digital camera scanning: how to fix dust

[Einst_Stein]

Advertisement (gone after registration)

Some film scanners can reduce dust with IR light source and scanner based Sw tools.

If you use digital camera to scan the film, how do you remove film dusts?  

[frame-it]

healing brush in lightroom..though nowadays i like to leave the dust because i like how it looks 😇

[frame-it]

or you can get this plugin, never used it in photoshop, but its lousy inside silverfast for big pieces of dust and scratches, works fine for tiny bits of dust

or try and find :

Polaroid Dust & Scratch Filter

http://web.archive.org/web/20071015015231/http://www.polaroid.com/service/software/poladsr/pdsr1_0.exe

Edited January 29, 2021 by frame-it

[wda]

Immediately before firing the shutter,  apply a carefully controlled puff of canned or blown air to remove loose particles of dust. 

[marchyman]

I use a flat bed scanner which is a dust magnet.   Before a scanning session I use an anti-static cloth to wipe the scanner down.   Before each scan I use an anti-static brush on both scanner glass and the item to be scanned, negative or print.  That is followed with judicious use of a rocket blower on glass and item to be scanned.

All of that minimizes dust.  It does not eliminate it.   I take care of the rest using the healing brush in Capture One.   I've found looking at the scan at 200% and using a brush with hardness set to around 60% works well. 

[250swb]

19 hours ago, Einst_Stein said:

Some film scanners can reduce dust with IR light source and scanner based Sw tools.

If you use digital camera to scan the film, how do you remove film dusts?  

Scanner software is pretty crude and can degrade a colour negative, or in the case of a B&W negative not work at all. So re-touching after the scan is often the best way anyway.

I have my negatives firmly clamped in a Lomo DigitaLIZA holder so a blast of compressed air across the surface (not directly onto the surface) gets rid of most dust, then in post processing I use a healing brush. Its a fast process if you've been careful to avoid dust, and much quicker than the long scan time using the IR function with a scanner.

[Guest]

Advertisement (gone after registration)

IR-based cleaning is by far the best method for slides and colour negatives (not for B&W negatives, critical for Kodachrome slides), both in terms of quality and user time. Even if one is very careful to avoid dust and scratches on the film, many old slides and negatives have defects which cannot be removed by cleaning.

Original:

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

After IR-based cleaning (SilverFast iSRD):

I do not see a realistic procedure to accomplish the same result manually, at least not for more than a couple of images.

Hermann-Josef

[Einst_Stein]

I  expect what can be done manually can be mimiced automatically by ML. What can be done by IR corection can be done by detecting absolute black or absolute white spots. Absolute dark and absolute white can be identified by extra over exposure scan pass and extra under rxposure scan pass.

I am surprised there is no such tool yet.

[Guest]

What is ML?

26 minutes ago, Einst_Stein said:

What can be done by IR corection can be done by detecting absolute black or absolute white spots.

I don't think this is true. How do you differentiate between defect and scene? This is exactly the problem with IR-based correction of Kodachrome. The main advantage of the IR-image is the fact that it shows the defects but not the scene (except for B&W and to some extend Kodachrome).

Hermann-Josef

[marchyman]

11 hours ago, Jossie said:

I do not see a realistic procedure

Patience.  If you don't have time constraints a little extra care pre-scan drastically reduces the spots to remove making post scan retouching a smaller burden.  Spots in sky, for example, are easy and quick to fix.  Other spots take more time.  Some a lot of time.

Scanning and retouching a roll of 24 color negatives typically takes me 2-4 hours.  Because I'm scanning many decade old and poorly stored negatives I sometimes spend more time doing color correction than dust removal. These are family snapshots, not fine art.  My goal is to reproduce on screen what would have been seen when the prints came back from the local drug store photo processor.  5-10 minutes per image is worthwhile for that.

Would I do the same if it was a job or taking time away from a job?  Probably not.

[Einst_Stein]

8 hours ago, Jossie said:

What is ML?

I don't think this is true. How do you differentiate between defect and scene? This is exactly the problem with IR-based correction of Kodachrome. The main advantage of the IR-image is the fact that it shows the defects but not the scene (except for B&W and to some extend Kodachrome).

Hermann-Josef

ML = Machine Learning.

Any dust removal is based on a method, whatever it is, to differentiate the dust and real feature. IR correction is based on the opaque vs. transparency of the dust vs. real image feature under the IR light source. That's what makes  non-C41 B&W or Kodachrome more difficult. I am willing to leave them as a separate problem. What I care most is the C41 or E6, especially old family pictures.  My thought is limited to these.

But no matter what, achieving 100% can only be dreamed, so far. My point is based on the "absolute dark" or "absolute white" property of dusts or scratches. This is where over-exposure scan path or under exposure scan pass  is about.  Nothing can be perfect, though. The further down improvement may be ML to differentiate the "mis-correction" according to the surrounding features. 

The recommended SDRx from Silverfast seems a useful tool. I am about to contact the vendor for pricing and trial.    

Edited January 30, 2021 by Einst_Stein

[kokoshawnuff]

Since I stated scanning with a dslr I’ve been able to avoid almost all dust by setting up a kind of assembly line: 

For camera scanning the entire roll of film uncut, so the scanning process is easier and faster. I store the long strips using a Print File 35-1m continuous roll  

I make sure everything (table, light source, mounts, etc) is as dust free as possible (wipe down, blower), then I feed the film through a kinetronics static vac (I imagine the cheaper static brush would do a good job too) directly into my light source (sunray box) and out. 
 

I can photograph a 36exp roll in about 2 min with maybe 5 min of LR healing brush work total per roll. 
 

 

[Doug A]

I use only distilled water for my entire developing process and washing (Ilford method). I snap the wet film between my hands to shake off most of the water and hang it in a shower stall that is used several times a day. I cut the film into 6-frame lengths and put it into Print File pages as soon as it dry. When I am ready to digitize it with my BEOON setup I run each strip between two washed and dried fingers before I pass it through the cutouts in the BEOON base.I have virtually no dust on my "scans" - less than a minute per frame in Affinity Photo.

[Stuart Richardson]

Dust removal is a chain that starts in camera. Make sure that your camera body clean and blow out the dust with a dust blower. This will help prevent pre-exposure dust (black on negatives, white on slides), when processing, try to do so in a clean environment, especially for drying and use clean, filtered water. In the processing room and scanning room, it is best to have a HEPA filter that has been running for at least an hour or so near the scanner or processing room.

When scanning, scan the film before it is put in plastic sheets, immediately after processing. The longer it has been sitting, the more it accumulates dust and grime, and the act of removing the film from the plastic causes static cling. Keep your humidity above 30 or 40% to keep static from being a major issue. When you are placing the film in your hold or in your setup, use the a light table light and hold your film on an angle to it so that the bright light illuminates the dust spots. Use a blower to remove the dust as much as possible. If it does not work, use an anti-static brush and after doing so, do it again, as brushes usually deposit a few bits of dust as they work. But they do loosen it so it is easier to remove with air.

Once you have finally done your scan, if all has gone well, there should be very little dust on the scan. If there is dust, I recommend using Photoshop over Lightroom. The content aware spot healing tool is much more sophisticated and performs better than the one in Lightroom. If you want help, create an adjustment layer with a curve and adjust the contrast so that it makes it easier to see the dust spots. You might have to to it in several different settings.

Spot the dust at 100% or so for most cases.

The problem with dust reduction technologies is that spotting dust from detail is a lot harder than it sounds. Is that a hair, or a distant fence? Is that dust, or white flowers in a field, grains of sand on someones leg? Basically, even to our brains very small bright detail or dark detail can be confusing. To machines it is hopeless. There are some programs that do this, but all that I have tried are more useful to rescue truly awful negatives than they are useful in general use. The one in Flex Color is ok, but it also likes to eat details like fences, wires, grass etc. Even the dust and scratches filter in photoshop can be useful in extreme cases, such as vintage negatives that are totally saturated with dust, but then it is a balance between loss of detail and presentation of the photographic image. The IR one is probably the best, but it requires a hardware solution and licensing, so it is only available in certain commercial scanners. Overall the best dust spotting technique is to prevent dust. Take it from me...I am the guy who has to clean it up for everyone else, so if there were an easier way to do it, I would be the first in line. When it really comes down to it, the only true solution is time spent, and few photos merit that treatment. The best work of that kind that I ever saw were huge William Eggleston prints at David Zwirner gallery a few years ago. They were vintage negatives and yet were totally spotless with no evidence at all that they had been touched. I do this for a living, so know how difficult that would achieve. The prints were selling for 30-40000 dollars each, so it was clearly worth it to pay someone to work on each photo for hours or days to bring them to that level.

[fotomas]

vor 30 Minuten schrieb Stuart Richardson:

The best work of that kind that I ever saw were huge William Eggleston prints at David Zwirner gallery a few years ago. They were vintage negatives and yet were totally spotless with no evidence at all that they had been touched. I do this for a living, so know how difficult that would achieve. The prints were selling for 30-40000 dollars each, so it was clearly worth it to pay someone to work on each photo for hours or days to bring them to that level.

As far as I know Eggleston prints are made as dye transfers. Also a total different beast. You need color matrixes in size of the print for this. Guess that there are also retouching possibilities during this process, but I don't know. Since it was available long before anything digital I would guess it must have been done the classical way, by removing any dust before, or maybe spotting things out with a brush.

In the analogue past I sometimes made prints on photographic paper from slides for customers. Don't really liked it and preferred printing from negs for my own stuff. Once for the tonality, but also because you have to remove any dust before, when printing from slides. No fun, but possible with patient. From negs it was much easier, because you could spot out the dust also afterwards.
Now with scanning I try to find a good balance by removing as much dust as possible in a few seconds before scanning and do the rest later on the computer.
With a brush of camel hair and a compressor I would estimate I get rid of 80 to 90% of the dust before. But mostly there is something left.
Another nasty thing is dirt in the washing water. Here you can do nearly nothing about it afterwards. Color film is affected much more then B & W. Here we have quite a lot dirt in the water. After installing a 5 microns filter for the washing water things get much better, but even now not 100% perfect.
Guess dust becomes a larger problem with digitizing films, because you can see everything when zooming in on the computer. Even small spot you wouldn't have noticed in an analog print because they was to small to see. 
On the computer I remove them all, but this took me longer then before when I spotted the, maybe up to 5, visible dust spots out with a brush on the analogue print.

[Guest]

1 hour ago, Stuart Richardson said:

The IR one is probably the best, but it requires a hardware solution and licensing, so it is only available in certain commercial scanners.

This is not correct. Even my rather cheap flat bed scanner is equipped with IR-based cleaning. The point is that colour negatives (not b&w negatives) and slides (not Kodachrome) are transparent to IR-radiation. Dust absorbs and scratches scatter IR-radiation, so both appear as dark areas in the IR-image. This is used to located the defects very efficiently (no ambiguity between dust and scene since the scene is not present) and software can then correct them in the RGB-image at exactly that position and shape.

IR-image of the example I showed above. Please not the almost complete absence of the street lamp at right!

Welcome, dear visitor! As registered member you'd see an image here…

Simply register for free here – We are always happy to welcome new members!

 

Everything dark is a defect and is perfectly corrected by the software using this IR-image.

Hermann-Josef

Edited February 14, 2021 by Jossie

[Stuart Richardson]

Hi, sorry, when I said commercial, I meant on the consumer market. It is not incorporated into the Imacon/Hasselblad scanners, drum scanners or the Creo/Kodak Eversmart and iQsmart scanners and cannot be used in camera based scanning, or things like the Phase One/Digital Transitions setup as far as I am aware. But I wish they were. It is a great solution. If I recall correctly, Imacon did not use it because the licensing was expensive and they thought it would make the expensive scanners even more so, and they were confident their own software based solution was good enough (it can work, but it works better with lower res scans.)

Fotomas -- These prints were digital and 40x60 inches or so. They were the first digital versions of Eggleston's work and it was quite controversial at the time precisely because they were not dye transfer and because they were very large, as opposed to most of his work which were printed small. Controversy aside, they were fantastic examples of digital prints. To the extent that I tried to find out who made them to see if I could meet them. I work as an exhibition printer, and it is not that often that I see prints that I am slightly baffled as to how they could be done so well. I did not see the originals of course, but the quality of the prints suggested to me that they must have been done with a massive amount of time and a lot of skill. Not a grain was out of place...

 

[Guest]

27 minutes ago, Stuart Richardson said:

If I recall correctly, Imacon did not use it because the licensing was expensive and they thought it would make the expensive scanners even more so, and they were confident their own software based solution was good enough

Imacon/Hasselblad used an optical element, a diffusor, to reduce the appearance of the defects.

Hermann-Josef

[Stuart Richardson]

Yes. I started with a 646 in my studio and later switched to the X5, and the diffusor does make a difference, but in my experience it is not as effective as a diffused head in the darkroom. The piece of acrylic they use is really quite small and not so thick. It helps, but the best anti-dust measure in the those scanners is the absence of glass. The other thing that helps is that if you enter the debugging menu in the scanner software (press d-e-b-g in sequence), you can choose the "monitor window" which gives a realtime readout of the illumination of the CCD. Any gaps in the histogram are generally pieces of dust or debris on the diffusor, lamp or ccd. This does not manifest as dust in the scan, but as banding, especially in color negative. It is good to have your eye on...

 

[Advertisement]

Hallo,

Look here: Interesting Leica gear

[DanielaIvanova]

On 1/30/2021 at 1:57 PM, Einst_Stein said:

I  expect what can be done manually can be mimiced automatically by ML. What can be done by IR corection can be done by detecting absolute black or absolute white spots. Absolute dark and absolute white can be identified by extra over exposure scan pass and extra under rxposure scan pass.

I am surprised there is no such tool yet.

I stumbled upon this reply while looking for discussion on dust removal approaches that is more recent. I am a PhD student working on using ML to remove dust, scratches and other dirt in film- it’s built around a passion project of mine since I also shoot, develop and scan my own film. I’ve already published a paper based on the results from my proof of concept  model if you are interested in reading about it. To train that model I added damage programmatically to already “clean” scans. I am currently looking to collect more examples of scans with film damage (dust, hair, scratches, dust) to define the problem better - it is very difficult to obtain those since many people, especially if they use IR based dust removal, do not even save the dusty scans. Collecting image pairs of dusty and cleaned up scans would be ideal- if anyone is willing to contribute such data please let me know!