Film Scanning Workflow
Introduction
I wanted to make some notes and comments on scanning of film. I have been mostly self taught since I started in 2009 with my brother's archive that I wanted to create a book from. I had a basic scanner for 35mm film and slides and used the included SW and Photoshop Elements 6. I could not and still cannot justify in my own mind a full-blown Photoshop license.I got reasonably good results. I learned the importance of cleaning dust before scanning as this saves a lot of time touching them up in Photoshop. (Some of his slides were filthy.) I scanned as TIFF files as these don't suffer losses from compression like JPG. I also tried scanning at a high resolution like 4000 DPI. I thought of myself as making archival scans as a result. The idea being to scan to capture all the information and then make modifications from that in Photoshop.
What I did not understand was the degree to which the scanner SW was making changes to make the scan 'better'. Eventually I moved to the system I have now which is an Epson V500 and Vuescan SW. At the time Vuescan looked better than Silverfast. Recently I saw Nick Carver's scanning video and the Silverfast looks quite impressive now. (There are other opinions however.) Also the EpsonScan SW that comes with the scanner can do some impressive work. Alex Burke uses it as part of his workflow described in his excellent eBook.
Part of the motivation for this post is to document for myself my workflow and the rationale for it. I often find writing down or discussing a subject clarifies problems and inconsistencies. Looking at what Nick and Alex do there is a lot of differences. They both achieve good results but they have different workflows for different reasons.
Other Photographer's Workflows
Nick's flow (as I see it). Nick's YouTube outlines a 'quick and dirty' approach. No wet scanning and getting results quickly. Obviously he mentions wet scanning and on other of his videos he uses anti-newton glass. But this is what gleaned from this video.Scanner Scanner SW Post_processing
Epson V800 SilverFast 8 Lightroom
Negative Inversion Color Correction Dust Removal Sharpening
SilverFast SilverFast/Lightroom Epson V800 ICE Lightroom
Nick relies on Silverfast for most of the inversion and color correction for negatives. They have a negative feature that includes lots of film profiles to get the conversion close in the first place. It seems impressive so I cannot fault this. He is a big fan of ICE (infrared dust removal) which I understand. I and others however have detected distortions or information loss from ICE so I don't use it unless the film is very dusty.
Alex uses EpsonScan rather extensively.
Scanner Scanner SW Post_processing
Epson V800 EpsonScan Lightroom
Negative Inversion Color Correction Dust Removal Sharpening
EpsonScan EpsonScan/Lightroom ??? Lightroom
EpsonScan does very good job at inversion from a color negative (better than Vuescan) and Alex then makes manual adjustments to reduce clipping that EpsonScan does by default. I have seen others actually do color correction and editing directly in EpsonScan with histograms and curves. In fact I think it should be possible to do extensive color work entirely within EpsonScan if one were particularly short of funds.
My Flow
From a flow description standpoint I would start with color negative work first. I think it is the most demanding and difficult aspect of the film-digital workflow. Transparency film I have less problems with as there is no orange color mask, you can view the transparency and see the colors there as a reference. Black and white is simplest as there is no white balance or color correction by definition.The other truth is that it is up to the photographer him/herself to decide on the final color balance of the image. Only they were there at the time and able to remember the scene and the mood. Nick makes reference to this inherent difficulty when he talks about going back and forth obsessively on color cast or white balance. I have this same problem.
I also suffer from a couple of other problems. If I review a series of corrected images from the same roll of film in the same location taken at the same time of day I can sometimes see wide differences in the overall color. This I need to work on in terms of discipline to apply the same basic settings and workflow to the image. I also have noted that images I have modified in the past and been happy with at the time, I can view later and see them as having a strange color cast. This is despite being 'happy' with the result I had. This is probably a lack of experience.
That brings me to where I am in my film scanning workflow.
Most of it is driven around the adoption of ColorFast SW for color film inversion. In the past my ability to get good color our of color negative film has been abysmal. I excitedly bought Kodak Ektar when it came out and was completely frustrated in my ability to get decent color from scanned images. Vuescan is lousy at color conversion. (I treat Vuescan as a scanning SW package only.)
Photoshop Elements is no better and the tools available combined with my knowledge made it a real struggle. It gets more difficult if there is no obvious color neutral area to adjust white balance from. So I abandoned color negative film and shot almost exclusively Ektachrome, Velvia, Astia, and Provia for my color work. This of course suffered many times from problems with dynamic range.
Eventually I discovered and decided to pay for ColorNeg software. The cost of this plus Photoshop Elements makes a poor man's proper Photoshop for me. This is the summary of my tools and workflow.
Scanner Scanner SW Post_processing
Epson V500 Vuescan Photoshop Elements
Negative Inversion Color Correction Dust Removal Sharpening
ColorNeg ColorNeg Photoshop Elements Photoshop Elements
Scanning
For me the objective of scanning is to capture as raw and complete a version of the negative as possible. Scanning is such a time consuming step my objective is to do it once. This means...- 48-bit--16 bits per color for maximum dynamic range
- TIFF file format -- No compression/No loss
- Linear Scan--No gamma curve applied.
- No ICE--No infrared dust removal scan step.
- High resolution-- Scan once and scale down later in photoshop.
I am in the process of adding BetterScanning negative holders to improve the results for my 120 film. I also want to look at a larger piece of ant-newton glass for my 4x5 negatives. I will probably just buy the glass and tape the negatives down and make some spacers. With the V500 the scanning area is only about 3 inches wide so 4x5 requires 2 passes and stitching. Some results of BetterScanning are here.
For the Vuescan settings I use the ColorNeg recommendations. 48-bit scans, output to 48-bit TIFF RAW files files. I am on the fence about multiple scans. This will reduce noise but you probably need at least 4 scans to see some small decrease in noise and some have recommended 16 passes. This takes a lot of time in highest resolution. I have done some tests here. There is a definite improvement in noise levels.
The next action for Vuescan is making sure the full range of the scan in each color is well centered in the scanners range. Per the ColorFast recommendations I made a series of test scans (at reduced resolution to save time.) In the first case the Analog gains in Vuscan are all set to 1 for R,G, and B channels. The exposure is also locked. Typically red will be brightest (histogram furthest to the right) followed by green and the blue when the color histograms are viewed in Photoshop.
Histograms--For those not familiar with statistics histograms can be a difficult concept to get your head around. Every scanned (or any other digital image) represents each point in the image (a pixel) as a range of numbers. For 8-bit per color images this is a value between 0 and 255 (2 to the power of 8 is 256 for the mathematical). 255 is the brightest value and 0 is the darkest. Each color pixel has 3 values (red green and blue) with each color having a value of 0 to 255. If all three colors are 255 in value then you have white for instance. For sake of simplicity let us consider a black and white image. Here there is just one value for pixel between 0 and 255 (white to black) and the middle values are levels of gray. If I take all of the pixels in the image and count how many pixels there are for each value (0-255) I can graph this and that is a histogram. Another way to visualize it is to have 255 bins arranged from right to left and fill the right most bin with all the pixels with a value of 255 (white) the bin to the left of that I put all the 254 valued pixels and I proceed to the left until the last bin on the far left I place all the 0 pixels (black). This gives me a view of the overall picture exposure.
I add more gain (exposure) to the green and blue channels in Vuescan, make another scan and then look at the histograms. The idea is to get these aligned and centered so that neither end of the histogram is is running into the left or right edge of the histogram box. (The test negative you choose should be reasonably well exposed and not over or under exposed.) In this way you can make sure the film fits the full range of the scanner. If the histogram for any given color is scrunched up at either end of the histogram chances are you are 'clipping' some pixels and these will be rendered as maximum brightness or darkness in one indistinguishable mass. I will save these settings and label them for the film type such as Kodak Portra 800.
ColorNeg
Once I have an image scanned with high resolution I bring it into Photoshop and I crop any film frame leaving only the image. (The film frame often confuses color correction SW including scanning SW if you are using the scanner to do the correction. Nick Carver mentions this as a revelation in Silverfast negative inversion results.)I then move to the Filter menu and select ColorFast and then work in ColorNeg. I then select ColorNeg if it isn’t already set there. There is a ColorPos option for positive images or transparency film and TouchUp for fixing an already inverted image. (In Photoshop Elements some operations like straightening only work on 8-bit images so I may make those changes later. JPGs will also only be 8-bit.)
Next I select the film type to match what I have scanned. At this point the image is usually pretty close in terms of color rendition. The next step for me is to adjust the exposure with the Black setting.
I then left click a color neutral area if I want to make some basic color/white balance correction. I often experiment with clicking different areas until I get what I want. Then because this affects the color saturation I use the White option which is a unique type of control. It adds and subtracts white to the image. White is added by moving the slider up and removed sliding it down. The effect is to lower or increase contrast.
Typically I am looking to improve contrast so I move it down but not much as color does not take much contrast change unlike black and white. Almost inevitably this will increase the apparent saturation of the colors. Alternatively or additionally one can use the Gamma entry to manipulate contrast. This gamma change does not impact the color but only the brightness or luminance.
Now I will look to add some saturation depending on the film and image. I typically start with a value of 120. I do the White or contrast setting first because it impacts the apparent saturation. If you set the saturation first then look to the contrast you can get cloying colors on your way to getting the contrast right.
Next I make more subtle color changes. I use the Ring CC tool. This initially confusing tool places a thumbnail of the current image in the center and surrounds it with different color-shifted versions (magenta, blue, yellow, red, green, cyan, and darker and lighter). This images are directions to push the image color and are thumbnails corrected by a unit of one in each direction. Clicking on a colored image moves the image in than direction by 1/2 the amount displayed. All the images are refreshed then to show the thumbnail thus corrected. You can cancel the effect by clicking the colored thumbnail directly opposite the central image.
It takes some getting used to but I am increasingly learning to use it. By clicking out of the Ring CC box you can see your full sized image again. These steps usually get me to a desired image color.
There are many more options in the SW most of which I am still learning about. There is a very nice highlight and shadow compression tool that is worth looking at. This makes tonal adjustments int he shadows and highlights to retain more detail.
Recently I have begun to look at their version of Curves/Zones in the context of B and W negatives. I find already there is a lot of control available in B and W negatives compared to Photoshop Elements.
After exiting ColorPerfect then I might make a small level adjust for final brightness and then sharpen the image with Unsharp Mask.
I can then save this as a 16-bit TIFF or convert to 8-bit before saving as JPEG. These will be large files and so not suitable for web or social media use. I often save a smaller JPG version by reducing the DPI of the image.
Photoshop Elements does not have the extensive layers feature that leave the original image in tact. Hence the reason I do not save over the original scan. I want that scan to be raw and pristine. One other downside of Photoshop Elements is that some operations like straightening and dust removal only work on 8-bit images. This means my raw scan cannot have dust removal performed on it and retain the 16-bit information. This means all the dust removal work I have done on the finished image must be repeated if I later pick up the original scan to change the image.
Comments