A few scanning tips


Processing Scanned Color Negatives

Color negatives are a special problem to remove the orange mask. Film scanners scan color negatives with a longer exposure time of the blue and green channels. This acts as an analog glass filter at the lens, and the longer exposure boosts the blue and green components and reduces the orange (as an analog operation, similar to a glass filter). Otherwise, if not filtered, then when inverted, the color negative mask appears near deep blue (the complement of light orange) instead of the correct colors.

Digital cameras with a macro lens can copy positive slides very well, but frankly, the best color negative work will be to use a real film scanner (for this reason). This is NOT saying you cannot post process the reversal of color negatives. People do it and some seem happy enough. There are inexpensive film scanners now using camera sensors (instead of carriage motors). That works for positive slides, but it is "make do" for color negatives, because the problems are serious, and a real film scanner is designed to avoid the issues.

This article is not intended as a tutorial about processing color negatives, as much as an explanation of why (if available) the film scanner can do it better.

B&W negatives are easy with the camera, just invert them and you're done (at a menu like Image - Adjustments - Invert). Except adding contrast does improve B&W photos significantly, so after simple inversion, in the Levels tool, move the histogram White Point down to near where the image data ends, and move the Histogram Black Point up to near where the image data starts (very much like shown below, except B&W only has the one grayscale channel). This increases contrast (blacker blacks, whiter whites) by filling the full range with actual image data (See more detail of an easy way).

Ansel Adams in his day did not use digital histograms, it was B&W work done in a darkroom enlarger. But one of his main rules was that something in the image needs to be solid black, and something needs to be pure white... which is contrast. Deep contrast may be less appealing in color prints (and clipping changes color values, but grayscale needs help with contrast, and thrives on a little clipping), but speaking of black and white photo prints, he said:

A print should exhibit a complete range of tones, as well as the extremes - white and black - in order to have strength and conviction. The amounts of pure white and solid black may be very small, but in almost every case, either one of both of these extremes should appear in order to "key" the values.

Ansel Adams, his book The Print, page 7, 1950.

But color negatives have the orange mask all over. This becomes deep blue when inverted to positive. It is a lot to deal with in digital. The orange mask is designed to help the filtering in the darkroom enlarger for the regular color print paper emulsion. Dark room printing and also film scanners handle this at the analog level, but digital edits do it as digital, which becomes 8-bit 255 level.

Light is analog. Regular analog color printing removes the orange with an orange filter (Magenta and Yellow) on the enlarger light. Real film scanners remove it (in Color Negative Mode) by individually varying the exposure (time duration) of the three RGB channels. For color negatives, the blue channel is exposed about 3.5x longer (than red), and the green channel is exposed about 2.5x longer (than the red channel). This simulates an analog optical filter. The important point is, both of these conventional methods to remove the orange are done with analog light, which has no limits (does not suffer clipping). Different film brands might vary the orange color a little, but the longer exposure boosts the blue and green, acting as if an analog filter were on the lens, holding back the orange complement.

Whereas camera sensor chips cannot do that exposure RGB variation. Digital color extreme shifting clips at digital 255 and 0. After the scanner or camera has digitized it, then the 255 end is a hard limit. If we shift the data very much then, the data just falls off of the 255 end (remains clipped at 255, which changes the color somewhat).

Other ideas for experimenting: Incandescent tungsten light is orange, which varies, but maybe around 3200K temperature, more or less. Using incandescent tungsten light to illuminate the film copy is just more orange, aggravating the orange mask problem. Assuming the copy is using sunlight or flash, what could be a partial correction is that the blue camera filters 80A (3200K, needs 2 EV more exposure) and 80B (3400K, needs 1.6 EV more exposure) were used on film cameras to correct incandescent lighting to daylight color film. Or a cameras custom White Balance setting of around 3200K or 3400K (with sunlight or flash illumination) could do the same job. Incandescent orange is not the same as the color negative mask orange, but both are strong similar orange. White Balance degrees K addresses only yellow vs. blue, but the film base has some Magenta in it too, which will also need added Green correction. So only partial correction, but this use would be a strong head start towards correcting the orange mask color (but it would then rule out other methods addressing the color film orange mask). If you have a color negative frame containing a 18% gray card or a standard White Balance card, it would be very helpful for determining a proper color conversion process (its final color should be neutral, with equal RGB components). Note that AFTER inversion, this mask is deep blue, not orange.

With the camera, you conceivably could add similar color printing filters on the lens or flash head, and then do simple invert. Not exact same filters, because enlarger was incandescent light (already orange), but the flash is daylight (white). Also different brands of film may vary the orange mask a little. Or if you might have access to an old enlarger color head (like Omega) - a light source with the dichroic filters, and three CMY dials to set the color of the light output - using this as the background light source could be a proper analog correction, and would be similar to a scanners corrections, but could still allow camera speed rate. Digital can handle slight shifts, but this big orange operation is best done while still analog (and not limited at 255). For example, we learn that even a Daylight white balance close up of a red rose will likely clip the red channel.

This is not to say you cannot use the photo editor for color negatives. It may come out good enough acceptable or usable, but the range is less, and really, it's just not the same as a film scanner... and certainly not as easy, and sometimes still a problem. Some will always disagree, claiming the editor "works for me". But a film scanner removes all doubt and eases much of the correction work (proper white balance as always is still a remaining issue).

For example, the Photoshop Curve tool even has a Color Negative preset, showing the inversion:

You could duplicate this curve in any curve tool.   Note that it specifies that the Blue, Green, and Red channel input values will get clipped at roughly 40%, 50%, and 75% input (with zero output past those points). The added blue and pink clipping frame is the same flower image as below, showing the clipping preview of those colors at those locations. It may remove the orange, but IMO, such clipping is not the most desirable way to do it. It is hard to get the colors right sometimes.

Inversion with Curve tools: Just grab the lower left end of the curve and pull it to the top, then pull the upper right end down to the bottom (as this example did). Then the RGB slopes run downward at 45 degrees, which inverts the image. Input at zero (black) is output at 100% (white), and input at 100% (white) is output at zero (black). This is the standard inversion technique. Inverts colors to their complement too, yellow becomes blue. Numerically, all RGB components become 255-previous (255-255 = 0, 255-0 = 255), which is inverted. See more about the Curve Tool.

The inverted curve makes the negative image become positive, and now the orange cast becomes blue (all colors become their complement color). The channel high ends were trimmed drastically to balance the colors (remove the orange, which will be inverted to blue). The clipping inverts to the positive zero end (loss of shadow detail). The curve numbers clearly show that the Red channel is clipped at 182, the Green at 132, and the Blue channel at 106 (before inversion). Any higher input values have the output clipped at zero (inversion). The "Show Clipping" preview feature shows only the pixels that are clipped (255 inverted to zero). However, you can select the Curves individual RGB channels, and move each bottom White Point (inverted to Black Point) to the right, just past clipping, eliminating all clipping. But this does completely change the color of orange that you hope to remove.

It's a good try, maybe dramatic, but a tough job digitally. Film scanners could do it that way too, but this clipping loss is why film scanners instead increase the analog blue and green channel scan timing (filters the orange color, as an analog operation, without digital clipping loss). So yes, there are these software options to do the orange reduction, but you may not always like the results. That effect is NOT what put film scanners out of business. :) This could be "good enough" for some things sometimes, but if you're more critical, there is the choice to use a real film scanner for color negatives. Analog correction is superior this time (has full range).

Here are some samples, some other method options. Three methods, which are all just dumb automation. Results will depend on individual images. I am not recommending these tools, they are simply available. None are necessarily correct color, but just are just tries for a more reasonable probable situation for the color channels.

The color negative image, orange

Using the Photoshop Curve
"Color Negative" Preset above

Then Levels histogram now.
Clipping inverted to 0 end.
Severe Loss of low end detail.

Other Tools

Instead, simply inverted, blue now

Then Photoshop menu Image - Auto Tone

Levels Histogram then.
Clipping inverted to 0 end

Or after Invert, and then in Photoshop Levels, Options, the Per Channel setting

Which automatically and tightly sets end points of the three individual channels, as shown, to remove end gaps and include only the actual data range. This is an older type of White Balance method, equalizing the top end of the bright data in all three channels (clipping a low level threshold specified in the tools dialog box shown), to each reach to 255 in all channels (which creates white, assuming there should be some white). Very nearly the same action as Auto Tones, and showing the histogram settings here was really the only reason to show it again.

Note: Changing the three individual RGB channels separately (as here) changes color. Each color has a RGB component, and these individual changes affect that resulting color. But after converting B&W to Grayscale, there is just one channel, and then similarly adjusting endpoints tightly to where the data starts increases contrast, no colors to be affected. Try it, you may like it.

That Levels result

This result is more like the first Curve, if as subsequently done here, we back out gamma (Levels center slider to 0.45, which is inverse of 2.2). I don't follow that logic (Gamma in the Curve is just brightness), it only appears true. I do prefer the less extreme variation.

Another method is to invert the color negative, and then use a White Balance tool to click in the unexposed border around the frames (was orange, but blue after inversion) to neutralize that color. Which is about the actual color cast present, and it ought to work best, but for me, only ballpark, not real close. Clicking something that was actually white in the scene may work better. But if it is already digitized, then clipping at 255 probably will be an issue.

Color negatives are a special problem, and a real film scanner will be best. But slides are quite easy with the digital camera and a good macro lens.

Menu of the other photo pages here

Copyright © 2007-2021 by Wayne Fulton - All rights are reserved.

Previous Main