David Townend wrote: > So do I take it as 'given' that if I am using non OEM inks and / or papers I > should be making a custom linearisation for each printer setting and paper > combo that I wish to then make a custom printer profile for ?
Yes, if you want the best possible results this is the way to go. > And that doing > this with a Greyscale Step Wedge 'by eye' is in fact not very satifactory ? Sometimes it's ok, sometimes it's not. However if you use an RGB driver you don't have any means (unless you are a very clever software engineer) of linearizing the printer driver, except to play with the printer settings (No colour adjustment, Photo realistic etc, and the paper settings). These settings are the best you can do in terms of linearization. > > As an aside, it appears you are also saying that it is not really possible > to get an accurate linearistion with non OEM inks and papers without going > through a RIP ? Yes, that is what I'm saying (except if the OEM inks are exactly similar nin colour to the original inks). You may still get a very nice profile with the above mentioned method though, which will be satisfactory in almost all areas on almost all prints. But occasionally you will get this one image that just comes out weirdly posterized in the shadows of a saturated colour or just plain wrong in a specific (usually dark saturated) colour, or with a slight colour bias in the neutral 3/4 tones or deep shadows (the areas where the light inks [if any] are decreasing and the dark inks are increasing - or the area where the 3/5 colour inks decrease rapidly and the black increase rapidly). This erroneous behaviour is a result of the profile not being accurate enough (too few samples), but the underlying reason for this is a bad linearization. In theory we could "simply" produce some RGB targets with 15-20.000 patches. This would space the colours of the patches closer together so the application wouldn't have to interpolate as much as with only 1000 parches. However there is still no guarantee that 20K patches would solve the problems entirely, but it is almost certain to introduce some heavy posterization. Oh and it would take about a day or two to measure the patches (provided that you don't fuck up the measurement and that all the patches are printed with perfectly clean nozzles and are dust free) so the price of a profile would probably be around �800-1200 and require you to print about 50-70 A4 pages of patches pending on instrument <G>. As an interesting aside Bill Atkinson (renowned nature photographer) did a lot of experimenting setting up his Epson 9600. It took him more than thousand hours, several home made software applications, and about 1/2 million spectral measurements to devise a way of producing accurate profiles for his Epson 9600. After creating his profiles he offered the 918 patch RGB profiling target (he ended up using) to Gretag Macbeth to incorporate in ProfileMaker Pro and now Eye One Photo. You'll notice that this is the preferred target by most people who profile printers remotely as it simply works better than anything I've seen around. If you want the full Bill A. story you need to do some digging at google... Best Regards Thomas Holm / Pixl ApS - Photographer & Colour Management Consultant - Adobe Certified Training Provider in Photoshop� - Apple Solutions Expert - Colour Management - Imacon Authorized Scanner Training Facility - Remote Profiling Service (Output ICC profiles) - Seminars speaker and tutor on CM and Digital Imaging etc. - Home Page: www.pixl.dk � Email: th[AT]pixl.dk -- =============================================================== GO TO http://www.prodig.org for ~ GUIDELINES ~ un/SUBSCRIBING ~ ITEMS for SALE
