<s>
Tone	B-Language
mapping	I-Language
is	O
a	O
technique	O
used	O
in	O
image	B-Algorithm
processing	I-Algorithm
and	O
computer	O
graphics	O
to	O
map	B-Algorithm
one	O
set	O
of	O
colors	O
to	O
another	O
to	O
approximate	O
the	O
appearance	O
of	O
high-dynamic-range	B-General_Concept
images	I-General_Concept
in	O
a	O
medium	O
that	O
has	O
a	O
more	O
limited	O
dynamic	O
range	O
.	O
</s>
<s>
Print-outs	B-General_Concept
,	O
CRT	B-Device
or	O
LCD	B-Device
monitors	I-Device
,	O
and	O
projectors	O
all	O
have	O
a	O
limited	O
dynamic	O
range	O
that	O
is	O
inadequate	O
to	O
reproduce	O
the	O
full	O
range	O
of	O
light	O
intensities	O
present	O
in	O
natural	O
scenes	O
.	O
</s>
<s>
Tone	B-Language
mapping	I-Language
addresses	O
the	O
problem	O
of	O
strong	O
contrast	O
reduction	O
from	O
the	O
scene	O
radiance	B-Application
to	O
the	O
displayable	O
range	O
while	O
preserving	O
the	O
image	O
details	O
and	O
color	O
appearance	O
important	O
to	O
appreciate	O
the	O
original	O
scene	O
content	O
.	O
</s>
<s>
Inverse	O
tone	B-Language
mapping	I-Language
is	O
the	O
inverse	O
technique	O
that	O
allows	O
to	O
expand	O
the	O
luminance	O
range	O
,	O
mapping	O
a	O
low	O
dynamic	O
range	O
image	O
into	O
a	O
higher	O
dynamic	O
range	O
image	O
.	O
</s>
<s>
It	O
is	O
notably	O
used	O
to	O
upscale	O
SDR	B-Device
videos	I-Device
to	O
HDR	B-Application
videos	I-Application
.	O
</s>
<s>
Early	O
film	O
developers	O
attempted	O
to	O
remedy	O
this	O
issue	O
by	O
designing	O
the	O
film	O
stocks	O
and	O
the	O
print	O
development	O
systems	O
that	O
gave	O
a	O
desired	O
S-shaped	B-Algorithm
tone	I-Algorithm
curve	I-Algorithm
with	O
slightly	O
enhanced	O
contrast	O
(	O
about	O
15%	O
)	O
in	O
the	O
middle	O
range	O
and	O
gradually	O
compressed	O
highlights	O
and	O
shadows	O
.	O
</s>
<s>
Despite	O
this	O
,	O
if	O
algorithms	O
could	O
not	O
sufficiently	O
map	B-Algorithm
tones	O
and	O
colors	O
,	O
a	O
skilled	O
artist	O
was	O
still	O
needed	O
,	O
as	O
is	O
the	O
case	O
with	O
cinematographic	O
movie	O
post-processing	O
.	O
</s>
<s>
Several	O
tone	B-Language
mapping	I-Language
operators	I-Language
were	O
developed	O
to	O
map	B-Algorithm
high	B-General_Concept
dynamic	I-General_Concept
range	I-General_Concept
(	O
HDR	O
)	O
images	O
to	O
standard	O
displays	O
.	O
</s>
<s>
Currently	O
,	O
image	O
reproduction	O
has	O
shifted	O
towards	O
display-driven	O
solutions	O
since	O
displays	O
now	O
possess	O
advanced	O
image	B-Algorithm
processing	I-Algorithm
algorithms	I-Algorithm
that	O
help	O
adapt	O
rendering	O
of	O
the	O
image	O
to	O
viewing	O
conditions	O
,	O
save	O
power	O
,	O
up-scale	O
color	O
gamut	O
and	O
dynamic	O
range	O
.	O
</s>
<s>
The	O
goals	O
of	O
tone	B-Language
mapping	I-Language
can	O
be	O
differently	O
stated	O
depending	O
on	O
the	O
particular	O
application	O
.	O
</s>
<s>
Various	O
tone	B-Language
mapping	I-Language
operators	I-Language
have	O
been	O
developed	O
in	O
the	O
recent	O
years	O
.	O
</s>
<s>
Those	O
techniques	O
are	O
simple	O
and	O
fast	O
(	O
since	O
they	O
can	O
be	O
implemented	O
using	O
look-up	B-Data_Structure
tables	I-Data_Structure
)	O
,	O
but	O
they	O
can	O
cause	O
a	O
loss	O
of	O
contrast	O
.	O
</s>
<s>
Examples	O
of	O
common	O
global	O
tone	B-Language
mapping	I-Language
methods	O
are	O
contrast	O
reduction	O
and	O
color	O
inversion	O
.	O
</s>
<s>
A	O
simple	O
example	O
of	O
global	O
tone	B-Language
mapping	I-Language
filter	O
is	O
(	O
Reinhard	O
)	O
,	O
where	O
Vin	O
is	O
the	O
luminance	O
of	O
the	O
original	O
pixel	O
and	O
Vout	O
is	O
the	O
luminance	O
of	O
the	O
filtered	O
pixel	O
.	O
</s>
<s>
This	O
function	O
will	O
map	B-Algorithm
the	O
luminance	O
Vin	O
in	O
the	O
domain	O
to	O
a	O
displayable	O
output	O
range	O
of	O
While	O
this	O
filter	O
provides	O
a	O
decent	O
contrast	O
for	O
parts	O
of	O
the	O
image	O
with	O
low	O
luminance	O
(	O
particularly	O
when	O
)	O
,	O
parts	O
of	O
the	O
image	O
with	O
higher	O
luminance	O
will	O
get	O
increasingly	O
lower	O
contrast	O
as	O
the	O
luminance	O
of	O
the	O
filtered	O
image	O
goes	O
to	O
1	O
.	O
</s>
<s>
A	O
perhaps	O
more	O
useful	O
global	O
tone	B-Language
mapping	I-Language
method	O
is	O
gamma	O
compression	O
,	O
which	O
has	O
the	O
filter	O
where	O
and	O
.	O
</s>
<s>
This	O
function	O
will	O
map	B-Algorithm
the	O
luminance	O
V	O
in	O
the	O
domain	O
to	O
the	O
output	O
range	O
γ	O
regulates	O
the	O
contrast	O
of	O
the	O
image	O
;	O
a	O
lower	O
value	O
for	O
lower	O
contrast	O
.	O
</s>
<s>
An	O
even	O
more	O
sophisticated	O
group	O
of	O
tone	B-Language
mapping	I-Language
algorithms	O
is	O
based	O
on	O
contrast	O
or	O
gradient	B-Algorithm
domain	I-Algorithm
methods	O
,	O
which	O
are	O
'	O
local	O
 '	O
.	O
</s>
<s>
Those	O
tone	B-Language
mapping	I-Language
methods	O
usually	O
produce	O
very	O
sharp	O
images	O
,	O
which	O
preserve	O
very	O
well	O
small	O
contrast	O
details	O
;	O
however	O
,	O
this	O
is	O
often	O
done	O
at	O
the	O
cost	O
of	O
flattening	O
an	O
overall	O
image	O
contrast	O
,	O
and	O
may	O
as	O
a	O
side	O
effect	O
produce	O
halo-like	O
glows	O
around	O
dark	O
objects	O
.	O
</s>
<s>
Examples	O
of	O
such	O
tone	B-Language
mapping	I-Language
methods	O
include	O
:	O
gradient	B-Algorithm
domain	I-Algorithm
high	I-Algorithm
dynamic	I-Algorithm
range	I-Algorithm
compression	I-Algorithm
and	O
A	O
Perceptual	O
Framework	O
for	O
Contrast	O
Processing	O
of	O
High	B-General_Concept
Dynamic	I-General_Concept
Range	I-General_Concept
Images	O
(	O
a	O
tone	B-Language
mapping	I-Language
is	O
one	O
of	O
the	O
applications	O
of	O
this	O
framework	O
)	O
.	O
</s>
<s>
Another	O
approach	O
to	O
tone	B-Language
mapping	I-Language
of	O
HDR	O
images	O
is	O
inspired	O
by	O
the	O
anchoring	O
theory	O
of	O
lightness	O
perception	O
.	O
</s>
<s>
The	O
key	O
concept	O
of	O
this	O
tone	B-Language
mapping	I-Language
method	O
(	O
Lightness	O
Perception	O
in	O
Tone	O
Reproduction	O
)	O
is	O
a	O
decomposition	O
of	O
an	O
HDR	O
image	O
into	O
areas	O
(	O
frameworks	O
)	O
of	O
consistent	O
illumination	O
and	O
the	O
local	O
calculation	O
of	O
the	O
lightness	O
values	O
.	O
</s>
<s>
This	O
approach	O
to	O
tone	B-Language
mapping	I-Language
does	O
not	O
affect	O
the	O
local	O
contrast	O
and	O
preserves	O
the	O
natural	O
colors	O
of	O
an	O
HDR	O
image	O
due	O
to	O
the	O
linear	O
handling	O
of	O
luminance	O
.	O
</s>
<s>
One	O
simple	O
form	O
of	O
tone	B-Language
mapping	I-Language
takes	O
a	O
standard	O
image	O
(	O
not	O
HDR	O
–	O
the	O
dynamic	O
range	O
already	O
compressed	O
)	O
and	O
applies	O
unsharp	B-Algorithm
masking	I-Algorithm
with	O
a	O
large	O
radius	O
,	O
which	O
increases	O
local	O
contrast	O
rather	O
than	O
sharpening	O
.	O
</s>
<s>
See	O
unsharp	B-Algorithm
masking	I-Algorithm
:	O
local	O
contrast	O
enhancement	O
for	O
details	O
.	O
</s>
<s>
One	O
of	O
the	O
commonly	O
used	O
tone	B-Language
mapping	I-Language
algorithms	O
is	O
the	O
iCAM06	O
which	O
is	O
based	O
on	O
both	O
the	O
color	O
appearance	O
model	O
and	O
hierarchical	O
mapping	O
.	O
</s>
<s>
Forms	O
of	O
tone	B-Language
mapping	I-Language
long	O
precede	O
digital	O
photography	O
.	O
</s>
<s>
The	O
manipulation	O
of	O
film	O
and	O
development	O
process	O
to	O
render	O
high	O
contrast	O
scenes	O
,	O
especially	O
those	O
shot	O
in	O
bright	O
sunlight	O
,	O
on	O
printing	O
paper	O
with	O
a	O
relatively	O
low	O
dynamic	O
range	O
,	O
is	O
effectively	O
a	O
form	O
of	O
tone	B-Language
mapping	I-Language
,	O
although	O
it	O
is	O
not	O
usually	O
called	O
that	O
.	O
</s>
<s>
The	O
normal	O
process	O
of	O
exposure	O
compensation	O
,	O
brightening	O
shadows	O
and	O
altering	O
contrast	O
applied	O
globally	O
to	O
digital	O
images	O
as	O
part	O
of	O
a	O
professional	O
or	O
serious	O
amateur	O
workflow	O
is	O
also	O
a	O
form	O
of	O
tone	B-Language
mapping	I-Language
.	O
</s>
<s>
However	O
,	O
HDR	O
tone	B-Language
mapping	I-Language
,	O
usually	O
using	O
local	O
operators	O
,	O
has	O
become	O
increasingly	O
popular	O
amongst	O
digital	O
photographers	O
as	O
a	O
post-processing	O
technique	O
,	O
where	O
several	O
exposures	O
at	O
different	O
shutter	O
speeds	O
are	O
combined	O
to	O
produce	O
an	O
HDR	O
image	O
and	O
a	O
tone	B-Language
mapping	I-Language
operator	I-Language
is	O
then	O
applied	O
to	O
the	O
result	O
.	O
</s>
<s>
In	O
some	O
cases	O
local	O
tone	B-Language
mapping	I-Language
is	O
used	O
even	O
though	O
the	O
dynamic	O
range	O
of	O
the	O
source	O
image	O
could	O
be	O
captured	O
on	O
the	O
target	O
media	O
,	O
either	O
to	O
produce	O
the	O
distinctive	O
appearance	O
of	O
a	O
locally	O
tone	O
mapped	O
image	O
,	O
or	O
to	O
produce	O
an	O
image	O
closer	O
to	O
the	O
photographer	O
's	O
artistic	O
vision	O
of	O
the	O
scene	O
by	O
removing	O
sharp	O
contrasts	O
,	O
which	O
often	O
look	O
unattractive	O
.	O
</s>
<s>
However	O
,	O
when	O
tone	B-Language
mapping	I-Language
is	O
applied	O
to	O
a	O
single	O
exposure	O
in	O
this	O
way	O
,	O
the	O
intermediate	O
image	O
has	O
only	O
normal	O
dynamic	O
range	O
,	O
and	O
the	O
amount	O
of	O
shadow	O
or	O
highlight	O
detail	O
that	O
can	O
be	O
rendered	O
is	O
only	O
that	O
which	O
was	O
captured	O
in	O
the	O
original	O
exposure	O
.	O
</s>
<s>
One	O
of	O
the	O
original	O
goals	O
of	O
tone	B-Language
mapping	I-Language
was	O
to	O
be	O
able	O
to	O
reproduce	O
a	O
given	O
scene	O
or	O
image	O
onto	O
a	O
display	O
device	O
such	O
that	O
the	O
brightness	O
sensation	O
of	O
the	O
image	O
to	O
a	O
human	O
viewer	O
closely	O
matches	O
the	O
real-world	O
brightness	O
sensation	O
.	O
</s>
<s>
For	O
this	O
method	O
,	O
models	O
for	O
the	O
Human	O
Visual	O
System	O
(	O
HVS	O
)	O
and	O
the	O
display	O
are	O
first	O
generated	O
,	O
along	O
with	O
a	O
simple	O
tone	B-Language
mapping	I-Language
operator	I-Language
.	O
</s>
<s>
The	O
images	O
on	O
the	O
right	O
show	O
the	O
interior	O
of	O
a	O
church	O
,	O
a	O
scene	O
which	O
has	O
a	O
variation	O
in	O
radiance	B-Application
much	O
larger	O
than	O
that	O
which	O
can	O
be	O
displayed	O
on	O
a	O
monitor	O
or	O
recorded	O
by	O
a	O
conventional	O
camera	O
.	O
</s>
<s>
The	O
six	O
individual	O
exposures	O
from	O
the	O
camera	O
show	O
the	O
radiance	B-Application
of	O
the	O
scene	O
in	O
some	O
range	O
transformed	O
to	O
the	O
range	O
of	O
brightnesses	O
that	O
can	O
be	O
displayed	O
on	O
a	O
monitor	O
.	O
</s>
<s>
The	O
range	O
of	O
radiances	B-Application
recorded	O
in	O
each	O
photo	O
is	O
limited	O
,	O
so	O
not	O
all	O
details	O
can	O
be	O
displayed	O
at	O
once	O
:	O
for	O
example	O
,	O
details	O
of	O
the	O
dark	O
church	O
interior	O
cannot	O
be	O
displayed	O
at	O
the	O
same	O
time	O
as	O
those	O
of	O
the	O
bright	O
stained-glass	O
window	O
.	O
</s>
<s>
An	O
algorithm	O
is	O
applied	O
to	O
the	O
six	O
images	O
to	O
recreate	O
the	O
high	B-General_Concept
dynamic	I-General_Concept
range	I-General_Concept
radiance	B-Application
map	B-Algorithm
of	O
the	O
original	O
scene	O
(	O
a	O
high	B-General_Concept
dynamic	I-General_Concept
range	I-General_Concept
image	I-General_Concept
)	O
.	O
</s>
<s>
Alternatively	O
,	O
some	O
higher-end	O
consumer	O
and	O
specialist	O
scientific	O
digital	O
cameras	O
are	O
able	O
to	O
record	O
a	O
high	B-General_Concept
dynamic	I-General_Concept
range	I-General_Concept
image	I-General_Concept
directly	O
,	O
for	O
example	O
with	O
RAW	B-Application
images	I-Application
.	O
</s>
<s>
In	O
the	O
ideal	O
case	O
,	O
a	O
camera	O
might	O
measure	O
luminance	O
directly	O
and	O
store	O
this	O
in	O
the	O
HDR	O
image	O
;	O
however	O
,	O
most	O
high	B-General_Concept
dynamic	I-General_Concept
range	I-General_Concept
images	O
produced	O
by	O
cameras	O
today	O
are	O
not	O
calibrated	O
or	O
even	O
proportional	O
to	O
luminance	O
,	O
due	O
to	O
practical	O
reasons	O
such	O
as	O
cost	O
and	O
time	O
required	O
to	O
measure	O
accurate	O
luminance	O
values	O
it	O
is	O
often	O
sufficient	O
for	O
artists	O
to	O
use	O
multiple	O
exposures	O
to	O
gain	O
an	O
"	O
HDR	O
image	O
"	O
which	O
grossly	O
approximates	O
the	O
true	O
luminance	O
signal	O
.	O
</s>
<s>
The	O
high	B-General_Concept
dynamic	I-General_Concept
range	I-General_Concept
image	I-General_Concept
is	O
passed	O
to	O
a	O
tone	B-Language
mapping	I-Language
operator	I-Language
,	O
in	O
this	O
case	O
a	O
local	O
operator	O
,	O
which	O
transforms	O
the	O
image	O
into	O
a	O
low	O
dynamic	O
range	O
image	O
suitable	O
for	O
viewing	O
on	O
a	O
monitor	O
.	O
</s>
<s>
Relative	O
to	O
the	O
church	O
interior	O
,	O
the	O
stained-glass	O
window	O
is	O
displayed	O
at	O
a	O
much	O
lower	O
brightness	O
than	O
a	O
linear	O
mapping	O
between	O
scene	O
radiance	B-Application
and	O
pixel	O
intensity	O
would	O
produce	O
.	O
</s>
<s>
Local	O
tone	B-Language
mapping	I-Language
technique	O
of	O
HDR	O
image	B-Algorithm
processing	I-Algorithm
often	O
produces	O
a	O
number	O
of	O
characteristic	O
effects	O
in	O
images	O
such	O
as	O
bright	O
halos	O
around	O
dark	O
objects	O
,	O
dark	O
halos	O
around	O
bright	O
objects	O
,	O
and	O
sometimes	O
a	O
"	O
cartoon-like	O
"	O
appearance	O
due	O
to	O
extremely	O
vivid	O
colors	O
and	O
lack	O
of	O
large-scale	O
color	O
variations	O
.	O
</s>
<s>
These	O
results	O
are	O
caused	O
by	O
application	O
of	O
geometric	O
space	O
distortion	O
of	O
captured	O
image	O
along	O
with	O
color	O
space	O
distortion	O
,	O
while	O
only	O
color	O
space	O
distortions	O
are	O
a	O
tone	B-Language
mapping	I-Language
effect	O
,	O
and	O
all	O
other	O
distortions	O
are	O
rather	O
a	O
custom	O
filtering	O
technique	O
than	O
any	O
tone	O
or	O
color	O
space	O
mapping	O
.	O
</s>
<s>
So	O
,	O
the	O
results	O
of	O
local	O
tone	B-Language
mapping	I-Language
are	O
often	O
judged	O
as	O
perverting	O
the	O
nature	O
of	O
a	O
documentalist	O
photographic	O
image	O
and	O
far	O
from	O
photographic	O
realism	O
.	O
</s>
<s>
Reducing	O
dynamic	O
range	O
with	O
tone	B-Language
mapping	I-Language
is	O
often	O
useful	O
in	O
bright	O
sunlit	O
scenes	O
,	O
where	O
the	O
difference	O
in	O
intensity	O
between	O
direct	O
illumination	O
and	O
shadow	O
is	O
great	O
.	O
</s>
<s>
Use	O
of	O
tone	B-Language
mapping	I-Language
in	O
this	O
context	O
may	O
not	O
be	O
apparent	O
from	O
the	O
final	O
image	O
:	O
</s>
<s>
Tone	B-Language
mapping	I-Language
can	O
also	O
produce	O
distinctive	O
visual	O
effects	O
in	O
the	O
final	O
image	O
,	O
such	O
as	O
the	O
visible	O
halo	O
around	O
the	O
tower	O
in	O
the	O
Cornell	O
Law	O
School	O
image	O
below	O
.	O
</s>
<s>
Halos	O
in	O
images	O
come	O
about	O
because	O
the	O
local	O
tone	B-Language
mapping	I-Language
operator	I-Language
will	O
brighten	O
areas	O
around	O
dark	O
objects	O
,	O
to	O
maintain	O
the	O
local	O
contrast	O
in	O
the	O
original	O
image	O
,	O
which	O
fools	O
the	O
human	O
visual	O
system	O
into	O
perceiving	O
the	O
dark	O
objects	O
as	O
being	O
dark	O
,	O
even	O
if	O
their	O
actual	O
luminance	O
is	O
the	O
same	O
as	O
that	O
of	O
areas	O
of	O
the	O
image	O
perceived	O
as	O
being	O
bright	O
.	O
</s>
