Great post, tdc123. The one thing I'd add is the the ICE technology, which uses an infrared channel to subtract out dirt (which scatter light differently when it hits dirt) doesn't really work with B&W, and is also not very good with Kodachrome. It works brilliantly, however, with most other color emulsions and with negatives (although few amateurs deal with color negative movie film). So, take that into account if the service you use promises to remove dirt using ICE.
+ Reply to Thread
Results 31 to 33 of 33
Thanks, johnmeyer. Yes, digital ice does not work on b&w. For slides in general, we just clean them off well physically and correct in Photoshop in almost all cases. Digital ICE isn't flawless, not to mention being quite slow. We find manual correction to be more accurate and quicker for a bit of dust. I trust my skills more than an automated tool. I know it has a following but it's just a matter of workflow and confidence in the product. Being that Digital ICE does sometimes produce its own correction artifacts correcting areas that shouldn't be corrected (fusing eyelids and other fine details), it's hard to be confident that some minute digital artifact wasn't missed in QA needing to be rescanned. Preserving direct from the original without changes and fixing by hand in Photoshop is just a lot safer. And an employee couldn't then claim it's the scanner's fault. It can be useful in some instances, but we don't use it as the first line of defense. Currently our options are Epson V700, Noritsu HS1800 and Hasselblad/Imacon. I understand the [discontinued] Nikon 9000 is well regarded, but for the reasons above, we avoid automated correction in this case.
Be aware that the quality of the results with ICE (a product that was invented by Applied Science Fiction, or ASF, and licensed to scanner companies) depends not only on ICE itself, but on the software that interprets the results. I found this out through extensive scanning of still photos using a Nikon Coolscan scanner. The Nikon software produced very spotty results, like what you report. Fortunately, I found an American expatriate living in England named Ed Hammrick who developed a brilliant scanner control software program called Vuescan. He adapted it to not only the Nikon line of scanners but also to Minolta, Canon, and many other scanners, including flatbeds. His software's interpretation of ICE produce excellent results in many cases, although there is nothing that is perfect when it comes to removing dirt and correctly replacing it with some clone of the adjacent pixels. However, his software ran circles around Nikon's software even though they were both using the same exact ICE data.
Last edited by johnmeyer; 31st Dec 2018 at 02:15. Reason: typo