Vuescan negative scanning - clipping shadows & highlights

Hi all,

I have been working with raw files of negatives I scanned many years ago in Vuescan (on a Canon 9950f flatbed scanner). If I look at the raw files as 'image' files they look like negatives do - inverted with the film base color. When I select 'color negative' as the source Vuescan does the right thing and inverts the colors and corrects for the film base. However, when Vuescan processes the negative, it clips the highlights and shadows. This is regardless of the fact I have the black point and white point set at zero. It is subtle for most images, but it is definitely there and is a problem for any under- or over-exposed images.

When I look at the raw file as an 'image' with the eyedropper tool I can see the values for the various areas and none is showing as

'000' or '255'. In other words, the raw image isn't clipped - this isn't something I got wrong when I performed the original scan. Rather, it appears that Vuescan is clipping the highlights and shadows as it processes the negative. I have played around with color balance and brightness settings but anything which is clipped remains clipped, which leads to some really ugly results if I am working with a negative which wasn't well exposed to begin with?

I am processing the files using the 'generic color negative', as I tend to find most of the film profiles produce unrealistic results. I do recall when working with b/w negatives that the film type did have a small impact on the amount of clipping, but haven't found a color film type which retains all the information.

Has anyone else seen a similar behavior? What if anything can I do to fix this, other than file a bug report?

Thanks in advance,

JJ

Nearly all scanner software does the same.

Use an image editor to invert the files and then levels or curves to align the RGB histograms and adjust the colour balance. IME that's the only way to get a really good positive image from colour negatives. Tedious, but it's like the difference between an automated mini-print, and a hand-produced colour print by a skilled printer.

• Using an image processor (e.g., Photoshop), first crop the negative image to remove the unexposed borders. The orange mask is strongest where unexposed and weakest where there is an image
  
• Do basic color correction in the negative state to remove the orange mask. One way is to use Levels (histogram) to set limits to the high and low tails of each color separately
  
• Invert the colors (ctl-I) and fine tune the color in the positive state.
  

This process can be automated, following cropping as in the first step, using the color processing stage of scanning software. I use SilverFast HD, which is otherwise used to process "RAW" images (i.e., unprocessed) from a scanner. SilverFast has the advantage of profiles for many types of film, which vary in the density of their dyes.

SilverFast has the advantage of profiles for many types of film, which vary in the density of their dyes.

Vuescan has the same facility.

Also, if you have a true RAW image that can be opened in ACR, CaptureOne or the like, then you can use the Colour Temperature and Hue (green/magenta) adjustments to neutralise the orange contrast mask by selecting an area of clear mask as a neutral white balance.

first crop the negative image to remove the unexposed borders.

Indeed. The clear mask is effectively 'blacker than black' in the positive, and will upset any auto-levels or auto-colour adjustments. It needs to be cropped away before using those, but a clear border, or the space between frames, is useful to get a sample of the unexposed mask.

It is the orange border on the edges and between frames which must be cropped out of the image. These areas are utterly useless as a test area for color or exposure.

RAW images, as applied to scanning, have nothing to do with ACR. They are simply as scanned, completely unprocessed. You could of course "scan" with a digital camera to a RAW camera format, but once processed, you would have a negative color image which requires further processing to a positive image. It is at that point I apply cropping and Silverfast HDR.

You can process an image without cropping, but the color and exposure are off, and exceedingly hard to correct. The film must be well exposed too, because over or under exposure has a profound effect on color balance.

If a camera is used for scanning, you must use the white balance based on the empty holder, diffuser or light source.

This is an image taken about 20 years ago, converted from a camera "scan." The magenta cast is resistant to ordinary color balance, much as reversal film behaves under overcast sky. It doesn't help to be at an altitude of nearly 10,000 feet.

I am a big fan of Vuescan, especially since it works with so many of my old scanners that are no longer supported by their manufacturers.

However, in the case of the Canoscan 9000F and the 9950F (I have both, for reasons ... , well, I have both) I do like the original Canon software also. If they will still run on your OS, I'd give them a look.

If a camera is used for scanning, you must use the white balance based on the empty holder, diffuser or light source.

Absolutely not!

My first response to this thread explains in detail the advantage of compensating for the orange mask. It's to prevent cramped histograms that must be aligned and their slopes equalised to get anywhere near to decent colour. Either by scanner software or in post processing.

Think about what happens in chemical colour printing. The mask is compensated for by the paper sensitivity - very blue sensitive - and the filter 'pack' added to the printing light. That's what needs to be emulated during scanning.

If a camera RAW image really was raw, there would be less issue, but every raw processor automatically adds a steep tone curve to the overly green, dark and low-contrast image got direct from the sensor. Such that if the cyan, yellow and magenta film dye images aren't brought into alignment before they get onto the sensor or are processed, that steep tone curve needs to be reversed, part-by-part, before the inverted red, green and blue channels can be matched.

Of course a dedicated film scanner does all this in hardware and software - hopefully. Even so, most scanners are designed to scan reversal film, with negatives a secondary consideration, and a 'raw' scan may not be as raw as could be desired.

I've scanned several hundred negatives using the method I described, with little or no additional color correction required. The need to crop away border areas was derived by trial and error, and is the best way to obtain consistent results with a dedicated scanner or with a digital camera.

It is likely that commercial printers determine exposure and color balance from a selected area within the image. Darkroom printing requires a great deal of trial-and-error before deciding on the best filter stack, and even then fine-tuning is required for best results.

Darkroom printing requires a great deal of trial-and-error before deciding on the best filter stack, and even then fine-tuning is required for best results.

I know, I did it for 40 years!

So why should producing a digital image from a film negative require any less skill and attention?

Use of a colour analyser in the darkroom only gets you so far, it doesn't completely automate the process. Likewise with scanner software or Auto Color and the like. The only tool that's totally reliable is the Mk 1 eyeball, coupled with a good familiarity with the Curves and Hue-Saturation tools.