Combining Multiple Exposures with SilverFast HDR®
I had the good fortune to obtain a copy of SilverFast Ai with my Epson scanner, and I mostly use it to scan medium-format color negatives. I spent a long time frustrated by the lack of control over the histogram while using SilverFast’s NegaFix utility, and originally wrote this page to accompany this thread in the SilverFast Forums. The official SilverFast documentation is so sparse for an application of this complexity, that one seriously wonders if even its developers fully understand it. (I even bought the little book.)
I have since, with a lot of online help such as Ian Lyons’ invaluable tutorials, managed to hammer out a way to create an image that maximizes the full potential of the negative while making use of the color adjustment provided by NegaFix. In short, one combines a number of scans at different “exposures” to create a full-scale image. The concept seems to be that a negative can record a greater range of values (the blue curve below) than are reproduced in the “normal” scale of a print, and simply compressing the values to fit them all into the final image produces a low-contrast, unexciting result. NegaFix is constrained to “expose” the negative within a fixed dynamic range (the red curves below), and there doesn’t seem to be anything that you or I can do about it; but this range can be moved up and down the tonal scale to produce darker or lighter images, which can then be combined in various ways to make the final print.
The simplest way to combine the images is just to stack them. In Photoshop, paste each image into its own layer, then set the opacities as follows: 100% for the background, 50% (1/2) for the next layer, 33% (1/3) for the next, 25% (1/4) for the next, 20% (1/5) for the next, and so on. Utilities such as Photomatix can do this automatically. The result is likely to be a slight improvement on the original image. More esoteric ways of doing it involve adaptive contrast and tone mapping; Photoshop CS2 and above can do this, as well as applications such as Photomatix and OptiPix; but to me the results are seldom an improvement. (Yes, you can get it to work; no, it doesn’t look very good. If you are curious, get a utility like EXIFutils and assign fictional exposure values to the images.)
The way I do it is to make exposures optimized for different areas of the image, such as highlights, midtones, and shadows, and then stack them; but I use layer masks to “paint” different areas of the image with different exposures. To create a layer mask, first select the relevant layer; click the square “Add layer mask” button at the bottom of the layers palette; click the layer mask thumbnail if necessary to make it active; then paint the image with black to hide the layer, white to reveal it, or gray for partial visibility.
Below is a photo taken from the north rim of the Grand Canyon. I wanted to lighten the foreground, and the dark area to the left in particular; so I made three exposures with SilverFast and masked them in Photoshop with the masks shown above. Click here to switch the images (JavaScript requred).
The most efficient way of making multiple exposures is to use SilverFast HDR. One first makes an HDR (high dynamic range) scan using the scanning software; this file represents the raw output of the scanner, and minimal adjustments are made at this point. Ideally, the scan need never be repeated. SilverFast HDR is then used to process the HDR file, converting it into a conventional image the way the scanning software normally would. It is at this point that NegaFix, brightness and contrast adjustments, and so on are applied. It is then a simple matter to process the image several times at different exposures.
If all you intend to do with the scanning software is make HDR files, the “entry-level” SilverFast SE is all you will need. SilverFast SE Plus and Ai Studio have multisampling and multi-exposure features, but I have not found them to be of practical benefit with negatives. The main advantage of the HDR Studio version is the ability to calibrate your scanner using an IT8 target; it works very well should you not already have an application that does this.
I do not intend to offer a comprehensive SilverFast tutorial here. These are just a few observations that are not addressed anywhere in the SilverFast documentation. When making the HDR scan of a negative, SilverFast can be set to scan the image as either a negative or a positive. The resulting image is a negative in either case; the important difference is that when scanning as a negative, NegaFix is active, and the histogram is expanded to fill a wider range of values. When scanning as a positive, the histogram is copied as-is, likely in compressed form. When processing the positive-scanned image, the final expanded histogram tends to look a little bit “combed,” indicating some data loss; a negative-scanned histogram is smoother, indicating that it is better to let the hardware and scanning software expand the histogram during capture.
Another difference between scanning as a negative vs. a positive is that in the latter case, a scanner transparency profile can be applied. The option is not otherwise available. This makes little actual difference in the final result.
I prefer to capture as much scanner data with as little modification as possible; so I scan as a negative, set NegaFix to linear, and do not apply color cast removal. The only adjustment made at this point is histogram expansion. Below are some screen shots. (In reality, even though they change the appearance of the scan in the SilverFast window, the NegaFix settings at this stage make surprisingly little difference in the final result. The HDR image need not look good; its data is stored in a different format from that of a conventional image file.)
1. The scan: I set NegaFix to <Other> <Other> <Linear>, and uncheck CCR (color cast removal).
2. Processing the HDR file: set NegaFix to your scanner and film type, and check CCR. In the example below, I have set the exposure to -1.0 stop.
