mendel_leisk Posted December 17, 2002 Share Posted December 17, 2002 I'm cleaning scans of moderately scratched and specked black and white negative film. I don't have ICE. If I did I don't believe it would be of use. I'm using all the tricks I can muster in photoshop: healing brush, history brush with dust and scratch/noise added snapshot, and, very occasionally, clone stamp. I have a lot of images and it's hard slogging. I've looked at a few existing programs/plugins that purport to automate the process. In my opinion, they seem rudimental and unable to discriminate between image and defect consistently and satisfactorily. Also, they apply an unacceptable overall degradation of image tone and quality. I was thinking about the thought process I go through when I'm looking at the image to determine what is a defect, the approach I take to remedy the defect, and wondering if software could be written using similar process, given todays PC's, ram, etcetera. When viewing an image, it usually very obvious to me (painfully so) what is a defect. But then, I can read hand scrawled text which would give an OCR program fits. Some criteria for I can think of for determining what is a defect: 1. Tone is abruptly different. 2. Adjacent area has no similar abrupt differences. For example, in scan of a pebbly beach, small specks just disappear, so what's the point of hunting for them. 3. Edges of these tone differences are abnormally sharply defined, as compared to image as a whole. 4. In the case of scratches, runs contrary to adjacent image, is too consistent, is sharp edged per item 3, and seems out of sync with rest of image. When I've determined something is a defect, and a readily observable defect, my last step is to disappear the defect as seamlessly as reasonably possible with adjoining tones and patterns. If there was a program or plugin that could achieve this consistently, without degrading image quality, I'm sure it would a gold mine for the developers. I know I'd buy it in a blink! Any thoughts, ideas? Link to comment Share on other sites More sharing options...
kumar_yajamanam Posted December 17, 2002 Share Posted December 17, 2002 Have you looked at Polaroid Dust & Scratch Removal software? It is an excellent tool, that you can fine tune per your needs. Best of all, it is a freeware!!! Link to comment Share on other sites More sharing options...
mendel_leisk Posted December 17, 2002 Author Share Posted December 17, 2002 I tried the Polaroid plug-in: found with my Tri-X scans it more or less obliterated grain before removing about 50% of defects. Smooth areas which had nice grain texture became smudged grey with isolated black specks. Kind of like the ocean bottom after a toxic spill. Also, it does a number on 16 bit histograms. 8 bit histograms seem ok. When I say intelligent software, I mean something that can assess whole image for defects, treat only the actual culprits, in a SEAMLESS AND UNNOTICEABLE manner, and leave the image pristine. I know I'm asking for the moon, but, I can do it by hand with photoshop, albeit in time-eating tendonitis-inflicting fashion, so... Link to comment Share on other sites More sharing options...
carl smith Posted December 17, 2002 Share Posted December 17, 2002 I've not found anything that can do that. ICE/FARE are the best things for removing from this image things that are on the film or scratches in the film fairly accurately. However it is always possible that the damage is so great that you have to do it yourself. I don't expect anything to come along for a good while that can intelligently determine image data type and effectively correct errors. In fact that's part of the craft, knowing how to clean up images, either digitally or in the darkroom or post darkroom touch up. Link to comment Share on other sites More sharing options...
steve_bingham Posted December 17, 2002 Share Posted December 17, 2002 Photoshop has a dust and scratch filter. However, to use it right does require some skill. Otherwise it tends to be overkill. Link to comment Share on other sites More sharing options...
mendel_leisk Posted December 18, 2002 Author Share Posted December 18, 2002 Hi Carl, Yeah, I guess if you look long and hard for something and don't find it, chances are it isn't there to be found, for now. And ICE wouldn't help me if I had it, with Tri-X. Looks like I've got myself a hobby for a bit longer, I've got very slightly less than 1800 images ahead. Some programmer could really clean up if he/she could make some sort of quantum leap in this field. Seems to me the key would be: anything truly part of image has grainy edges, whereas defects have sharper edges. Hi Steve, I use healing brush and history brush (sourcing from dust&scratch, noise added snapshot) about equally. Occasionally, non-aligned clone stamp saves my bacon, when nothing else works. History brush is the only way I apply dust&scratch, and only for spot application of tricky dust motes on complex patterns or edges of contrast. I'm basically following Ian Lyons tutorial, per: (http://www.computer-darkroom.com/tutorials/tutorial_5_1.htm) Someone here was talking about using dust and scratch targeted on entire shadow area, to clean up noise, with following edit|fade|darken only. I tried it but found it too invasive. Maybe with some more tweaking, but I don't think so. I don't think it would be saving me any time anyway. One thing I've found with the healing brush, even though it is supposed to NOT be suitable in areas of strong contrast, I've found it behaves a lot like clone stamp, in that it doesn't "suck" tone from adjacent area, for something like a scratch crossing a sudden tone change edge, as long as you align your source and target points carefully on the tone edge, and pull right across. Also, I've started trying the lasso with healing brush, to isolate the affect. I usually don't need that though. Something that's saving me from going blind trying to see the "damn near invisible" diameter representation brushes, is to set the display to standard, which gives you clear albeit goofy, picture of the tool for a pointer, visible against any tone. You lose pictorial representation of brush dia., though. If I've got a moderately clean image, and it's busy, ie: lots of grass, foliage, waves, sand or other random pattern to break up defects, I'll get through it in 10 minutes, even less. But typically, I'll spend close to 1/2 hour on clean up alone, and that's over and above scanning. Anyway, thanks all for input. Link to comment Share on other sites More sharing options...
ron_russell Posted December 18, 2002 Share Posted December 18, 2002 My own limited experience shows dust removal makes the image less sharp; however, scratch removal is accomplished by using the scratch removal tool on each scratch such as featured on programs like Paint Shop Pro 7, which is about 50 times easier to use than Photoshop and uses far less resources but is excellent. Link to comment Share on other sites More sharing options...
bill crookston Posted December 18, 2002 Share Posted December 18, 2002 You don't say how you initially clean the negaives before you scan them. I used to use jet air and brushes on my negatives but I now use a film cleaner called a Booflet (www.boofey.com). Since then dust has not been a big problem on the scans that I do. Link to comment Share on other sites More sharing options...
mendel_leisk Posted December 18, 2002 Author Share Posted December 18, 2002 Hi Bill, I give both sides a gentle wipe with a regular blower brush bristle, then vigorous blow with another blower brush. I've tried canned compressed air, with some trepidation, not a lot of difference. I don't believe I'm missing a lot of the loose dust. Most of the dust I'm dealing with does not budge from scan to scan. I've experimented: clean, scan, clean, scan again, compare. I use ACDSee viewer program to toggle between images, looking for variations. Akin to looking for planet/astroids in star field. My problems are stuck or embedded dust, unidentified "crud" which could be sediment from water supply and most bothersome: scratches. I think a lot of these scratches stem from my bad practice with enlarger way back when, moving strip within holder instead of taking holder out of head and opening it fully. I would say 99% of the stuff is on film back. I'm scanning (with Dimage Scan Dual II) with film emulsion facing scanning mechanism (up) and doing focus with each scan (Vuescan). Link to comment Share on other sites More sharing options...
Recommended Posts
Create an account or sign in to comment
You need to be a member in order to leave a comment
Create an account
Sign up for a new account in our community. It's easy!
Register a new accountSign in
Already have an account? Sign in here.
Sign In Now