• Incident light meters, usually placed at the card
  and pointed at the camera, measure the light
  falling on the card and determine the exposure
  required to render the card as a middle gray in
  the image.
• Note that in either case, changing the intensity
  of the light source illuminating the card will
  still result in a middle gray image of the card.
• Note that substituting a 90 white card for the
  gray card will still result in a middle gray
  image when using a reflected light meter, but
  using an incident light meter will result in a
  white card in the image.

12
Reflected Light Meters Want To Make Everything
Mid-Gray
White Card Gray Card
Gray Card White Card
f/5.6 1/50
f/5.6 1/250
Metered Spot
Reflectance ratio is 90/18 5. Exposure ratio
5.
13
Built-In Metering Systems

• All built-in metering systems are reflected-light
  meters.
• Built-in light meters usually allow a selection
  of uniform full-frame averaging, center-weighted
  full-frame averaging, spot metering, or
  evaluative metering.
• We all know that focus and exposure are separate
  issues, but the camera manufacturers have decided
  to help us out by linking the two systems. They
  provide multiple focus points across the frame,
  and link the exposure determination to the focus
  point in use. This can be disabled so that only
  the center point is used for exposure.
• Perform experiments to learn how your camera
  metering system works.
• HOMEWORK Cut out a paper white circle, paste it
  onto a black paper background. Or use a small
  flashlight against a dark backgroud. Manually
  focus your camera on the white circle or
  flashlight, move the camera around to move the
  white circle around in the viewfinder. See how
  the exposure indicator responds, using each
  metering mode.

14
Hand-Held Meters

• Can be incident or reflected, or both.
• Can be continuous light or flash, or both.
• Very useful when incident light readings are
  desired, or when the camera is fixed on a tripod
  and you need to get up close to the subject and
  meter different portions of it.

15
General Background

• In photography, we often speak of light intensity
  using the EV system, which is referenced so that
  EV0 corresponds to f/1 at 1 sec at ISO 100. Each
  unit of EV is a doubling or halving of light.
  (Also see the APEX system)
• We also use the term stop or f-stop to refer
  to a factor of two change in light. Even though
  the term stop originally applied only to
  lenses, we now use it as applying to anything
  that changes the light by a factor of two. A
  change of one EV is a change of one stop.
• This use of factors of two is based on tradition
  only there is nothing absolute about it. The
  system could just as well have based on factors
  of three, or ten, or whatever. But two it is.
• A full stop is a factor of two increase or
  decrease in light. A half stop is a 41 change.
  A third stop is a 26 change.
• The high and low limits of exposure for a given
  ISO are usually based on the ability to discern
  detail in the highlights and shadows. Detail is
  represented by small fluctuations in light value.
  For purposes of this presentation, I have chosen
  1/3 stop as the threshold of perception. To keep
  it simple, color changes were not used. Color
  variations can aid in perception of detail.

16
Digital Camera Architecture Refresher
17
Photosensor Saturates Before ADC
Voltage Output
Light Input
Photosensor
A/D Converter
18
Other Color Schemes

• The previous illustration showed a Bayer sensor.
• There are a couple more color sensing schemes
• Sigma uses the Foveon sensor, which uses
  vertically layered color filters, somewhat
  analogous to film, and which makes each photosite
  a full pixel.
• Sony has introduced a slightly different color
  filter array, which instead of RGB (more
  accurately, RGGB), uses Red/Green/Blue/Emerald
  (RGBE). Supposedly this is for more accurate
  color rendition.

19
The General Problem of Exposure

• The scene to be photographed may have a
  brightness range of any value. It may be two
  stops, for a dreamy, misty scene, or 10 or more
  stops for a sunlit waterfall in front of a dark
  rock overhang.
• The final viewing media, whether it be a
  photographic print, a press-printed image, a
  computer screen or a projected image, will have a
  limited capability of representing brightness
  range, maybe more, maybe less than the original
  scene.
• The exposure problem is twofold
• To capture the full range, or the desired range,
  of scene details on the film or digital media so
  that they can be manipulated for the desired
  presentation, and
• To fit the scene brightness range onto the
  viewing media in a visually pleasing manner.
• Between the scene and the viewing media is the
  camera, the film, or the digital equipment and
  software.

20
Media Capability
This image was created in Photoshop, half max
white and half max black.
A glossy print was made at Costco. Using the
built-in exposure meter in my camera, I measure a
reflectance range of 4.6 stops. On the projection
screen used here, I measure a range of ???.
21
Sensor Characteristics

• Now that we have examined the usable brightness
  range of our output media, we ought to look at
  how our camera sensor can handle the scene
  brightness range. This will be restricted to
  digital sensors.
• The next issue will be how to fit the scene
  brightness range onto the output media range,
  which is a central issue in exposure.
• Experiment Use a gray card and a white card in
  the same photos. Use a wide range of exposures.
  In Adobe Camera Raw, measure the values of the
  gray and white cards. Determine the sensor
  range.

22
Sensor (?) Experiment Results
Top Left Curve ACR Defaults
Bottom Left ACR Sliders To Zero
Bottom Right Same as Bottom Left, Log-Log
23
Sensor Experiment Results What Did I Learn?

• Mainly that I have a lot more to learn!
• I naively expected the values measured in ACR to
  be linear with the relative exposure. Instead, I
  learned that the values are determined by the
  settings of the ACR or Photoshop sliders.
• Setting the ACR sliders to zero gets rid of any
  tone curve, other than a gamma curve applied by
  ACR that cannot be eliminated that I know of.
• The bottom curves on the previous slide are close
  to a square root function. Somewhat less
  naively, I would expect a gamma of 2.2, rather
  than 2.
• Cannot explain two stops rather than 2.32 stops
  between gray and white cards.
• The large lesson is that you cannot expect to
  know anything about light ratios by looking at
  tone values in Photoshop. If you change a tone
  value by a factor of two, that will not
  correspond to a stop change in light.

24
White Balance
Sunlight Daylight WB
Open Shade Daylight WB

• As these two photos illustrate, the color of
  light varies with the light source.
• Highly-adaptable human vision doesnt notice any
  difference, but the camera faithfully records the
  difference.

NOTE Histograms were obtained from a selection
on the card to eliminate the grass.
25
White Balance (contd)

• In my experience, electronic flash is close
  enough in color to bright sunlight or open shade
  so as to not cause an objectionable mixed light
  situation. More critical viewers or applications
  may require filtration of the flash for a better
  match.
• How can you color match the flash and ambient
  without a color temperature meter, especially one
  that works with flash?
• First, using a gray card or other reference, set
  an accurate white balance camera setting for the
  ambient light (using presets or custom WB).
• Next, apply filtration to the flash and make test
  exposures using the gray card using different
  filters until the color histogram shows alignment
  of the colors. Try to shade the gray card from
  the ambient light so that the flash dominates.

26
White Balance (contd)

• Automatic White Balance acts in a manner
  determined by the various camera manufacturers,
  and they do not tell us the details. All we know
  is that the camera adapts its color balance
  setting in response to the scene.
• It is often said that AWB works in an analogous
  fashion to full-frame averaging for exposure,
  which would mean that it averages the entire
  scene and applies color correction to make it
  come out neutral gray. To test that hypothesis,
  I performed the following experiment

Daylight WB
Auto WB
27
White Balance (contd)

• My camera manual says that the camera makes an
  adjustment in software to make white areas look
  white. So my suspicion is that it doesnt
  respond to large areas of strong color, but
  searches for bright areas that it thinks should
  be white and then adjusts that area for
  neutrality. Tests are inconclusive.
• The following photos were taken using
  center-weighted exposure averaging, and both
  Daylight WB and Automatic WB settings.

28
White Balance (contd)
Sunlight Gray Card Daylight WB
Sunlight White Card Daylight WB
29
White Balance (contd)

• For general snapshots, AWB wont hurt much but
  probably wont help much. Using the camera
  presets for sunlight, open shade, cloudy,
  tungsten, fluorescent, etc., is not much more
  trouble and can give more consistent results
  without being influenced by the scene coloration.
  
• In some situations, such as photographing
  wildflowers under a forest canopy using ambient
  lighting, using custom white balance, either in
  camera or in post processing, can give superior
  results. Either method requires photographing a
  gray or white reference.
• Using batch WB correction of photos taken in the
  same lighting condition is useful and efficient.
  But.. only if a fixed manual WB setting is used
  for all the photos. This will not work with
  AWB.
• Camera settings for white balance affect JPEGs
  only, they do not change the raw data.
• My conclusion is that AWB is a mystery and
  doesnt seem to do anything useful.

30
Mixed Lighting
Histograms derived from gray card only.
31
Mixed Lighting (contd)
828
Daylight WB
Histograms derived from gray card only.
32
Mixed Lighting (contd)
For reference, the last image is repeated.
Background tungsten, foreground flash. Tungsten
WB.
Now in an attempt to correct the mixed lighting
situation, we introduce a filter over the flash
to warm its color
NOTE Histograms from gray card only
33
Mixed Lighting Homework

• Test your flash against your camera white balance
  settings. See if it better matches flash or
  daylight setting.
• Consider getting filters for your flash to match
  tungsten and fluorescent lighting.
• If you decide to try using filters to match your
  flash to the ambient light, consider the
  following cases
• Tungsten
• Fluorescent (unfortunately, there are many
  colors)
• Open Shade
• Direct Sunlight
• Late afternoon (Golden Hour) lighting

34
Histogram

• Perhaps the most important exposure tool we have
  today, in digital photography, is the histogram.
  The histogram is a plot showing how the
  brightness values are distributed in the data
  captured by the sensor.
• The horizontal axis represents light value, with
  black on the left and white on the right. Each
  point on the horizontal axis represents a value
  of light, described by a number from zero (black)
  to 255 (white). At each of those points, a
  vertical line is drawn, having a height
  proportional to the number of pixels having that
  light value.
• Some examples

35
Histogram (contd)

• Question What Should The Histogram Look Like
  For Proper Exposure?
• (a) Most tones concentrated toward the left
• (b) Most tones concentrated around the center
• (c) Most tones concentrated toward the right
• (d) None of the above
• Answer (d) ! It depends entirely on the scene
  and how it is visualized to be rendered, and on
  the workflow of the photographer.
• There are two fundamental schools of exposure in
  digital photography
• Get it Right In The Camera
• Expose To The Right (ETTR)

36
Expose To The Right (ETTR)

• This technique advocates using as much exposure
  as possible without blocking up the highlights
  (clipping, saturation).
• It assumes postprocessing.
• ETTR was popularized by an article on the
  Luminous Landscape website. The article says
  that ETTR brings two benefits improved
  signal-to-noise ratio, and reduced posterization.
  But it doesnt say how the SNR improvement
  occurs, and implies that it is due to the
  increase in encoding levels available for
  stronger light signals.
• The truth of the matter lies more in two things
  one is the discrete nature of light, and the
  other is the electronic circuit noise that
  competes with low-level signals.
• The number of photons arriving at the photodiode
  per unit time is random and is described
  statistically by a Poisson Distribution. The
  number of photons collected can be described by
  the average number and by how much that number
  varies, called the standard deviation.
• In a Poisson process, the standard deviation is
  equal to the square root of the mean.
• The end result is that the signal to noise ratio
  improves as the light intensity increases due
  simply to the properties of light itself. The
  more light, the less the relative influence of
  the random component of the light.

37
ETTR (contd)

• Always try to get as much light as you can into
  the sensor without overfilling it. Use the
  lowest ISO setting you can.
• You may have constraints in how large an aperture
  you can use or how slow a shutter speed you can
  use. After reaching those limits, crank up the
  ISO to reduce noise in the darker areas (due to
  some subtle technical factors).

38
More Light Less Noise, So ETTR
Canon A650 IS, f/3.5, ISO 800, Daylight WB Both
images normalized in postprocessing for same gray
card value.
0.4 sec
39
Types of Histograms

• Different cameras and different software packages
  will display different kinds of histograms.
• Brightness
• Luminance L .59G .30R .11B
• RGB
• Separate R, G, and B
• Single-channel clipping cannot necessarily be
  detected in luminance histograms.
• The Highlight Alert or Flasher or Blinker may
  operate on single-channel clipping.
• ETTR requires an understanding of the histogram
  and clipping alerts in the camera to be used
• HOMEWORK Experiment to learn how much to adjust
  the exposure compensation by, based on the
  vertical grid lines on the camera histogram.

40
Raw vs JPEG

• Most digicams will only output JPEG files
• Some higher-end PS will output raw and maybe
  TIFF
• Probably all DSLRs of all brands will output raw
  files
• Usually you can select to get JPEG only, Raw
  only, or Raw JPEG
• The question is, if raw is available from your
  camera, should you use it?
• But even before that question, a more fundamental
  question is, What is Raw? For that matter,
  What is JPEG?

41
What is Raw?

• The Raw data format contains unprocessed image
  sensor data, the A/D converter outputs for each
  of the R, G, and B sensels. It has not been
  demosoaiced, gamma corrected, or had any other
  processing.
• The raw data file also contains metadata that
  records all the camera settings used in creating
  the file, including date and time, as well as
  other sensor data such as black reference, maybe
  a blackframe, and a small JPEG image for viewing.
• All image processing is performed in external
  software.
• Lossless compression algorithms are used to
  reduce file size. Nikon employs a compression
  technique that is theoretically lossy but is
  practically transparent.

42
What is JPEG?

• JPEG (or .jpg) is the most commonly used image
  file format.
• It uses a lossy compression algorithm with
  variable parameters to achieve a much smaller
  file size than the uncompressed original, or even
  relative to the same image compressed with
  lossless techniques.
• Lossy vs lossless data compression
• The lossy part of JPEG encoding is based on 8X8
  pixel blocks (nonoverlapping).
• JPEG uses 24-bit color, or 8 bits per color.
• Recommendation If you save your original image
  files as .jpg, it is usually a good practice to
  save any processed derivative files with a
  different file name, so that the original file
  remains unchanged. Lost information cannot be
  recovered.

43
JPEG Qualities

• Advantages of JPEG
• Instantly available, can even be printed directly
  from the camera or shown on a TV
• Does not require postprocessing
• Universally accepted
• Smaller file size
• Flexible as to quality and file size
• Disadvantages of JPEG
• Repetitive opening and saving of the file can
  result in artifacts
• Editing options somewhat limited
• 8-bit color word size limits editing
• Headroom for highlight recovery lost
• Camera-produced JPEGs use in-camera algorithms
  optimized for speed. JPEG production using
  PC-based development of raw data can use more
  sophisticated and flexible software.
• NOTE If you dont want to use raw and dont
  mind some postprocessing, you might can reduce
  the contrast and color saturation settings in
  your camera to allow more scene brightness range
  to be packed into the JPEG 8-bit range, then you
  will have a little more flexibility in your
  postprocessing.

44
ISO

• High-end Canon DSLRs (5D, 1D) use hardware ISO
  gains at all settings except 50 and 3200.
• Lower models use hardware gains at 100, 200, 400,
  800, 1600, but software multiplication or
  division at all in-between settings.
• Recommendation For all but the high-end DSLRs,
  use only the primary ISO settings. Use the
  lowest ISO setting consistent with usable shutter
  speed and f-number settings. Use exposure
  compensation (change the light) when necessary.
• Note Use of the Auto exposure setting will
  allow the camera to
• automatically choose any ISO setting it
  wishes. Avoid this when
• possible.

45
Thats All, Folks!
Discussion Welcome stan_at_sprevost.net