Jorge writes: > ...just to see him checking the Histogram of one of > images, showing obvious signs of posterization.
Was this _visible_ posterization on the monitor or hard proof, or just a gappy histogram? This makes all the difference. > Next, he just went to Channels and selected each individual channel, applied > a rather small Gaussian Blur ( like 0.4) to each Channel, went back to the > overall HIstogram and Voila!! No posterization. > A new unsharp masking to override the softness introduced by the Blurr and > this was it. > The scene was repeated with all files with no major checking. > > I don't know if this is standard technique, I found it an "easy way out" to > counter damage infringed to a digital file, and he said he just does it all > the time. > > I am asking you all , because I found it hard to accept. Does this really > not affect printing ( not Epsons), I mean real -4 colour- printing quality?? > And what about colors?? Does this not introduce changes in colors?? There are many ways to 'mung and blur' a file, if producing a better histogram than image is required (enter Lee Varis?). Profile conversion, dithering, noise, rotation/unrotation, interpolation, jpeg artifacts, sub pixel nudges, quantization and other things can slightly change the statistical report on a file (histograms, unique levels and colours etc). What can statistacally appear as 'more data' can actually be less data. I personally do not spend much time with the histogram, never for our own scans anyway (I know how our drum behaves). For a shot that is not from our section, then I may initially inspect the histogram as one small part of my overall checks, but afterwards it would be rare for me to bother. Once I have placed fixed colour samplers on the endpoints and other key tones, I always know at a glance what is happening for the tones which are critical to me. When doing commercial imaging work (as opposed to internal magazine work) then some clients do care about the histogram and other data concerns and this may be more of an issue - but at the end of the day it's the output that matters and histograms are not the best indicator of output quality. A very good article on histograms can be found here: http://www.marginalsoftware.com/HowtoScan/tutorial_page_1.htm A bit like the recent wheel spinning over gamut warnings...they have a point but how the image looks should be the major consideration. Stephen Marsh. =============================================================== GO TO http://www.prodig.org for ~ GUIDELINES ~ un/SUBSCRIBING ~ ITEMS for SALE
