Professional Documents
Culture Documents
Erik Reinhard Abstract — With interest in high-dynamic-range imaging mounting, techniques for displaying such
Timo Kunkel images on conventional display devices are gaining in importance. Conversely, high-dynamic-range
Yoann Marion display hardware is creating the need for display algorithms that prepare images for such displays. In
Jonathan Brouillat this paper, the current state of the art in dynamic-range reduction and expansion is reviewed, and in
particular the theoretical and practical need to structure tone reproduction as a combination of a
Rémi Cozot
forward and a reverse pass is passed.
Kadi Bouatouch
Keywords — High-dynamic-range imaging reduction, expansion, tone mapping, inverse tone
mapping.
Extended revised version of a paper presented at the SID Symposium, Seminar & Exhibition (SID ‘07) held in Long Beach, California, May 20–25, 2007.
E. Reinhard is with the University of Bristol, Department of Computer Science, Merchant Venturers Bodg., Woodland Rd., Bristol N/A BS8 1UB,
U.K.; telephone +44-117-954-5140, fax –5208, e-mail: reinhard@cs.ucf.edu. He is also with IRISA, Campus de Beaulieu, France, and the University
of Central Florida, Orlando, FL.
T. Kunkel is with the University of Bristol, U.K.
Y. Marion, J. Brouillat, R. Cozot, and K. Bouatouch are with IRISA, Campus de Beaulieu, Rennes, France.
© Copyright 2007 Society for Information Display 1071-0922/07/1512-0997$1.00
which is the same for each pixel.7–13 The term global stems
bottom image in Fig. 2 is tone-mapped non-linearly showing
from the fact that many such functions need to be anchored
detail in both the light and dark regions.
to some values that are computed by analyzing the full
In general, linear scaling will not be appropriate for
image. In practice, most operators use the geometric aver-
tone reproduction. The key issue in tone reproduction is
age Lv over the N image pixels to steer the compression:
then to compress an image while at the same time preserving
one or more attributes of the image. Different tone-repro- F 1 log d + L (x, y) I .
duction algorithms focus on different attributes such as con- Lv = exp GH N xÂ,y b v gJK (1)
trast, visible detail, brightness, or appearance. Ideally,
displaying a tone-mapped image on a low-dynamic-range The small constant δ is introduced to prevent the average
display device would recreate the same visual response in to become zero in the presence of black pixels. The lumi-
the observer as the original scene. Given the limitations of nance of each pixel is indicated with Lv, which can be com-
display devices, this is, in general, not achievable, although puted from RGB values if the color space is known. If the
we may approximate this goal as closely as possible. color space in which the image is specified is unknown, then
the second best alternative would be to assume that the
image uses sRGB primaries and white point, so that the
3 Spatial operators luminance of a pixel is given by
In the following sections we discuss tone-reproduction Lv (x, y) = 0.2125R(x, y) + 07154
. G(x, y) + 0.0721B(x, y).
operators which apply compression directly on pixels. Often
global and local operators are distinguished. Tone-repro- The geometric average is normally mapped to a pre-
duction operators in the former class change each pixel’s defined display value. The main challenge faced in the design
luminance values according to a compressive function of a global operator lies in the choice of compressive func-
998 Reinhard et al. / Image display algorithms for high- and low-dynamic-range display devices
different than the brightness of a pixel in a dark neighbor-
hood. Design challenges for local operators involve choos-
in g th e co mpressive function, the size of the local
neighborhood for each pixel, and the manner in which local
pixel values are used. In general, local operators are able to
achieve better compression than global operators (Fig. 3),
albeit at a higher computational cost.
Both global and local operators are often inspired by
the human visual system. Most operators employ one of two
distinct compressive functions, which is orthogonal to the
distinction between local and global operators. Display val-
ues Ld(x, y) are most commonly derived from image lumi-
nances Lv(x, y) by the following two functional forms:
FIGURE 3 — A local tone-reproduction operator (left) and a global
tone-reproduction operator (right).14 The local operator shows more Lv (x, y)
detail, as, for instance, seen in the insets. Ld (x, y) = , (2)
f (x, y)
tion. Many functions are possible, which are for instance Lnv (x, y)
Ld ( x, y) = n . (3)
based on the image’s histogram 12 (Section 4.2) or on data Lv (x, y) + g n (x, y)
gathered from psychophysics (Section 4.3).
On the other hand, local operators compress each In these equations, f(x, y) and g(x, y) may either be
pixel according to a specific compression function which is constant or a function which varies per pixel. In the former
modulated by information derived from a selection of neigh- case, we have a global operator, whereas a spatially varying
boring pixels, rather than the full image.14–25 The rationale function results in a local operator. The exponent n is a con-
is that the brightness of a pixel in a light neighborhood is stant which is either fixed or set differently per image.
FIGURE 4 — Halos are artifacts commonly associated with local tone-reproduction operators. Chiu’s operator is used here without
smoothing iterations to demonstrate the effect of division (left).
1000 Reinhard et al. / Image display algorithms for high- and low-dynamic-range display devices
FIGURE 5 — Tone-mapping function created by reshaping the
cumulative histogram of the image shown in Fig. 6.
1002 Reinhard et al. / Image display algorithms for high- and low-dynamic-range display devices
with respect to the pixel under consideration. It is therefore
important that the local average is computed over pixel val-
ues that are not significantly different from the pixel that is
being filtered.
This suggests a strategy whereby an image is filtered
such that no blurring over such edges occurs. A simple but
computationally expensive way is to compute a stack of
Gaussian blurred images with different kernel sizes, i.e., an
image pyramid. For each pixel, we may choose the largest
Gaussian which does not overlap with a significant gradi-
ent. The scale at which this happens can be computed as
follows.
In a relatively uniform neighborhood, the value of a
Gaussian blurred pixel should be the same regardless of the
filter kernel size. Thus, in this case, the difference between
a pixel filtered with two different Gaussians should be
around zero. This difference will only change significantly if
the wider filter kernel overlaps with a neighborhood con-
taining a sharp contrast step, whereas the smaller filter ker-
nel does not. A difference of Gaussians (DoG) signal
LiDoG(x, y) at scale i can be computed as follows:
LDoG
i ( x, y) = Ris (x, y) - R2is (x, y). (8)
LDoG DoG
s,i ( x, y) = Li (x, y) G(x, y). (13)
1004 Reinhard et al. / Image display algorithms for high- and low-dynamic-range display devices
FIGURE 11 — Tone reproduction using a sub-band architecture, compu-
ted here using a Haar filter.36
b g
—L = Lv (x + 1, y) - Lv (x, y), Lv (x, y + 1) - Lv (x, y) . (17)
FIGURE 13 — An image tone-mapped using bilateral filtering. The base
and detail layers shown in Fig. 12 are recombined after compressing the Here, ∇L is a vector-valued gradient field. By attenu-
base layer. ating large gradients more than small gradients, a tone
reproduction operator may be constructed.41 Afterwards,
in a spatially varying or local tone reproduction operator. A an image can be reconstructed by integrating the gradient
local operator based on sigmoidal compression can for field to form a tone-mapped image. Such integration must
instance be created by substituting Lblur(x, y) = LB(x, y) in be approximated by numerical techniques, which is achieved
Eq. (10). by solving a Poisson equation using the Full Multi-Grid
Alternatively, the semi-saturation constant g(x, y) in Method.42 The resulting image then needs to be linearly
Eq. (3) may be seen as a local adaptation constant and can scaled to fit the range of the target display device. An exam-
therefore be locally approximated with LB(x, y),40 or with ple of gradient domain compression is shown in Fig. 14.
any of the other filters mentioned above. As shown in Fig. 7, This form of dynamic-range reduction is especially useful in
the choice of semi-saturation constant shifts the curve hori- applications where detail visibility is of importance.
FIGURE 14 — The image on the left is tone-mapped using gradient domain compression. The magnitude of the gradients ||∇Lw|| is
mapped to a gray scale in the right image (white is a gradient of 0; black is the maximum gradient in the image).
1006 Reinhard et al. / Image display algorithms for high- and low-dynamic-range display devices
FIGURE 15 — Per-channel gamma correction may desaturate the image. The images were desaturated with values of s = 0.2, s = 0.5,
and s = 0.8.
s
computational model of lightness perception, which can be
(x, y) = G
F I (x, y) I
B
extended for the purpose of tone reproduction.
One of the strengths of using a computational model
I R,d
H L (x, y)JK
v
Ld (x, y).
of lightness perception for the purpose of tone reproduc- Alternatively, the saturation constant s may be chosen
tion, is that conventionally difficult phenomena such as the to be smaller than 1. Such per-channel gamma correction
1008 Reinhard et al. / Image display algorithms for high- and low-dynamic-range display devices
FIGURE 16 — Example of clamping. Both images were tone-mapped using photographic tone reproduction. The left image is not
clamped, whereas 7% of the darkest and 2% of the lightest pixels were clamped in the right image.
Nonetheless, there are other issues related to dynamic- which is consistent with the finding that, at least for rela-
range expansion, which will have to be taken into account. tively short exposure times, humans prefer to view linear
These do not relate to the expansion of values per se, but are up-scaled image over non-linearly scaled images.51 The
related to artifacts in the source material that may be ampli- appearance of artifacts is minimized by application of noise
fied to become more visible. For instance, by expanding the filtering and quantization reduction through the use of a
luminance range of an image, the lossy compression applied bilateral filter. Pixel encodings of 235 or higher in video formats
to JPEG images may become visible. Second, the non-linear are normally assumed to indicate light sources or highlights.
encoding of pixel values may, after expansion, lead to visible Those pixels can be enhanced separately. Alternatively,
quantization artifacts. Finally, under- and over-exposed areas highlights could be detected with a dedicated algorithm
may require separate processing to salvage a lack of data in before being scaled separately from the remainder of the
these regions.48 image.52
Some solutions for these problems have been pro- In summary, the main problems associated with dis-
posed. For instance, Banterle et al. invert the photographic playing conventional images on HDR display devices revolve
operator for dynamic-range expansion.49 As this effectively around avoiding the visibility of artifacts in the capture and
results in an inverse sigmoid, this makes the implicit but encoding of legacy content. So far, simple functions have
erroneous assumption that the input image is given in units proved adequate for dynamic-range expansion, although we
which correspond to either photo-receptor output or some would not draw the conclusion that this would be the case
perceptual quantity. Blocky artifacts, for instance, those for all images and all display conditions.
arising form JPEG encoding, are avoided by determining
pixels belonging to light sources in the input image and apply-
ing a different interpolation scheme for those pixels.
Rempel et al. have found that linear up-scaling can 8 Tone reproduction and inverse tone
typically be performed up until a contrast of 5000:1 before reproduction
the image takes on an unnatural appearance.50 Their solu- Many tone-reproduction operators are modeled after some
tion is therefore to rely predominantly on linear up-scaling, aspects of human vision. The computed display values
1010 Reinhard et al. / Image display algorithms for high- and low-dynamic-range display devices
FIGURE 19 — Photographic operator with key k = 0.3 (top) and k = 0.08
FIGURE 18 — Forward model only with n = 0.7 and k = 0.3 (top) and (bottom). By using the photographic operator, we have effectively
n = 0.7 and k = 0.08 (bottom). changed the exponent to n = 1.
1012 Reinhard et al. / Image display algorithms for high- and low-dynamic-range display devices
26 J M DiCarlo and B A Wandell, “Rendering high dynamic range im-
References ages,” Proc SPIE Electronic Imaging 2000 Conference 3965, 392–401
1 L Spillmann and J S Werner, eds., Visual Perception: The Neurological (2000).
Foundations (Academic Press, San Diego, 1990). 27 J Duan, G Qiu, and M Chen, “Comprehensive fast tone mapping for
2 J A Ferwerda, “Elements of early vision for computer graphics,” IEEE high dynamic range image visualization,” Proc Pacific Graphics (2005).
Computer Graphics Applications 21(5), 22–33 (2001). 28 K I Naka and W A H Rushton, “S-potentials from luminosity units in
3 H Seetzen, L A Whitehead, and G Ward, “A high dynamic range display the retina of fish (cyprinidae),” J Physiology (London) 185(3), 587–599
using low and high resolution modulators,” SID Symposium Digest (1966).
Tech Papers 34, 1450–1453 (2003). 29 J Kleinschmidt and J E Dowling, “Intracellular recordings from gecko
4 H Seetzen, W Heidrich, W Stuerzlinger, G Ward, L Whitehead, M photoreceptors during light and dark adaptation,” J General Physiology
Trentacoste, A Ghosh, and A Vorozcovs, “High dynamic range display 66(5), 617–648 (1975).
systems,” ACM Trans Graphics 23(3), 760–768 (2004). 30 D C Hood, M A Finkelstein, and E Buckingham, “Psychophysical tests
5 E Reinhard, G Ward, S Pattanaik, and P Debevec, High Dynamic of models of the response function,” Vision Research 19(4), 401–406
Range Imaging: Acquisition, Display and Image-Based Lighting (Mor- (1979).
gan Kaufmann Publishers, San Francisco, 2005). 31 D C Hood and M A Finkelstein, “Comparison of changes in sensitivity
6 E Reinhard, “Overview of dynamic range reduction,” SID Symposium and sensation: Implications for the response-intensity function of the
Digest Tech Papers 38, 1049–1052 (2007). human photopic system,” J Experimental Psychol: Human Perceptual
7 N J Miller, P Y Ngai, and D D Miller, “The application of computer Performance 5(3), 391–405 (1979).
graphics in lighting design,” J IES, 14, 6–26 (1984). 32 E Reinhard, “Parameter estimation for photographic tone reproduc-
8 J Tumblin and H Rushmeier, “Tone reproduction for computer gener- tion,” J Graphics Tools 7(1), 45–51 (2003).
ated images,” IEEE Computer Graphics Applications 13(6), 42–48 33 E Reinhard and K Devlin, “Dynamic range reduction inspired by
(1993). photoreceptor physiology,” IEEE Trans Visualization Computer
9 G Ward, “A contrast-based scalefactor for luminance display,” Paul Graphics 11(1), 13–24 (2005).
Heckbert, ed., Graphics Gems IV (Academic Press, Boston, 1994), pp. 34 J Tumblin and G Turk, “LCIS: A boundary hierarchy for detail-preserv-
415–421. ing contrast reduction,” Alyn Rockwood, ed., SIGGRAPH 1999, Com-
10 J A Ferwerda, S Pattanaik, P Shirley, and D P Greenberg, “A model of puter Graphics Proceedings, 83–90 (1999).
visual adaptation for realistic image synthesis,” SIGGRAPH Conf Proc,
35 D Comaniciu and P Meer, “Mean shift: a robust approach toward
249–258 (1996).
feature space analysis,” IEEE Trans Pattern Analysis Machine Intelli-
11 F Drago, K Myszkowski, T Annen, and N Chiba, “Adaptive logarithmic
gence 24(5), 603–619 (2002).
mapping for displaying high contrast scenes,” Computer Graphics
Forum 22(3) (2003). 36 Y Li, L Sharan, and E H Adelson, “Compressing and companding high
dynamic range images with subband architectures,” ACM Trans
12 G Ward, H Rushmeier, and C Piatko, “A visibility matching tone
Graphics 24(3), 836–844 (2005).
reproduction operator for high dynamic range scenes,” IEEE Trans
Visualization Computer Graphics 3(4) (1997). 37 C Tomasi and R Manduchi, “Bilateral filtering for gray and color
13 C Schlick, “Quantization techniques for the visualization of high dynamic images,” Proc IEEE International Conference Computer Vision,
range pictures,” Peter Shirley, Georgios Sakas, and Stefan Müller, eds., 836–846 (1998).
Photorealistic Rendering Techniques (Springer-Verlag, Berlin, Ger- 38 S Paris and F Durand, “A fast approximation of the bilateral filter using
many, 1994), pp. 7–20. a signal processing approach,” European Conference on Computer
14 E Reinhard, M Stark, P Shirley, and J Ferwerda, “Photographic tone Vision (2006).
reproduction for digital images,” ACM Trans Graphics 21(3), 267–276 39 B Weiss, “Fast median and bilateral filtering,” ACM Trans Graphics
(2002). 25(3) (2006).
15 K Chiu, M Herf, P Shirley, S Swamy, C Wang, and K Zimmerman, 40 P Ledda, L-P Santos, and A Chalmers, “A local model of eye adaptation
“Spatially nonuniform scaling functions for high contrast images,” Proc for high dynamic range images,” Proc ACM Afrigraph, 151–160 (2004).
Graphics Interface ‘93, 245–253 (1993). 41 R Fattal, D Lischinski, and M Werman, “Gradient domain high dy-
16 D J Jobson, Z Rahman, and G A Woodell, “Retinex image processing: namic range compression,” ACM Trans Graphics 21(3), 249–256
Improved fidelity to direct visual observation,” Proc IS&T 4th Color (2002).
Imaging Conference: Color Science, Systems, and Applications 4, 42 W H Press, S A Teukolsky, W T Vetterling, and B P Flannery, Numerical
124–125 (1995). Recipes in C: The Art of Scientific Computing, 2nd edn. (Cambridge
17 Z Rahman, D J Jobson, and G A Woodell, “A multiscale retinex for University Press, 1992).
color rendition and dynamic range compression,” SPIE Proc: Applica- 43 G Krawczyk, R Mantiuk, K Myszkowski, and H-P Seidel, “Lightness
tions of Digital Image Processing XIX, 2847 (1996). perception inspired tone mapping,” Proc First ACM Symposium on
18 Z Rahman, G A Woodell, and D J Jobson, “A comparison of the Applied Perception in Graphics and Visualization, 172 (2004).
multiscale retinex with other image enhancement techniques,” IS&T’s 44 G Krawczyk, K Myszkowski, and H-P Seidel, “Lightness perception in
50th Annual Conference: A Celebration of All Imaging 50, 426–431 tone reproduction for high dynamic range images,” Computer Graph-
(1997). ics Forum 24(3), 635–645 (2005).
19 S N Pattanaik, J A Ferwerda, M D Fairchild, and D P Greenberg, “A
45 G Krawczyk, K Myszkowski, and H-P Seidel, “Computational model of
multiscale model of adaptation and spatial vision for realistic image
lightness perception in high dynamic range imaging,” Proc IS&T/SPIE
display,” SIGGRAPH ‘98 Conf Proc, 287–298 (1998).
Human Vision and Electronic Imaging 6057 (2006).
20 M D Fairchild and G M Johnson, “Meet iCAM: An image color
appearance model,” IS&T/SID 10th Color Imaging Conference, 33–38 46 M D Fairchild and G M Johnson, “The iCAM framework for image
(2002). appearance, image differences, and image quality,” J Electronic Imag-
21 M Ashikhmin, “A tone mapping algorithm for high contrast images,” ing (2004).
Proc 13th Eurographics Workshop on Rendering, 145–155 (2002). 47 N Moroney and I Tastl, “A comparison of retinex and iCAM for scene
22 S N Pattanaik and H Yee, “Adaptive gain control for high dynamic rendering,” J Electronic Imaging 13(1) (2004).
range image display,” Proc Spring Conf in Computer Graphics, 24–27 48 L Wang, L-Y Wei, K Zhou, B Guo, and H-Y Shum, “High dynamic
(2002). range image hallucination,” Eurographics Symposium on Rendering
23 A V Oppenheim, R Schafer, and T Stockham, “Nonlinear filtering of (2007).
multiplied and convolved signals,” Proc IEEE 56(8), 1264–1291 49 F Banterle, P Ledda, K Debattista, and A Chalmers, “Inverse tone
(1968). mapping,” GRAPHITE ‘06: Proc 4th Intl Conf on Computer Graphics
24 F Durand and J Dorsey, “Fast bilateral filtering for the display of and Interactive Techniques in Australasia and Southeast Asia, 349–356
high-dynamic-range images,” ACM Trans Graphics 21(3), 257–266 (2006).
(2002). 50 A Rempel, M Trentacoste, H Seetzen, D Young, W Heidrich, L White-
25 P Choudhury and J Tumblin, “The trilateral filter for high contrast head, and G Ward, “On-the-fly reverse tone mapping of legacy video
images and meshes,” Proc Eurographics Symposium on Rendering, and photographics,” ACM Trans Graphics (Proc SIGGRAPH 2007)
186–196 (2003). 26(3) (2007).
1014 Reinhard et al. / Image display algorithms for high- and low-dynamic-range display devices