Thoughts on color negative scanning

I have revisited this topic now that I am scanning with a digital camera. See Color film scan processing.

I’ve been thinking about color negative scanning for a quite a bit. I scan with a Nikon Coolscan V and typically shoot some form of Kodak Portra with a bit of Ektar thrown in. Obtaining satisfactory color from a color negative scan should be easy. Unfortunately, it’s not always straightforward.

My goal in all of this is to work efficiently in a minimum amount of programs and get the best color I can. A secondary goal is to be able to archive a master version of my scan. A raw scan has some distinct advantages over a corrected file in this respect.

The software

The options that I’ve seen for the Coolscan are Nikon Scan, Vuescan, and Silver Fast. Each offers some distinct advantages and disadvantages.

Nikon Scan

Nikon Scan comes with the scanner and usually delivers reasonably good color. It’s not been updated in ages and doesn’t always run correctly on newer Macs. Fortunately for me, it does run ok. However, it’s dog slow and doesn’t give you a lot of control over things. ICE is nice, but sometimes too heavy handed. Sometimes the color of your scan is just off, meaning you are going to spend some time in Photoshop anyway making basic color corrections, not just aesthetic ones.

Vuescan

A lot has been written about Vuescan. I’m not going to rehash it all. I use it for all my B&W scanning. I’ve found that out of the box, it’s color can leave a bit to be desired. I’ve recently spent some time trying to figure out how to get the best out of it. I’ve managed to get a lot better results from Vuescan with color, some of it based off of widely available tutorials for locking exposure to the base, etc. I’ll summarize that in a bit.

Vuescan’s big disadvantage is a confusing interface. However, it’s fast, works with a ton of scanners, and is constantly updated. It’s very easy to get a ‘raw’ scan out of it too.

Silverfast

Frankly, I don’t get Silverfast. Some people rave over it. It’s really expensive and has a poor interface, though it’s better than Vuescan in this department. It’s big advantages to me are it’s apparently decent profiling function, good handling of Kodachrome, and NegaFix. NegaFix is nice—it’s nothing ground breaking, but it does provide a decent interface for dealing with mask negation and color balance. However, there are several options which are poorly explained, and frankly, I’m not going to spend several hundred dollars to purchase Silverfast and the accompanying HDR program to deal with raw scan files. Thanks but no thanks.

A note about pro scanner software

Those of us who use ‘consumer’ scanners like Nikons, Epsons, Minoltas, etc., often have issues with color negative color correction. Those who use scanners that a step above, Flextights, drum scanners, etc., often tell us to just fix it in the scanner software. I’ve never used a scanner as nice as those, so I can only assume FlexColor and other programs for professional scanners do things right. Us little guys have to figure something else out…

All in the scanner software?

There’s something to be said for getting a flat scan with no corrections and doing the rest in Photoshop. Photoshop has a lot of controls, often easier to use than your scanner software. However, a flat scan isn’t always the best way to get ‘maximum’ data to Photoshop. You might be better off tweaking certain settings that control the scanner hardware itself to deliver a file with most data.

The orange mask

There’s a lot junk on the internet about how hard it is to compensate for the orange mask of color negative film, and that this is somehow the source of difficulty for color negative color correction. It’s not. It’s pretty easy. The orange mask is there for a reason too.

Removing the mask with Vuescan

Vuescan seems to do a good job of doing this by itself. You can help it along though if you have a scanner with individual control over the R, G, and B exposures. Vuescan already does a guess at this but you can fine tune it.

The procedure that I’m using (gleaned from the internet) is the following:

1. Preview your negative, making sure you have enough orange mask visible.
2. Check ‘Lock exposure’ in the ‘Input’ tab.
3. Preview again.
4. Check ‘Lock film base color’ in the ‘Input’ tab.
5. Go to the ‘Color’ tab and record the film base values.
6. Assuming the ‘film base red’ number is the highest (it should be), divide it by ‘film base green’. This is your green analog gain value. Set it in the ‘Input’ tab.
7. Do the same for blue.
8. Uncheck ‘Lock film base color’.
9. Make your scan.

This setting should be good for the whole roll. It should also be good for other films of the same or similar types. I’m not going to be a stickler and make sure it’s perfect for EVERY roll. Being in the ball park let’s you ensure your film’s content fits in the scanner’s dynamic range, channel by channel. I don’t think being exact buys you much over that.

This method should also have the advantage of more or less canceling out the orange mask in the raw scan. It’s essentially like filtering out the orange mask base layer with a filter pack in darkroom printing. We’ll get the last little bit of it later in Photoshop.

Dealing with the raw scan file

If it’s not clear yet, I’m saving these as raw scans from Vuescan. If you get a good raw scan, you can do any of the following techniques at a later time. That’s the power. At the very worst, you can do the first method and deal with the file in Vuescan and generate a corrected file; essentially the same as if you skipped the raw step to begin with. However, the first Photoshop method is my preferred one.

Inverting in Photoshop

A detailed explanation of this is method is located at Inverting raw scans.

As I stated earlier, when I first tried this I was met with failure. That’s because I was trying to deal with a linear raw scan, and not a gamma corrected scan. Gamma correcting the scan using levels or curves leads to issues.

After happening upon negfix7 and reading some interesting documents on the CF Systems website and, I decided to make my own inversion routine using ImageMagick. I spent a bit of time tweaking this script and writing some Photoshop actions to finish off color correction. I was pretty happy with my results and was waiting to write them up when I started to investigate other options, which led me to the process I’m about to describe. I might put up the script for completeness’s sake at some point, but not for now. For now though, look at negfix7 or negfix8. These scripts get the inversion out of the way and the bulk of the color balance. They can be slow on large scans.

The key to the Photoshop technique is to assign your raw scan a profile that has its gamma set to 1.0 and then convert it to its counterpart profile with proper gamma. Let color management do the gamma conversion for you. I made gamma 1.0 counterpart profiles for AdobeRGB and BruceRGB, but you could easily do it for sRGB or ProPhoto if you wanted to. Without this step, the inversion never looks right and you fight with trying to fix the gamma of the image with inadequate tools (curves and levels). The rest of the steps are straightforward and easily automated.

Refer to Inverting raw scans for the detailed explanation and Photoshop actions.

If you want to scan your negative as a positive, as long as your scanning software converts the linear scan data to a reasonable working gamma (2.2, etc.), then you should be able to invert and color correct in Photoshop with no problems.

The Vuescan raw workflow

I don’t foresee using this method much myself, but I did look into it to see if it worked for me. It works pretty well, but is a bit too fiddly for me. I was also never totally satisfied with some of the colors I got.

The basic idea is to run Vuescan with no scanner attached, open your raw file in Vuescan, setting the media to ‘color negative’, and select ‘Scan to file’. Hitting the preview button will generate your preview. I find it is helpful to periodically toggle on ‘Graph image’ to see what the RGB histogram is doing with the corrections.

You have several options for color correction. The most straight forward is to go to the ‘Color’ tab, select ‘Manual’ for the balance, and right click on a neutral part of your image. This will provide values for the ‘Neutral red/blue/green’ settings. You can modify them manually if you choose.

Do not be afraid to use the ‘Brightness’ slider here. Adjust it so your histogram isn’t slammed up against either side and so that the image looks decent. If there is a color cast, you can also move the individual RGB brightness sliders. The brightness sliders act as gamma changes on the overall image or the individual channels.

At this point, you can stop. Or you can move into more manual corrections. I’m not sure if it’s worth it, but here goes. Back in the ‘Input’ tab, select ‘Lock film base color’. Once you do that, ‘Lock image color’ appears as an option; check it.

Moving back to the ‘Color’ tab, we now have the ability to manual control individual white and black RGB points. If this is useful, go at it. I’m not convinced that this is the best place to do it though…

Through all of this, I keep the film settings to ‘generic color negative’. I don’t necessarily think any of the specific settings are a good idea. Mainly because it’s unclear what they actually do. However, if you like using them, feel free too.

ColorPerfect

I’m not going to into too much detail here on ColorPerfect. It used to be called ColorNeg. It’s not that expensive and I was able to obtain pretty good color out of it. It has a horrible interface (notice a trend here?). Though the final color was great, sometimes there was a funny look to it and it was a bit too saturated in odd places.

The basic usage is simple. Bring your linear raw scan into Photoshop and run the plug in. Select your film if you desire. Clicking on a neutral point will help set your white balance. At this point you can send it back to Photoshop for further tweaking. However, there are some tools in ColorPerfect that can help you with difficult negatives; unfortunately they are a bit obtuse in use. It can be advantageous to tweak the gamma and black points in the plugin as well as messing around with the highlight preservation feature.

Notes

‘Raw’ scans?

Most scanner software provides the ability to output a ‘raw’ scan. This is different from digital camera RAW files. Scanners don’t have Bayer arrays, so each pixel has a real recorded R, G, and B value with no interpolation going on. RAW development programs do a lot of things, but the real ‘development’ step is in interpolating the Bayer array data into a full color image. 99% of the rest of the things that programs like this can do could be done on any image data. That’s why Camera Raw, Lightroom, Aperture, etc. let you bring in tiff or jpeg files to edit.

A scanner raw file is just the data right off the sensor. It hasn’t been gamma corrected; it’s linear data with a gamma of 1.0. If the scan is from a negative, it also will be inverted.

I was initially told that you could work with raw scan files fine in Photoshop. If any of you have tried, you probably experienced the same thing as I did—it doesn’t quite work right. While there’s nothing wrong with the idea, if done incorrectly (like I did back in the day), it’s not going to be a good experience.

DNG files?

No.

There’s no real reason to use a DNG file for your scan. Lightroom and Camera Raw can edit TIFF files just fine. So can most other programs. Keep your scans as TIFFs. If you want to edit them in Camera Raw, either access them through Bridge, or go to File>Open in Photoshop, select your file, and choose ‘Camera Raw’ as the type. After you save Camera Raw settings in your TIFF, it will default to opening in Camera Raw in the future. You can change this behavior by accessing the Camera Raw settings.

Lightroom/Camera Raw

Again, I’m going to say no. They are great tools and allow for non-destructive editing. Unfortunately, they don’t offer full RGB curve functionality. While they do let you white balance easily, the corrections are some what limited. The white balancing tool does shift the RGB curves relative to each other, which is good, but finer control is more difficult to achieve. It is more appropriate if you have an already inverted and somewhat color corrected scan (like that which is obtained from the negfix script, but honestly, I’d rather use Photoshop.

I find that trying to avoid Photoshop can sometimes make more work. Of course if you are happy with these programs, stick with them. They are useful tools in your arsenal. I prefer to use them for specific tasks. In the negfix, processing a 4000 dpi scan can take a good deal of time. It’s usually much faster for me to just pop it open in Photoshop and run a predefined action.

Dust & scratch removal

But, but, but, if we use raw scans, we can’t use ICE or any of the other automatic dust removal routines! Correct. However, I save my raw scans as RGBI files. The infrared channel is there waiting to be processed. I have a set of actions that process this and enable easy dust correction in Photoshop. All of this is detailed on the Dust & scratch removal page.