Digital BW, The Print

--- In DigitalBlackandWhiteThePrint@yahoogroups.com, "Loris Medici" 
<lorism@t...> wrote:
> Please forgive my limited English understanding (and expression 
ability)
> but when one talks about a LUT I understand "a list consisting of
> limited amnt. of rows consisting from actual value -> converted 
value",
> if one talks about a transfer *function* there are no limited 
samples
> such a LUT but a "mathematical equation giving converted values for
> every possible actual values"... A tonal adjustment curve is a 
transfer
> function not a LUT (if it were so, we would have different set of 
curves
> for 8 and 16 bit modes - and the 16bit mode curve files' size 
would be
> 256K)

I work as an image-processing engineer in an R&D lab and there is 
incomplete agreement among my clleagues whether "transfer function"  
continous values - most people accept that it could be discrete 
values but a few prefer "mapping funtion" for anything implemented 
in a LUT.  My guess is that most of the curves we see in Photoshop 
and profile-creation tools are actually implemented somewhere as 
LUTs simple because they are easy to implement that way, easy to 
edit, and readily accomodate pixels values that are already in a 
discrete domain (i.e., a pixel might be 253 or 254 but not 253.5).  
But, nonetheless, I chose the term "transfer function" instead 
of "curves" or "LUT", because I wanted a generic term that wasn't 
specific to a particular tool or implementation, and because it's a 
term regularly used in image and signal processing, which IS after 
all, what we're talking about here.