Professional Documents
Culture Documents
HDRContrastCTydtgat Et Al
HDRContrastCTydtgat Et Al
HDRContrastCTydtgat Et Al
net/publication/303556353
CITATION READS
1 201
4 authors, including:
Goran Stojmenovik
Barco
24 PUBLICATIONS 129 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Claude Tydtgat on 03 March 2018.
Claude Tydtgat
Barco NV.
Dirk Maes
Barco NV.
Goran Stojmenovik
Barco NV.
Augustin Grillet
Barco NV.
Abstract.
There is a growing interest in High Dynamic Range Imaging for cinema projection applications. It has
been reported that the observable cinema contrast ratio is not only determined by the native
projector contrast, but even more by various parameters such as the projection lens, port window,
screen, theatre setting and audiences. We have derived a mathematical model to characterize the
influence of these parameters and to assess their relative importance as a function of the projected
image content. Further a fast measurement method was developed to determine the various
parameters across different theatres, providing a good match with our model.
Keywords.
The authors are solely responsible for the content of this technical presentation. The technical presentation does not necessarily reflect the
official position of the Society of Motion Picture and Television Engineers (SMPTE), and its printing and distribution does not constitute an
endorsement of views which may be expressed. This technical presentation is subject to a formal peer-review process by the SMPTE
Board of Editors, upon completion of the conference. Citation of this work should state that it is a SMPTE meeting paper. EXAMPLE:
Author's Last Name, Initials. 2011. Title of Presentation, Meeting name and location.: SMPTE. For information about securing permission
to reprint or reproduce a technical presentation, please contact SMPTE at jwelch@smpte.org or 914-761-1100 (3 Barker Ave., White
Plains, NY 10601).
Mathematical model
A mathematical model to predict theatre contrast is done in two steps:
1. modeling projector contrast
2. modeling theatre reflectivity
1. The first step involves modeling the ‘projector contrast’ which depends on the imaging device
contrast, imaging optics, dust inside the projector, dust on the projection lens… but we take also
scattering due to the port window into account. The general effect is that starting from an
imager, some light will be scattered all over the projected image reducing the projector contrast.
The combination of Equation eqs.1 and 2 can be approximated to the more familiar typical
‘parallel resistor’ form for values of ‘‘ and ‘s’ <<1 (as shown in the Appendix):
1 1
= + 𝛼(𝑠 + 𝜌)
𝐶𝑅𝑡ℎ𝑒𝑎𝑡𝑟𝑒 𝐶𝑅𝑠𝑒𝑞
eq.4
It is important to notice that in this approximate equation (eq.4), the scatter fraction ‘s’ and the
theatre reflectivity ‘’ simply add, this indeed shows why ‘s’ and ‘’ are sensitive to
measurement uncertainties but ‘s + ’ not.
In Table 1 a summary of the parameters introduced in the model:
Parameter Definition
% of white content of the projected image
(image brightness)
s Projector + port window scattering
coefficient
Theater reflectivity coefficient
Table 1. Parameters of the mathematical model
Measurement method
In order to verify the mathematical model and obtain data for a variety of theatres worldwide we
have developed a fast measurement method.
In order to simulate the varying white image content , we developed a series of test patterns
inspired by the ANSI test pattern with white squares arranged in a 4x4 grid but with a varying
© 2015 Society of Motion Picture & Television Engineers® (SMPTE®) 4
size of the white squares, corresponding to a white content fraction α ranging from 1 to 50%.
The distribution of the white content in a 4x4 grid helps to achieve a uniform illumination of the
auditorium, while the size of the black and white squares is sufficiently large to be able to
perform reliable measurements in the center. In Figure 1, the test pattern for a 15% white
content fraction is illustrated.
Figure 2. Left image with 25% white fraction captured with long camera exposure time of 13 s,
right image with 25% white fraction captured with short camera exposure time of 1/15 s.
First, some preprocessing of the images is needed to remove the noise floor added by the
camera. Therefor dark images are taken with the lens covered by the lens cap, one using the
short and another using the long exposure time. These offset darks are then subtracted from all
corresponding images.
© 2015 Society of Motion Picture & Television Engineers® (SMPTE®) 5
It goes without saying that the camera optics by itself needs to be as clean as possible to avoid
added veiling glare. Keep the lens and other optics clean and avoid fingerprints. A cross check
of the derived contrast can be quickly performed using an LMT meter. For instance, the contrast
of the 50% ANSI pattern can be measured using an LMT and cross checked with the camera
image derived contrast values.
The images were sampled at the center of the different squares (cfr. measurement locations of
the conventional ANSI pattern) and the average contrast ratio was calculated taking into
account the ratio between the black level exposure time and the white level exposure time.
An occlusion contrast for the 50% ANSI pattern is also measured by obscuring some dark parts
of blacks in the pattern. In this case the light incident on the screen from the projector is
blocked and only light reflected by the auditorium is measured [4]. An example of such a
measurement is shown at the bottom left in Figure 3.
Figure 3. Example occlusion contrast measurement clearly showing that occluded parts are
darker than the non-occluded darks.
We were able to find the reflectivity and scattering of several theatres, a representative selection
of them is shown in the table below.
Theatre s s+
1 0.015 0.019 0.034
2 0.033 0.025 0.058
3 0.010 0.025 0.035
4 0.005 0.011 0.016
5 0.018 0.022 0.040
6 0.009 0.018 0.027
7 0.008 0.021 0.029
8 0.010 0.013 0.023
Figure 5. Theatre contrast calculations using real world parameters (theatre reflectivity and
scatter) and statistics of several movies for a sequential projector contrast of 2000:1. Blue
curves show the theatre contrast.
From the graphs one can notice that in most of the area of interest, and especially with a
median brightness about 5%, the achievable theatre contrast ranges between 500:1 and 700:1
even with a very high contrast projector.
Let’s now focus on dark images. As a figure of merit we have chosen the line of 1.5% white
content, below which we estimate only about 10% of the movie content is found. For an average
theatre the s+ factor is 0.028.The achievable effective theatre contrast for a 2000:1 projector
contrast is around 1150:1. For the 1M:1 projector, the achievable theatre contrast for the same
scattering and reflectivity value is around 2400:1, (not shown in the picture, but can easily be
calculated with Eq.4). Thus increasing the projector contrast by a factor of 500:1 effectively
increases the effective theatre contrast only by about of factor 2. Even further, due to the
diminishing returns in this case, the largest improvements in contrast happen with the first steps
of increasing the projector contrast above 2000:1.
So we can draw three conclusions:
1. Auditorium reflectivity and scattering largely influence the effective theatre contrast for all
image content, especially for median image brightnesses for which the projector choice
makes little difference (700:1 for 1M:1 projector vs. 500:1 for a standard DLP projector).
2. For average auditoriums and pretty dark image content (1.5%, defining a 10:90 split of
the frame count), a 1M:1 projector only offers a twofold improvement of image contrast.
Most of the contrast improvements happens with relatively lower projector contrast
improvements due to the hyperbolic nature of the curve.
Conclusion
In this paper we derived a mathematical model for the determination of effective theatre contrast
as function of the projector sequential contrast, projector optical and port window scattering
effects as well as variables anteater reflectivity. In approximation for small scattering and
reflectivity values, this model is compatible with earlier models for theatre contrast [3]. We
proved that, using a new fast measurement method, the model fits the measurements perfectly.
By testing our models on a number of real theatres we were able to get a realistic range for the
parameters influencing in-theatre contrast. From these parameters we can predict the
achievable contrast for different white image content values found in most feature film
presentations while also accommodating for different projector contrasts modeled in the same
auditorium.
The results indicate that in an average cinema room, the contrast can still be improved for dark
images, either by improving methods for the reduction of auditorium reflectivity+scattering, or a
higher projector contrast, or both. However this improvement soon enters a law of diminishing
returns as the projector contrast increases beyond a point where the room variables impact the
audiences ability to view any meaningful improvement. Next to projector brightness, projector
contrast has a large impact on system cost, be it for throwing away light, requiring specialized
optics, extensive modifications to the viewing environment, etc. There are limits to how far
projector technology can be pushed to provide a visible improvement in contrast levels, while
still maintaining a good economic value proposition for the end user.
The projector contrast is the contrast measured in a perfect theatre (a ‘black hole’) and is the
ratio of E’max to E’min:
𝐸𝑚𝑎𝑥
(1 − 𝑠 + 𝑠𝛼)𝐸𝑚𝑎𝑥 + 𝑠(1 − 𝛼)𝐸𝑚𝑖𝑛 (1 − 𝑠 + 𝑠𝛼) 𝐸𝑚𝑖𝑛 + 𝑠(1 − 𝛼)
𝐶𝑅𝑝𝑟𝑜𝑗 = =
𝑠𝛼𝐸𝑚𝑎𝑥 + (1 − 𝑠𝛼)𝐸𝑚𝑖𝑛 𝐸
𝑠𝛼 𝐸𝑚𝑎𝑥 + (1 − 𝑠𝛼)
𝑚𝑖𝑛
Now, the ratio Emax / Emin is nothing but the sequential contrast:
𝐸 ′ 𝑚𝑎𝑥 (𝛼 = 1) 𝐸𝑚𝑎𝑥
𝐶𝑅𝑠𝑒𝑞 = =
𝐸 ′ 𝑚𝑖𝑛 (𝛼 = 0) 𝐸𝑚𝑖𝑛
So we get a first important equation for the projector contrast:
(1 − 𝑠 + 𝑠𝛼)𝐶𝑅𝑠𝑒𝑞 + 𝑠(1 − 𝛼)
𝐶𝑅𝑝𝑟𝑜𝑗 =
𝑠𝛼𝐶𝑅𝑠𝑒𝑞 + (1 − 𝑠𝛼)
eq.1
When the area fraction is 0.5 we find the ANSI contrast.
We now place this projector in a real theatre, not a ‘black hole’. Light reflected from the
projection screen will be partially absorbed and partially reflected from the walls, the ceiling,
floor, seats, people … in all directions. The net effect is that a portion of this back-reflected light
will fall back upon the screen. However for this back-reflected light the image information is
completely destroyed and we assume that this back reflected light is uniformly spread upon the
front screen to simplify the problem.
© 2015 Society of Motion Picture & Television Engineers® (SMPTE®) 11
In principle this back reflected light will again reflect from the front screen and have a second
step to be back reflected again… In the end this means that the total back reflected light will be
proportional to the total projected light. This proportionality constant is called the theatre
reflectivity .
The total back reflected luminous flux is then:
𝐿𝑏𝑎𝑐𝑘_𝑟𝑒𝑓𝑙𝑒𝑐𝑡𝑒𝑑 = 𝜌(𝐴𝑚𝑎𝑥 𝐸′𝑚𝑎𝑥 + 𝐴𝑚𝑖𝑛 𝐸′𝑚𝑖𝑛 )
This luminous flux is uniformly spread over the total area, so an offset should be added to the
perceived illuminance:
𝐴𝑚𝑎𝑥 𝐸′𝑚𝑎𝑥 + 𝐴𝑚𝑖𝑛 𝐸′𝑚𝑖𝑛
𝐸𝑜𝑓𝑓𝑠𝑒𝑡 = 𝜌 = 𝜌[𝛼𝐸′𝑚𝑎𝑥 + (1 − 𝛼)𝐸′𝑚𝑖𝑛 ]
𝐴𝑚𝑎𝑥 + 𝐴𝑚𝑎𝑥
And finally the theatre contrast will be:
𝐸 ′ 𝑚𝑎𝑥 + 𝐸𝑜𝑓𝑓𝑠𝑒𝑡 (1 + 𝛼𝜌)𝐸′𝑚𝑎𝑥 + (1 − 𝛼)𝜌𝐸′𝑚𝑖𝑛
𝐶𝑅𝑡ℎ𝑒𝑎𝑡𝑟𝑒 = =
𝐸 ′ 𝑚𝑖𝑛 + 𝐸𝑜𝑓𝑓𝑠𝑒𝑡 𝛼𝜌𝐸′𝑚𝑎𝑥 + (1 + 𝜌 − 𝛼𝜌)𝐸′𝑚𝑖𝑛
Dividing denominator and nominator by E’min we finally find the equation for the theatre contrast,
the second important equation:
(1 + 𝛼𝜌)𝐶𝑅𝑝𝑟𝑜𝑗 + (1 − 𝛼)𝜌
𝐶𝑅𝑡ℎ𝑒𝑎𝑡𝑟𝑒 =
𝛼𝜌𝐶𝑅𝑝𝑟𝑜𝑗 + (1 + 𝜌 − 𝛼𝜌)
eq.2
Now we work out an approximation of this last equation and prove that this results into the well
known ‘parallel resistor’ formula for contrasts.
We know that the theatre reflectivity in practical situations will be very small and that takes
values between 0 and 1. Thus we get with a reasonable projector contrast CRproj:
(1 + 𝛼𝜌)𝐶𝑅𝑝𝑟𝑜𝑗 + (1 − 𝛼)𝜌 ≈ 𝐶𝑅𝑝𝑟𝑜𝑗
The denominator of CRtheatre can be approximated by
𝛼𝜌𝐶𝑅𝑝𝑟𝑜𝑗 + (1 + 𝜌 − 𝛼𝜌) ≈ 𝐶𝑅𝑝𝑟𝑜𝑗 𝛼𝜌 + 1
Using these approximations we find
1 1 1
= +
𝐶𝑅𝑡ℎ𝑒𝑎𝑡𝑟𝑒 𝐶𝑅𝑝𝑟𝑜𝑗 1/𝛼𝜌
The same approximation can be applied to the projector contrast CRproj:
References
[1] T. Kunkel and E. Reinhard. “A reassessment of the simultaneous dynamic range of the
human visual system”. Proceedings of the 7th Symposium on Applied Perception in Graphics
and Visualization. 17–24 (2010).
[2] S.Farrell, T. Kunkel, S. Daly “A Cinema Luminance Range by the People, for the People:
Viewer Preferences on Luminance Limits for a Large Screen Environment”, Proc. SMPTE
Annual Technical Conference October 2014.
[3] P. Ludé. “High dynamic range cinema, How black is black”. NAB technology summit on
cinema , April 2015
[4] M. Schuck, S. Gilman, G. Sharp, "3D Image Quality in the PLF Environment”, Display
Summit, June 2015