Dear Martin, On Feb 24, 2011, at 8:27 AM, Idea Digital Imaging wrote: > On 24 Feb 2011, at 12:30, Iliah Borg wrote: > >> But of course, chart is on the film. Reproduction here is of a physical >> object, that is why we are operating measured values, not the desired >> values. So, in your words - "the chart on the film" is the subject of >> reproduction. This allows to approximate how the particular film responds to >> the light. > > > How useful are the results of this process? > > I looked at trying to profile negatives on our drum scanners more than a > decade ago but the results were very poor. > > You could build a good profile from a correctly exposed test image but > variations in over/under exposure in real world images rendered the profiles > useless. > With input profiles for "moving target" media like film or digital captures (moving in the sense they depend on the colour of light, exposure, development, aging, etc) the task of the profiles is to convert colours, just colours - not the tone. Ironically, even auto-levels before applying a matrix-based profile do a better job than an elaborative hand-edited LUT profile with an embedded tone curve. Alternatively, profiles done from targets shot using different exposures are in wide use; operator just selects the proper proper profile based on visual. > And, as you also need to incorporate a wide choice of print papers into > colour managment, I didn't really see how profiling would be faster or better > than having a good scanner operator getting the colour right during scanning? I do not see how papers come into the blend. Applying output profiles is accomplished in a much later stage. Input profiles are helpers, they cut the time an operator spends working on an image; but most importantly they lessen the fatigue. -- Iliah Borg ib@xxxxxxxxxxx