Practical HDRI

Practical HDRI


242 Pages


Practical HDRI, 2nd Edition, by pro photographer Jack Howard, leads you into the the new frontier of High Dynamic Range Imaging, a multi-shot technique to digitally capture, store, and edit the full luminosity range of a scene in ways not possible in a single captured image. Fully updated for 2010, the 2nd Edition covers the HDR process from image capture through post-processing for web and print. Practical HDRI, 2nd Edition is richly illustrated with step-by-step tutorials for creating professional results using the leading HDR software titles, including the latest versions of Adobe Photoshop, Photomatix Pro, FDRTools, Dynamic Photo HDR, and HDR PhotoStudio.

Howard instructs from experience as a photographer and a writer, with an emphasis on making the HDR process work for you.

Topics include:

Step-by-step tutorialsBasic and advanced workflows and workaroundsWeb and print optimizationFile management, keywordingAdvice on cameras, gear and software

HDR imaging can be challenging and frustrating for the uninitiated, as it involves both field and computer expertise. However, Jack Howard explains this complex subject matter in a practical way that will not intimidate the beginner, nor bore the experienced photographer. The emphasis is on the creative process and how to make it work for you-rather than the science behind it.



Published by
Published 28 August 2010
Reads 55
EAN13 9781457165924
Language English
Document size 50 MB

Legal information: rental price per page €. This information is given for information only in accordance with current legislation.

Report a problem

Practical HDRI
High Dynamic Range Imaging for Photographers
2nd Edition
Jack Howard
PHDRI-2e_Book.indb 1 6/10/10 2:53 PMJack Howard,
Editor: Joan Dixon
Copyeditor: Judy Flynn
Layout and Type: Terri Wright Design,
Cover Design: Helmut Kraus,
Printer: Tara TPS Co., Ltd. through Four Colour Print Group
Printed in Korea
ISBN 978-1-933952-63-5
2nd Edition
© 2010 by John M. Howard (aka Jack Howard)
Rocky Nook Inc.
26 West Mission Street Ste 3
Santa Barbara, CA 93101
Library of Congress Cataloging-in-Publication Data
Howard, Jack, 1972–
Practical HDRI : high dynamic range imaging for photographers / Jack Howard. — 2nd ed.
p. cm.
Includes bibliographical references and index.
ISBN 978-1-933952-63-5 (alk. paper)
1. Photography--Digital techniques. 2. Image processing--Digital techniques.
3. Single-lens refex cameras. 4.  High dynamic range imaging.  I. Title.
 TR267.H688 2010
Distributed by O‘Reilly Media
1005 Gravenstein Highway North
Sebastopol, CA 95472
Many of the designations used by manufacturers and sellers to distinguish their products are claimed as
trademarks of their respective companies. Where those designations appear in this book, and Rocky Nook
was aware of a trademark claim, the designations have been printed in caps or initial caps. Tey are used in
editorial fashion only and not for the beneft of such companies, nor are they intended to convey endors-e
ment or other afliation with this book.
No part of the material protected by this copyright notice may be reproduced or utilized in any form,
electronic or mechanical, including photocopying, recording, or by any information storage and retrieval
system, without written permission of the copyright owner. While reasonable care has been exercised in
the preparation of this book, the publisher and author assume no responsibility for errors or omissions, or
for damages resulting from the use of the information contained herein
Tis book is printed on acid-free paper.
PHDRI-2e_Book.indb 2 6/10/10 2:53 PMFor the one and only Corey Leigh—my wife, best friend, confdant,
and partner in all things. I stand by her and she stands by me and we now walk
this world together with our beautiful daughter, Avery Rose.
Our journey continues.
PHDRI-2e_Book.indb 3 6/10/10 2:53 PMTable of Contents
xi Preface
1 T e Big Lie about High Dynamic Range Images
7 T e Little Lie about Low Dynamic Range Images
11 Crunching All the Brightness into a Much Smaller Space
14 CHAPTER 1 Cameras and Gear for High Dynamic Range Imaging
15 A Little Bit of Good News, a Little Bit of Bad News, and Some
Serious Gear Envy
15 Two DSLR Features for Easier HDR Shooting: Auto Exposure
Bracketing and Burst Mode
17 A Trusty Tripod Is a Good T ing!
17 Let’s Not Forget about Memory
18 Lens Envy
18 Filters
18 Cable Releases, Tethered Shooting, and Wireless Remotes
19 Equipment Essentials
20 CHAPTER 2 Composition, Framing, and Exposure Basics
21 T e Rule of T irds
22 Lines, Layers, and Textures
24 T e T ree Intertwined Exposure-Setting Functions of your DSLR
24 Aperture and Depth of Field
24 ISO Sensor Sensitivities and Sensibilities
24 Shutter Speed: How to Change Exposure with the Least Impact for
Your HDRs
27 Relative and Absolute Exposure Values
PHDRI-2e_Book.indb 4 6/10/10 2:53 PM28 CHAPTER 3 Popular Breeds of Lenses for High Dynamic
Range Imaging
29 Rectilinear Wide-Angle Lenses
30 Two Flavors of Fisheyes
31 Swinging, Bending Optics
31 Big, Expensive, Perspective-Fixing Lenses
33 An Economical and Fun Way to Mess Around with Focal Plane
Shift, Curved Field Lenses, and Fisheye Eff ects
33 Macros: Make the Little World as Big as All Get-Out!
35 Normal and Telephoto Lenses—Both Zooms and Primes
36 CHAPTER 4 Capturing Images for High Dynamic
Range Imaging
37 RAW vs. JPEG vs. RAW plus JPEG (plus TIFFs!)
38 Recognizing High Dynamic Range Photo Opportunities and
38 See the Gull? T is Is Not an HDRI Op
39 Oh Grand! An HDRI Opportunity!
40 Metering for High Dynamic Range Imaging
40 First T ings First: Take Your Camera Out of “Easy” Mode!
41 Determining Exposure: In-Camera Metering is Your Friend but
May Not Always Be Right
43 Evaluative and Center-Weighted Average Metering
43 Spot and Partial Metering
43 Exposure Bracketing
43 First T ings Burst
44 Check Your Camera Manual for AEB Options
46 Full Manual Mode: Retro with a Modern Twist
46 Be Still, My Bracketing Camera
47 Ready for Some More Gambling? Moving Objects, Deghosting,
and More Risky Techniques!
48 File Management, Storage, and Back-Up
48 A Little Bit of Organization Goes a Long Way
49 Bracketed Source Image Importing and Keywording
52 Image Organization, Stacks and Ratings, and Even More
56 Saving Back to the Stack
PHDRI-2e_Book.indb 5 6/10/10 2:53 PMvi Contents
60 CHAPTER 5 HDR Generation from your Bracketed Photos
61 Before We Start: A Visual T umbnail Viewer Is a Good T ing!
61 Patience Is a Virtue!
62 A Little Bit of Goldilocks Syndrome Going On
63 Converting RAWs to Minimally Processed TIFFs
64 Merging of Images to HDR in Adobe Photoshop CS5, Photomatix
Pro 3, FDRTools Advanced 2.3, Dynamic Photo HDR 4, and
HDR PhotoStudio 2
64 Merging to HDR in Adobe Photoshop CS5
70 Merging to HDR in Photomatix Pro 3.2
74 Merging to HDR in FDRTools Advanced 2.3
78 Merging to HDR in Dynamic Photo HDR 4.6
81 Merging to HDR in HDR PhotoStudio 2
84 CHAPTER 6 Advanced HDR Merging Techniques & Tricks
85 Photomatix Pro 3.2 Deghosting and Other Tricks
86 Photomatix Pro’s Batch Processor
87 T e Single File Conversion Automator
88 FDRTools Advanced 2.3 — Advanced Tricks and Deghosting
90 FDRTools’s Creative HDR Merge Function
90 FDRTools xDOF Merge Function
90 Batch Importing of Source Images with FDRTools
92 Dynamic Photo HDR 4 — Deghosting and Advanced
Merge Techniques
94 Smoother, Sharper, or Custom Curves
95 Batch Merging with HDR PhotoStudio 2
96 Deghosting with Adobe Photoshop CS5
96 Adobe Photoshop CS5 Extended – Deghosting with Smart Object
Stacks at Each Exposure Value Level before HDR Merging
102 CHAPTER 7 Working on Your HDRs in 32-Bit Space
103 Photoshop’s 32-bit Toolkit
103 Making Selections with the Lassos, Wands, and Pen Tools
103 T e Lasso Tools
106 T e Quick Selection Tool
107 Complex Masks Using Color Range and Bit-depth Drop-down
109 Making Selections from Paths
110 Image Editing Tips in Adobe Photoshop in 32-bit Space
110 Image > Adjust > Exposure
112 Saturation Ef ects via Levels > Midtones Adjustments
PHDRI-2e_Book.indb 6 6/10/10 2:54 PMContents vii
113 Built-in Photo Filters and Customizable Photo Filters
114 Adjusting White Balance with Levels
115 Light-Painting Brushes
117 Dodging/Adding Light
119 T e Rubber Stamp for Spotting and Cleanup
120 Tuning and Toning HDRs in HDR PhotoStudio
120 Getting a Feel for the HDR PhotoStudio Work Paths
122 Working with Recipes
123 Rotating, Cropping, Resizing, and Whatnot
124 Toning and Tuning your Image with the HDR PhotoStudio Toolkit
124 Display Brightness Up/Down
124 Veiling Glare Tool
125 Def ne a Black Point with the Veiling Glare Eyedropper
126 T e Brightness/Contrast Tool
129 T e Shadow/Highlight Command
130 T e Sharpness Tool
130 T e White Balance and Color Tuner Tools
132 T e Saturation Tool
133 T e Tint Tool
134 T e Noise Elimination Tool
135 Saving Your HDR Image in HDR PhotoStudio
136 CHAPTER 8 Tone Mapping High Dynamic Range Images
137 T inking of Your 32-Bit HDR Image as a RAW File
137 Two Flavors of Tone Mappers: T ink Globally or Act Locally
139 Faking It
140 Halos, Inversions, Hypersaturation, and Other Weirdness with
Tone Mapping
140 Goodbye High Dynamic Range Image—Hello Tone-Mapped Image!
143 Tone Mapping High Dynamic Range Images with Adobe
Photoshop CS5
143 Build Yourself Some Keyboard Shortcuts
144 Photoshop’s “Boring” Global Operators
144 Exposure and Gamma
146 Highlight Compression
146 Equalize Histogram
147 Local Adaptation: Photoshop’s Local Operator with a Cornered
Curve Control and Lots of Sliders
149 Tone Mapping with FDRTools Advanced 2.3
150 Tone Mapping with Simplex
150 Tone Mapping with Receptor in FDRTools Advanced
151 Tone Mapping with Compressor in FDRTools Advanced
156 T e FDRTools Advanced Photoshop Plug-In
PHDRI-2e_Book.indb 7 6/10/10 2:54 PMviii Contents
157 Tone Mapping with Photomatix Pro
157 Tone Compressor: Photomatix Pro’s Global Operator
161 Tone Mapping with Details Enhancer in Photomatix Pro
164 Exposure Fusion in Photomatix Pro
167 Bit-Dropping and Tone Mapping with HDR PhotoStudio 2
169 Dynamic Photo HDR: Tons of Tone Mappers and More!
170 T e Tone Mappers
174 A Host of Features, Buttons, Orbs, Presets, and Whatnots
182 Little Planet via Process and Edit
184 CHAPTER 9 Post Tone Mapping Image Optimization
185 Post-Tone Map Tutorials that Make No Sense to Us
186 T ree Big Issues with Tone-Mapped Images
187 Color Prof les
189 Histogram Spread
190 Ultrasaturation, Hypervividity, and Gamut Warnings
192 Rekindling the Wet Darkroom Days: Digital Test Strips
194 Making Adobe Camera Raw 6 Your First-line Image Editor for
Tone-Mapped Images
197 Working Up Your Tone-Mapped File in Adobe Camera Raw
198 ACR’s Tools
202 T ree Cool New Tools Added to ACR Since the First Edition
204 Tone Curve and Basic Exposure Adjustment Settings in ACR
209 T e Hue, Saturation, and Luminance Sliders
209 Noise and Sharpening
212 Saving and Loading Presets
212 Some Final Tweaks with Photoshop
213 Dealing with T ose Pesky HDRtifacts When Something Moves
215 Healing and Cloning in Photoshop Instead of ACR
217 T e Ultra-Subtle Sweetening Layer in Photoshop CS3
222 Epilogue
222 Where Do We Go from Here?
227 Index
PHDRI-2e_Book.indb 8 6/10/10 2:54 PMPreface
ince the frst edition ofP ractical HDRI hit that, quite simply put, is the purpose of this
the bookshelves in September 2008, much book. I‘m going to stay as far away from the Shas happened in the world of high dynamic mind-boggling mathematics running in the
range imaging (HDRI). HDRI is now, in certain background as is humanly possible. I focus on
quarters, a four-letter word, so to speak, due to real-world advice, both in the feld and in the
an overwhelming association of the term HDR digital darkroom. Tere are times when I will
with (and exclusively with) a certain style of hy- have to get ultra-geeky to explain this concept
per-saturated, ultra-detailed tone mapping. But or that operation, but I promise I‘ll try to keep it
high dynamic range imaging encompasses so to a minimum. Tis is a specialized photography
1much more than this single style of tone map- how-to book, not a college math text!
ping. In this book, we’ll holistically explore the Tis book is not meant to be a comprehensive
HDRI process, including tone mapping, giving introduction to all aspects of digital SLR
you the skills and know-how to make your im- photography. Te focus is on what you need to
ages photorealistic, surrealistic, or somewhere know and what works for HDRI photography.
in between. So, for example, there’s virtually nothing in this
Maybe you‘ve already been experimenting book about autofocus tracking of fast-moving
with HDRI techniques, or maybe you simply subjects, but I do touch on when to use
singlerecognize it as the next big thing and want to shot autofocus and when manual focus is better
ramp up on this cool imaging technology as since this is critical to HDRI workfows.
soon as possible. By the time you read this book I presuppose a basic understanding of still
cover-to-cover, you‘ll have the foundation for photography with a DSLR or advanced compact
making HDRI work for you. camera. I’m not suggesting you need to be a
High dynamic range imaging with a digital grizzled pro with decades of imaging
experisingle lens refex (DSLR) camera is a slightly ence to grasp the advice and workflows in this
more complicated process than simple single- book, but I expect most readers to have more
shot capture and processing. Tere are new than an inkling of how to take some creative
terms to grasp, new software workfows to mas- control of their camera. Yes, I know all too
ter, and a host of image-degrading challenges to well that camera manuals aren’t exciting by
overcome at every step along the way. Sounds a any stretch of the imagination, but if you take
bit scary, doesn‘t it?
1 If you do want to do a deep-dive into the numbers, math, and Don‘t worry. Once you get comfortable with
science behind HDRI, I highly recommend Te HDRI Handbook: the HDR workfow and mindset, it‘ll become
High Dynamic Range Imaging for Photographers and CG Artists by
second nature. You‘ll fnd your HDR-eye. And my friend Christian Bloch, also published by Rocky Nook.
PHDRI-2e_Book.indb 9 6/10/10 2:54 PMx Preface
the time to really get to know your camera, Adobe Photoshop CS5 have been signifcantly
you will get a lot more out of your shooting upgraded for the frst time since CS2, and
experiences, both with HDRI and traditional many amazingly useful 32-bit features that
single-shot imaging. were previously available only in the Extended
Tis book is DSLR-intensive. Tere are a version of Photoshop have now migrated to
handful of advanced compact interchangeable the non-Extended version for the frst time.
lens and integrated lens electronic viewfnder Te powerful imaging engines behind Adobe
cameras that can produce high-quality source Photoshop now have a much-improved user
fles for high dynamic range imaging, and most interface with a deghosting option during
of the advice and tips given can be translated to merge. It is a marked improvement to the user
the handful of non-SLR digicams with just some experience, both in terms of the expanded
minor adjustments. Even today’s exceptionally feature set and interfaces. Tis is great news for
afordable entry-level DSLRs and interchange- HDR photographers!
able lens compacts ofer great image quality Next in the book we’ll explore tone mapping.
(especially at their lowest ISOs) and the base set Tis is where all the information in the HDR
of camera controls necessary to get started in image is translated back down to traditional bit
top-notch high dynamic range imaging. spaces for display on conventional computer
Get ready to cover a lot of ground in the monitors, for sharing on the Web, and printing
following pages! Te frst chapters cover in-the- for display. We’ll explore the full scope of tone
feld shooting techniques, tips, and tricks for mapping for various results ranging from subtle
successful HDR source image capture as well to surreal.
as lens, gear, and gadget tips geared toward Finally, we’ll look at using the tools available
the HDRI experience. We’ll also take a look at in Adobe Camera RAW and Adobe Photoshop
fle-management issues specifc to HDRI, in CS5 to optimize your HDR images after tone
a section ofc hapter 4 that is much expanded mapping for both Web and print.
upon from the frst edition. Ten we’ll jump I touch briefy on Adobe’s popular Lightroom
into HDR merging of the source fles to high-bit for HDR merge initiation and fle management,
high dynamic range images for both basic and but, it should be noted, Lightroom isn’t
curchallenging situations. rently set up to thumbnail most 32-bit fles, and
What was a small section ofc hapter 6 in the it simply passes of fles to other programs for
frst edition is now a full chapter in the second the heavy lifting involved in HDRI, so it isn’t
edition that focuses on image editing in 32-bit really a major player in HDRI.
space with Adobe Photoshop CS5 and HDR All the software explored in depth in this
ediPhotoStudio, either prior to, or instead of, the tion ofP ractical HDRI is cross-platform software,
bit-dropping tone-map processes we’ll explore and user experiences are virtually identical
later in the book. whether you’re running the software on a Mac
HDR PhotoStudio is an exciting new player in or a Windows PC. Tere may be some slight
the feld of HDR photography, and it’s the only cosmetic variations between the platforms, but
new software title to be included in the second this isn’t anything to worry about. It is quite
edition of this book. Te HDRI workfows of interesting to see how Mediachance has made
PHDRI-2e_Book.indb 10 6/10/10 2:54 PMDynamic Photo HDR cross platform: It ships the to you, you will have established a solid
foundaMac version inside a stand-alone PC emulator! tion by the time you work through the chapters.
We haven’t included software downloads on Tis book focuses on pragmatic workfows, a
a disk because the smaller software companies healthy dose of tips and tricks, and real-world
update and upgrade at such frequent rates that advice from my own trials, errors, and
sucthere will likely be a newer version available cesses, all with the intention of helping to make
online before this book is shipped to you. But your once-impossible photos look as amazing as
I do provide links to each program’s free trial possible.
download site so you can try before you buy— I have revisited every work path and screen
and you’ll be absolutely certain that you’re shot in the frst edition and updated them as
installing the freshest version of each program. necessary. When there has been no meaningful
Also, I have always believed that the majority change to a given operation since the frst
of photographers are most invested in their edition was published, you may catch some
own images, so I’m not including downloadable shots of older versions of programs. I have
image samples in this book. Instead, I encour- clarifed some passages that may have been a bit
age you to go out and create your own images unclear and have also made a number of small
following the advice in the early chapters—and typographic corrections for the second edition.
reap the rewards by practicing with your own In short, I have built on the strengths of the
amazingly worked-up HDR images that you’re frst edition to make an even stronger second
ready to share with the world! edition.
It is my goal to get you ramped up on HDRI as I truly hope you fnd this book informative,
quickly and as painlessly as possible. If you’ve useful, and enlightening.
already been experimenting with HDRI, the tips,
tricks, and new ideas found in this book will help — Jack Howard
you improve and refne your craft. If HDRI is new June 2010
PHDRI-2e_Book.indb 11 6/10/10 2:54 PMIntroduction
PHDRI-2e_Book.indb 12 6/10/10 2:54 PMIntroduction 1
The Big Lie about High Dynamic clippings of the luminance values—either
due to the limitations of the digital camera’s Range Images
sensor to capture light across a broad exposure
here is not a single high dynamic range range or as a deliberate ef ect applied in
post(HDR) image published in this book, de- production. Tspite the title. Not a one. Sure, there are In either event, high-key (highlight-clipped)
screen shots of 32-bit high dynamic range im- and silhouetted (shadow-clipped) images can
ages, but these were grabbed from traditional have impact. High-key imagery is very popular
monitors, which cannot display the full lumi- in fashion, food, and advertising photography
nance and tonality of an HDR image in all its to draw the viewer’s eye directly to the
photogglory. rapher’s chosen focus point—and who hasn’t
You are looking at either 8-bit per channel snapped a shot of the shadowed outline of a
screen grabs from LCD computer screens or distinctive object such as a cactus, palm tree, or
16-bit per channel TIFFs of image f les that architecturally signif cant structure against a
were sent to prepress and converted into CMYK dramatic sky? Both of these styles of
photograhalftones optimized for display on this paper phy take deliberate advantage of the limitations
stock. of the camera’s sensor and employ white clip
Every brightest white you see in the images in and black clip as an aesthetic consideration. But
this book has an equal RGB value of 255, 255, outside of certain situations where a conscious
255; a LAB value of 100, 0, 0; and a CMYK value decision is made to exploit the capture chip’s
of 3%, 2%, 1%, 0% (for a very light, neutral ink limited sensitivity, the dynamic range of a
senlaydown just a touch below the paper stock’s sor can be a vision-stealing hurdle between you
brightness rating of about 92). And every and your dream shot.
darkest shadow has an RGB value of 0, 0, 0, a
LAB value of 0, 0, 0, and a CMYK value of 80%,
69%, 69%, 92% (dot gain makes this amount of
ink appear a true enough black without
overinking). In truth, these two extremes represent
A BIT MORE ABOUT “BITS” Technically, I’m talking about the number of
bits per channel, so when I say 8-bit, 16-bit, and 32-bit throughout this book, I
am “shorthanding” for x bits per channel, which would generally be 24, 48, and
96 bits total per format. But in practice, some of the bits are used for things
other than color information (or not used at all). And some HDR formats use
half-type data, so they don’t actually employ 32 bits for storage and
compression. The original 8-bit color format was truly only eight colors, with 3 bits that
could be on or of : 0,0,0; 0,0,1; 0,1,1; 0,1,0; 1,0,0; 1,1,0; 1,0,1; 1,1,1. Eight bits per
channel space, meanwhile, has over 16 million color values. And the number of
possible color and brightness values in some 32-bit spaces boggles the mind. ■
PHDRI-2e_Book.indb 1 6/10/10 2:54 PMIntroduction2

I envisioned this shot as a silhouette from the very start. I even underexposed by a full stop
to ensure that the horse and fence would be complete shadows to emphasize the outline of
both in contrast to the late afternoon directional sun streaming across the green grass of the
paddocks. Here’s where the limitations of the imaging sensor were deliberately exploited for
an intended aesthetic ef ect. Even without detail in the animal, you recognize this as a horse
from its contour lines alone. Notice the shape of the histogram, indicating major shadow

Here I let much of the
image go to pure whites in
the high-key style. Your eye
is naturally drawn exactly
where I intend: to the three
blobs of paint at the mouths
of the tubes of watercolor
paint. Notice the shape of
the histogram, indicating
major highlight clipping.
PHDRI-2e_Book.indb 2 6/10/10 2:54 PMIntroduction 3

This is a middle shot from a burst to capture high dynamic range
source images. I tweaked the exposure a touch to give pure blacks
and pure whites in this frame. Notice how the two control points
correspond to white clipped and black clipped areas of the image.

When you click the arrows atop the histogram in Adobe Camera Raw,
it will show the white and black clipped pixels in your image. You can
choose one or the other, or both.
Got that so far? Any pure white (RGB: 255, 255, signif cant detail from these areas. Sure, you
255) or pure black (RGB: 0, 0, 0) in a standard can squeeze these away from the ends of the
low dynamic range (LDR) digital photograph ef- histogram, but all you’ll get are patches of detail
fectively contains no additional useful color or lacking gray instead. Messing around with
indiluminance information. vidual channels to get away from a true neutral
Whether your image is formatted as an 8-bit (equal RGB values: 255, 255, 255 or 162, 162,
per channel JPEG, a 14-bit-per-channel RAW, 162, for example) will build some color
informaor a 16-bit per channel TIFF, and no matter tion into the bald patches, but these are just
how much you push or pull the exposure in an paint chip samples. Let’s call a light greenish
image-editing program, you cannot retrieve any gray Saratoga Spring (RGB: 205, 233, 205), a
PHDRI-2e_Book.indb 3 6/10/10 2:54 PMIntroduction4

Here’s a perfectly gray square with equal values in all three
channels: 175, 175, 175. You can make darker or lighter grays
from 0, 0, 0 all the way up to 255, 255, 255 as long as you keep all
three channels equal.

By changing one of the color channels to a nonequal value (or all nonequal
values, for that matter), you get away from neutrals and into colors. But there’s
still no way to rebuild detail information when the captured pixels are beyond
the dynamic range of the sensor. You can forget about Havana Mauve and
Baltimore Blue, but we’ll be revisiting Saratoga Spring in a few pages.
mauvish gray Havana Sunset (226, 190, 190),
and a cool blue gray Baltimore Blue (180, 190,
190). Perhaps I don’t have a future career as a
paint color namer, but you get the point. And,
making unequal neutrals can introduce color
casting to an image—either unintentionally, or
for a purposeful addition of warmth or coolness
to the overall feel.
PHDRI-2e_Book.indb 4 6/10/10 2:55 PMIntroduction 5
 
Here is a pretty extreme example of exposure
range compression at capture. This image
was made during a blizzard in Yellowstone
National Park, Wyoming, USA, and the
deepest blacks and hottest whites all fall
within a couple of exposure values (EVs). In
fact, this image needs to be spread across
the entire histogram to give it normalized
contrast and pop. Notice the large number
of control points to make sure I don’t
accidentally introduce white or black clip
during my global tweaks in Adobe Camera
Raw. On a bright, sunny day, this image
would have had major clipping issues at
one end or the other, or potentially both,
depending on the exposure settings.
When the deepest shadows and hottest whites would call a proper exposure for the given
imhit the walls of the histogram in a traditional age—good detail throughout the tonal range
LDR image, there’s no detail information that with an emphasis on the primary subject of
can be retrieved whatsoever. With traditional interest without excessive blocked shadows and
single-image photography, dealing with black blown highlights distracting the aesthetic eye
and white clip can be handled to certain degrees away from the main thrust of the image. Sure, it
of success by adding some combination of f ll- might need some minor tweaking in the digital
f ash, graduated neutral density f lters, or wait- darkroom for optimization, but it doesn’t need
ing patiently for a certain time of day or weather major surgery before sharing.
pattern that manages to compress the overall But there are times, no matter how skilled
exposure range of the scene into luminance a photographer you may be, and regardless of
values that are smaller than the camera sensor’s what you’ve got in your bag of tricks, when
dynamic range. And this is what most people it is simply impossible to capture all of the
PHDRI-2e_Book.indb 5 6/10/10 2:55 PMIntroduction6
brightness values in a scene presented before squeezed straight from the tube, that’s a pretty
you in a single exposure. Here’s where high big set of paints, right of the bat. But the 32-bit
dynamic range imaging can come to the rescue. high dynamic range space adds the ability to
By taking a bracketed sequence of frames of the dilute each tube with water to make a lighter
scene, ranging from severe underexposure to tint or add an ultrablack neutral to make a
severe overexposure (based upon the median darker hue. And it can do this in f oating-point
meter reading), and merging these images into numbers—meaning each additional drop of
a single 32-bit per channel high dynamic range tint or hue from an eyedropper has its own
image, it is possible to cover a much broader ex- distinct value. Add one microliter of water to a
posure range because this 32-bit space is much gram of ultramarine paint (mix well!) and it’s
larger than 8- and 16-bit spaces and can there- got its own 32-bit f oating-point value. Add 1
fore hold much more information. T ink of it gram of ultramarine to a liter of water and it’s
this way: If 8-bit space is a page in this book and got a dif erent 32-bit f oating-point value. Add
16-bit space is a couple of chapters, 32-bit space equal parts ultramarine and “ultrablack” and
is an entire f oor-to-ceiling bookshelf. One it will have a dif erent 32-bit f oating-point
32-bit format is so robust that it can represent value, although all three of these will have the
a series of luminance values that exceeds the same low-bit RGB numbers, if that makes any
tonal range of the entire known visible universe sense. I’m talking a huge quantity of numbers
a couple of times over! Another HDR f le format representing colors and brightnesses here, if I
has a billion colors for each luminance level over haven’t yet beaten that point into the ground!
a 25 EV range, which is quite practical enough (And yes, from a very technical perspective,
here on this little earth. I’m talking huge, awe- paints are a subtractive color model, while the
inspiring color spaces here. HDR space is an additive color model, but it
If you think of every single low-bit RGB does get the point across, yes?)
combination as a tube of watercolor paint

The pigment in each of these three spots of ultramarine
watercolor paint is the same, but one is full strength, one is
slightly thinned, and one is dramatically thinned. In 32-bit
space, each of these three “colors” of ultramarine would
have its own f oating-point values but the same RGB values,
as shown on the following three screen shots of the color
“Saratoga Spring” show.
PHDRI-2e_Book.indb 6 6/10/10 2:55 PMIntroduction 7
The Little Lie about Low Dynamic
Range Images
Every single 8- and 16-bit image is built from
the same bits of the same relative brightness
and chromaticity. Slice and dice it as Lab color,
or RGB if you want, but there is a def nite upper
and lower limit. No pixel can have a value less
than 0, 0, 0 or greater than 255, 255, 255. It is
a closed set, and every traditionally captured
image represents its values with these 256 by 
3 channels values—and that’s where black and Here’s a paint chip of Saratoga Spring, jumped up into 32-bit space. Notice that the
RGB values are the same for all three but that there are dif erent 32-bit values. Notice white clip come from. T ere’s no -27, -27, -27
the position of the slider underneath the images. In the last one, I have shifted the for the “ultrablack” shadows deeper in that cave
exposure value preview brighter, although it is the identical color in all three images. or 314, 276, 320 for a f abberghastingly bright
Yes, 32-bit space is a lot to wrap your head around!
“infra-lavender” you discovered in a f eld in
Burgundy after several glasses of wine when the
sun was shining just so on that magical spring moment, a very good gray card proxy. It’s a piece
afternoon. Nope. Every color, shade, and tone of f agstone no matter the ambient lighting
must be no less than 0, 0, 0 and no more than conditions. And this f agstone lives in a ring
255, 255, 255 or they’re kicked of the histo- around a tree by a little decorative bench in the
gram for not coloring within the lines. photographs shown here, captured in bright,
In the low-bit world of digital imaging, midday sunlight and by crescent moonlight
all 18% gray in an image is the same 18% with stray dif use tungsten lights.
gray—never mind the real-world luminance T e f agstone is the f agstone, and it has the
values. T ink about a piece of f agstone for a same ref ectance characteristics (right around
PHDRI-2e_Book.indb 7 6/10/10 2:55 PMIntroduction8
18%) no matter how it is illuminated. But you’ll In both low-bit photographs under these
need dramatically dif erent exposures to capture dif erent lighting conditions, the brightest
the scene under these two lighting conditions. details in the scene that aren’t clipped will
In afternoon sunlight, I used an exposure of be in the 245 to 254 range, and the deepest
1/13 second @ f/16 at ISO 100 to yield a frame shadows that aren’t completely blocked will
with minimal clipping. After the sun went down, be in the 2 to 15 range for each RGB channel.
I needed an exposure of 9:24 @ f/4 at ISO 100 And this works reasonably well in representing
to make a normal exposure of the same scene the scene on a normal computer monitor and
(without any clipping—I had to spread the for print output. We expect the image to have
histogram a bit to normalize this low-contrast contrast and a pleasing tonal range spread
version of this scene). Shooting the nighttime across the histogram. Even if we look at the two
scene at 1/13 @ f/16 would yield a frame of photographs side by side, we accept the 0, 0,
nothing but blacks: 0, 0, 0. And shooting the 0 to 255, 255, 255 representation of the scene
bright sunny day scene at over nine minutes and perceive and posit dif erent light sources
would yield a completely burnt frame of nothing and dif erent times of day. We perceive the ink
but whites: 255, 255, 255. laydown through the contextual clues of the
imNonetheless, it is the same f agstone, and in age and combine this with an understanding of
each image you’ll want its RGB values to be right the lighting in the natural world, and we accept 
around 175, 175, 175, which is a Lab value of that one image is nighttime and one is daytime. This same scene requires
72%, 0, 0 to be a true neutral and a gray card But the maximum and minimum ink laydown very dif erent exposures
when shot in the afternoon proxy for balancing the exposure in post-pro- on a print, or RGB numbers on a monitor, are
and on a nearly moonless cessing. Any neutral brighter than the f agstone within the same small range of values in both
night. But in both cases, will have a greater lightness percentage: 84%, 0, cases. However, in the physical world, even the
I’m going to white-balance
0 or 92%, 0, 0, for example. Darker neutrals will darkest shadow values that aren’t clipped in the
of a section of f agstone
have a lower lightness percentage: 52%, 0, 0 and daylight photograph are actually much, much that stands in nicely for an
so on. Colors, of course, have nonequal values in brighter than the hottest nonclipped highlights 18% gray card. And in both
cases, all the color values at least one color channel in the RGB model and in the nighttime photo.
live within that 0, 0, 0 to in the “a” and “b” axes in the Lab model.
255, 255, 255 range.
PHDRI-2e_Book.indb 8 6/10/10 2:55 PMIntroduction 9

When I take these shots and
merge them into a single high
dynamic range image, the
histogram in Photomatix Pro
looks quite dif erent. Notice that
the estimated dynamic range is
more than 40,000:1!
T e same base set of numbers representing small set of monitor colors and an even smaller
dif erent color and tonal values works in both set of ink colors to represent so many dif erent
these cases, and in all low dynamic range im- lighting conditions and brightness values. From
ages, be they 8-bit per channel JPEGs, 16-bit a technical perspective, these monitors and the
per channel TIFFs from 14-bit per channel printers (and the inks, and by extension, the
RAW f les, or any of the other standard 8- and prints) are called “device-referred”. We know
16-bit f le formats. Based upon our perception there is no set of inks that can truly display all
and understanding, we accept this small set of the colors and brightness levels in the world,
numbers to represent the dif ering luminosity but we can conveniently crunch them into a set
values of pretty much all images we display on that does a reasonable job of representing much
our computer monitors and as photographic of the world.
prints. It’s a working fallacy that we accept a
PHDRI-2e_Book.indb 9 6/10/10 2:55 PMIntroduction10
But in the real world, that same scene of the create a single image that can hold all of this
bench under nighttime or daytime conditions color and luminance information—no more
has distinctively dif erent overall luminosities. blown highlights or blocked shadows. Yippee!
And it’s not just the same scene under dif erent True high dynamic range images live in a much
lighting situations that can have dramatically bigger space than the single-shot 8- and 16-bit
dif erent luminance values. T ere is many a time low dynamic range images you are familiar with.
when the range of light and shadow through the T e 32 bits per channel HDR format allows
viewf nder simply exceeds the dynamic range for such a ridiculously huge set of numbers
of even today’s top-of-the-line digital cameras. representing dif erent color and luminance
And that’s where high dynamic range imaging values that it can preserve a truer representation
comes into play. By making a series of images of the real-world relative brightness over a much
across a range of exposures to capture the entire larger range—it can describe candlelight and the
luminance range of a scene, we can use the blazing desert sun in the same image without
magic of insanely complicated mathematics to clipping, for example. Brief y revisiting our bench
and f agstone, these two images live in very
dif erent places on the HDR brightness scale and
histogram—which makes sense, since these were
captured at very dif erent exposures, and the
whole point of HDR imaging is to better
represent the real-world lighting conditions. T ere are
a number of dif erent 32-bit formats, with a ton
of crazy math employed to make each one work,
but from a practical photographic perspective,

Both of these screen shots from Adobe Photoshop CS3 show
the same high dynamic range image at dif erent exposure value
levels. You can see the seven low dynamic range source images
in the left column. The Set White Point Preview slider in the right
column sets the exposure level for the high dynamic range image.
Anything that pushes past 0, 0, 0 or 255, 255, 255 based on the
white point preview level goes to black or white on your monitor,
but all that information is held in the HDR image. By adjusting the
white point preview under the histogram, we can reveal detail at
various luminance levels.
PHDRI-2e_Book.indb 10 6/10/10 2:55 PMIntroduction 11
you need to understand this as much as you need Of course, not many of us have the exorbitantly
to understand how the Healing Brush in Adobe expensive monitors necessary to fully appreciate
Photoshop is like a thermal transference fun-c a high dynamic range image in all its luminous
tion—in other words, pretty much not at all. glory. So why bother if we don’t have monitors
For HDR images, the two formats we are capable of displaying HDR images or printers
most interested in from a practical perspective capable of reproducing them?
are OpenEXR (.exr) and Radiance (.hdr) for
cross-program compatibility. Tese formats
Crunching All the Brightness into a are not truly infnite, but there is not much of a
color or an exposure value in the universe that Much Smaller Space
can’t be held within one of them. I will revisit
this later, but for now, just take away that both Normal computer monitors cannot display the
of these formats have some really cool street full range of exposure of a true high dynamic
cred—Radiance was invented by Greg Ward, range image, nor can our printers output the
HDR pioneer par excellence, and OpenEXR is an full luminosity range of an HDR image, so we
open-source format courtesy of Industrial Light must crunch that 32-bit photograph back down
& Magic—you know, the folks who madeS tar into smaller bit space. Tere are a number of
Wars! We’ll also touch on HDR PhotoStudio’s ways this is accomplished, and I will use the
new BEF format and PSDs for saving layered umbrella term tone mapping in this book when
HDR images in Adobe Photoshop. speaking generically about dropping bit depth
From a technical perspective, high dynamic from 32 to either 8 or 16 bits per channel. (I’ll
range images are “scene-referred”, as their lumi- also use crunch, compress, and a couple of other
nance values are not output device dependent. terms to break up the monotony.)

This tone-mapped image
resulting from seven source
fles preserves much more
of the overall luminosity of
the scene than any single
captured image. After being
merged to a 32-bit high
dynamic range image, it was
converted back to 16-bit
space using the Compressor
Tonemapper tool in the
FDRTools Advanced
software program.
PHDRI-2e_Book.indb 11 6/10/10 2:55 PM