ðòïåëôù 


  áòèé÷ 


Apache-Talk @lexa.ru 

Inet-Admins @info.east.ru 

Filmscanners @halftone.co.uk 

Security-alerts @yandex-team.ru 

nginx-ru @sysoev.ru 

  óôáôøé 


  ðåòóïîáìøîïå 


  ðòïçòáííù 



ðéûéôå
ðéóøíá












     áòèé÷ :: Filmscanners
Filmscanners mailing list archive (filmscanners@halftone.co.uk)

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: filmscanners: Setting screen gamma problem



Sorry to be difficult, but I don't believe that this is correct, and this 
is exactly what I would like some confirmation of - either way.  The whole 
point of the non-linear transformations or mapping that leads to the 
'gamma' that we are discussing is to make equal digital brightness steps 
cause equal PERCEIVED brightness changes on screen.  The means of doing 
this and the non-linear response of the eye vs the non-linear response of 
the tv tube and how it all ties together is complicated, but the end result 
AIUI is that one digital bit at the bottom end of the greyscale should 
represent the same APPARENT visual difference as one bit at the bright end 
of the greyscale.

The reason this is done is to maximise the utility of the miserable 256 
brightness levels we have available.  If we didn't perform this logarithmic 
mapping, a bit at the high end would represent a very small difference in 
perceived brightness and we might have more resolution than we need, while 
at the low end a bit would represent a very large difference in perceived 
brightness, so in dark areas we would not have enough resolution.  To avoid 
this situation we map the ACTUAL screen brightness to a power law of the 
digital levels.  This power law matches the eye's logarithmic response so 
that the PERCEIVED screen brightness is directly proportional to the 
original digital levels.

If this is true then a 16-bits step at the bottom (dark) end should give 
the same apparent brightness difference as 16 bits at the top (bright) end, 
and the same in the middle of the range.  And so my 'linear' (linear in 
terms of digital values) ramp should look evenly spaced in PERCEIVED 
brightness on screen.

I believe that on this first-order analysis that the ramp should look 
evenly spaced.  When I asked the question I was wondering whether there is 
some other second-order reason why such a ramp should NOT look evenly 
spaced.  I can't think of a reason why, but there might be one.

My original reason for asking this was that I can NOT get my ramp to look 
linear without setting the gamma to silly levels - about 
1.4.  Alternatively, if I set gamma correctly, the ramp does not look 
linear - small steps at the bottom and big ones at the top. Hence I 
wondered whether my understanding was incorrect, or if there was something 
else wrong with the set up of my monitor.  It seemed a simple enough 
question at the time.

Does anyone know?

Julian

PS as a check, I set my monitor gamma to 1.0 (i.e. linear mapping of bits 
to perceived brightness), and the ramp looks very unevenly spaced.  This 
time the big steps are at the bottom, with virtually no visible difference 
between the top levels, as I would expect if things work as I understand.

Also, when adjusting gamma I have - after reading many posts and emails - 
very carefully set my black point because incorrect black point will make 
any attempt to set gamma fairly meaningless.


Julian asked...
|Hence my
| original question - should such a step wedge look evenly spaced on a well
| set up monitor?

Maris replied...
>No, it should not.  Monitors use gamma of necessity as the guns do not
>display light linearly.  Gamma is logarithmic - hence a non-logarithmic step
>wedge with even tone spacing will not and should not *appear* evenly spaced
>in tone on your monitor.
>
>Maris


Julian Robinson
in usually sunny, smog free Canberra, Australia




 




Copyright © Lexa Software, 1996-2009.