How to Remove Banding on Negative Scans?

[Nebbit]

Hey there! I just had a huge batch of film processed locally and was very disappointed to find banding across many of my images. At first I thought it was dust caught in the Flexlight scanner I had been using however, the lines were in different places across each image. It turns out it was an after-effect of the dip processing method used at the lab I sent the film to. They say it has been a recurring issue with the equipment they inherited and that they do their best to minimize it (which is normally the case). Unfortunately, this is the most trusted film lab in Toronto.

Consequently, this leaves me with a lot of work in post to get rid of these issues. Does anyone know of a best practice?Ordinarily, I would just use the healing brush tool for such problems, but because these bands span the entire length of the image and often go through sky, using this is not a best approach considering the healing brush does not work well on very large areas, or on gradients. Any efficient methods out there for removing streaks and lines, that hopefully don't involve hours of work ?

I've included an image where you can see the bands in question in the top portion of the sky, running the length of the frame horizontally.


Thanks!

 

[Lachlan Young]

Frequency Separation approaches work well in my experience with this sort of problem.

 

[railwayman3]

I have to say that I am surprised that any "trusted" lab would admit that they have a visible recurring processing fault caused by their equipment, and which they don't seem able to eliminate, particularly as dip-and-dunk is probably the most established system used now, and for many years, for B&W, C41 and E6. Doesn't give any confidence, and you should perhaps complain and ask for, at least, replacement film and refund of processing costs ??

 

[Mainecoonmaniac]

Banding comes from missing data. If you get your scans are in JPEG, the compression algorithm throws out data. Another issue could be the bit depth of the original scan. The more data, the less possibility of banding.

 

[NortheastPhotographic]

I have to say that I am surprised that any "trusted" lab would admit that they have a visible recurring processing fault caused by their equipment, and which they don't seem able to eliminate, particularly as dip-and-dunk is probably the most established system used now, and for many years, for B&W, C41 and E6. Doesn't give any confidence, and you should perhaps complain and ask for, at least, replacement film and refund of processing costs ??

Just goes to show there is really no perfect way to process color film. Our Noritsu V30 (leader card type) requires regular up keep and cleanings, but I love the results I get from it (C41). High volume shops need dip/dunk but the leader card machines can be great for those of us doing less than 1000 rolls a week. For sheet film we are moving to Expert Drums. I personally wouldn't want to have to maintain a dip and dunk line. The big shops that run them need them but it introduces a lot of complexity.

 

[mshchem]

What the heck is a healing brush? Will it work on hair loss?

 

[Nebbit]

I have to say that I am surprised that any "trusted" lab would admit that they have a visible recurring processing fault caused by their equipment, and which they don't seem able to eliminate, particularly as dip-and-dunk is probably the most established system used now, and for many years, for B&W, C41 and E6. Doesn't give any confidence, and you should perhaps complain and ask for, at least, replacement film and refund of processing costs ??

I was to haha. The business is or was partially owned by Edward Burtynsky at some point, so one would expect the best, but sadly it seems to be something they have not been able to resolve completely. Out of the two major (and to my knowledge only) labs for processing in the city, it is the better. Fortunately, they gave me a full refund.

 

[Kino]

Yeah, I'm not seeing any problem with the image you posted. Can you draw an arrow to the defect?

I'm a bit confused here: did the lab say the banding was a processing issue or a scanner issue?

Mach banding is where density shifts occur when the bit depth is too low to accurately transition on subtle density graduations. Most typically, it happens on 8 bit images; 10 bit being the minimum to keep this artifact from occurring.

If it is the scan (and not the negative) with banding issues, then the lab is probably down-converting the image to 8 bit and screwing the pooch to keep data wrangling to a minimum. If this is the case I would go back and tell them this is unacceptable and ask them to give you at least 10 bit images.

An old trick to minimize banding in 8 bit images is to lightly add noise to the image (yes, random noise). This breaks up the gradient enough to minimize the artifact and was extensively used in the early years of CGI in motion pictures when 8 bit was the only game in town.

 

[Nebbit]

Frequency Separation approaches work well in my experience with this sort of problem.

Thanks Laclan. I'll look into it and see if it can help me

 

[Nebbit]

Banding comes from missing data. If you get your scans are in JPEG, the compression algorithm throws out data. Another issue could be the bit depth of the original scan. The more data, the less possibility of banding.

Hey Mainecoonmaniac, thanks for your reply. It is definitely not a scanning issue as I'm scanning raw on Flexlight scanner at 3200 dpi. The light banding in the top sky is due to the tech issues in my lab's dip and dunk process.

 

[Nebbit]

Just goes to show there is really no perfect way to process color film. Our Noritsu V30 (leader card type) requires regular up keep and cleanings, but I love the results I get from it (C41). High volume shops need dip/dunk but the leader card machines can be great for those of us doing less than 1000 rolls a week. For sheet film we are moving to Expert Drums. I personally wouldn't want to have to maintain a dip and dunk line. The big shops that run them need them but it introduces a lot of complexity.

Thanks! I'll look into seeing if there are any smaller labs around that use leader card processing

 

[Nebbit]

Hey Mainecoonmaniac, thanks for your reply. It is definitely not a scanning issue as I'm scanning raw on Flexlight scanner at 3200 dpi. The light banding in the top sky is due to the tech issues in my lab's dip and dunk process.

Additionally, I should add it's not banding as in the digital variety. There are bands of very thin lines running across the images, wasn't sure what else to call it.

 

[Nebbit]

Yeah, I'm not seeing any problem with the image you posted. Can you draw an arrow to the defect?

I'm a bit confused here: did the lab say the banding was a processing issue or a scanner issue?

Mach banding is where density shifts occur when the bit depth is too low to accurately transition on subtle density graduations. Most typically, it happens on 8 bit images; 10 bit being the minimum to keep this artifact from occurring.

If it is the scan (and not the negative) with banding issues, then the lab is probably down-converting the image to 8 bit and screwing the pooch to keep data wrangling to a minimum. If this is the case I would go back and tell them this is unacceptable and ask them to give you at least 10 bit images.

An old trick to minimize banding in 8 bit images is to lightly add noise to the image (yes, random noise). This breaks up the gradient enough to minimize the artifact and was extensively used in the early years of CGI in motion pictures when 8 bit was the only game in town.

Thanks for your response @Kino ! Unfortunately, it won't let me upload anything with higher resolution, but I have attached a close up screen cap to give you an idea. The lines are very faint, but will definitely show when printed at a reasonable size. It is definitely more difficult to notice at lower resolutions.

 

[Nebbit]

Just for additional clarity ...

I'm scanning the negatives myself, uncompressed on an Imacon Flexlight scanner at 3200dpi. The hazy, thin lines that run across the top portion of the image are what I am referring to. They are unrelated to the scanning as they are on the negative and not the consequence of anything digital. Normally, I use photoshop's spot healing brush to get rid of dust and other such details, however that tool is notorious for struggling with large sections and across gradients, such as the sky in an image. I'm just looking for the simplest and most effective way to get rid of them to avoid spending hours fixing up each effected neg.

Thanks again for everyone's quick responses!

 

[Kino]

Ugh. That's a tough one.

If the lines are very straight and very well defined, and you have a very large scan to work with, you could draw a box around the entire line, go into curves and manipulate the RGB curves to try to bring it in line with the surrounding pixels.

Might have to experiment a bit with the feathering of the box you draw, but if you manipulate the RGB and gamma data, I'll bet you can minimize the line.

I had to digitally restore 7 films from 1907 to 1926 that were tinted and toned and you would not believe some of the repairs I had to do and how I had to do them. Of course, I used both Photoshop 32bit and After Effects extensively with many plugins, but you can do some amazing stuff with a bit of exploration...

 

[Kino]

You could try this software on a trial basis: http://akvis.com/en/retoucher/download-image-restoration.php

The films I did were in 2008-2009 and it was all pretty much manual masking, pixel stealing and adjustment, there has been a lot of progress in defect removal algorithms...

 

[Patrick Robert James]

If you use the healing brush or even the clone tool you can fix that no problem. To get it to go faster, pick one spot, then you can go to the next area that has a break in the information (change of density/color/structure) of the line and hold the shift key down and click. PS will fill the line in for you. If there is a big density change between the line and the surrounding area sometimes it is better to use the clone tool to get the density closer then the healing brush. You can just stack the layers if you want. Sometimes the healing brush tool works a lot better if you break the line into small pieces.

I can think of some other ways to deal with this too, but no need to overcomplicate it.

I'm assuming you know this, but I'll mention it for others. Always retouch first. Always use a separate layer for the retouching.

Hope that helps you.

 

[Lachlan Young]

Additionally, I should add it's not banding as in the digital variety. There are bands of very thin lines running across the images, wasn't sure what else to call it.

They might be intra-layer drying marks - they can come from the negative being dried too fast - often from too much heat relative to ambient humidity.

Leader card machines can introduce all sorts of scratches and physical damage - the best solution of all is likely processing in a Jobo and carefully controlled drying time.

 

[Mainecoonmaniac]

Hey Mainecoonmaniac, thanks for your reply. It is definitely not a scanning issue as I'm scanning raw on Flexlight scanner at 3200 dpi. The light banding in the top sky is due to the tech issues in my lab's dip and dunk process.

Wow. Flexlight are the Cadillac of scanners. Also, dip and dunk is the better film processors. They beat the roller film processors. Have you tried scanning them yourself to check that it's not the scanner?

 

[removed account4]

If you use the healing brush or even the clone tool you can fix that no problem. To get it to go faster, pick one spot, then you can go to the next area that has a break in the information (change of density/color/structure) of the line and hold the shift key down and click. PS will fill the line in for you. If there is a big density change between the line and the surrounding area sometimes it is better to use the clone tool to get the density closer then the healing brush. You can just stack the layers if you want. Sometimes the healing brush tool works a lot better if you break the line into small pieces.

I can think of some other ways to deal with this too, but no need to overcomplicate it.

I'm assuming you know this, but I'll mention it for others. Always retouch first. Always use a separate layer for the retouching.

Hope that helps you.

+1

i'd use the rubber stamp and clean up using the healing brush .. once you get a system down it takes no time at all....

have fuN !
john

 

[cramej]

To me, this doesn't look like a processing problem...

 

[Mainecoonmaniac]

To me, this doesn't look like a processing problem...
View attachment 235746

This looks like dirty scanner glass if the scanner used glass.

 

[ced]

Try one scan from top to bottom (i.e. different orientation) & see...

 

[Kino]

Banding on Imacon/Hasselblad Scanners:
http://www.thedigitalcoach.com/ewExternalFiles/Banding Tutorial.pdf

and

https://www.largeformatphotography....php?133659-Banding-in-Skies-on-Imacon-scanner

 

[Kino]

And finally, this: http://www.martinzimelka.com/homepa.../3_Imacon_343_manual_CCD_pixel_remapping.html

Might be of help; who knows?