ADJUSTING LEVELS IN PHOTOSHOP VS NIKON SCAN

The popular advice given for scanning is to make as many of the necessary adjustments in the scanning software as possible. This is due to the belief that the levels adjustments in the scanning software tie into the analog controls of the scanner. The following test was performed to determine if these perceptions were in fact true.

The methodology involved scanning the same slide twice in 14 bit mode, sRGB color space, and 4000 dpi resolution on a Nikon Super Coolscan 4000. One scan had it's levels adjusted in Nikon Scan and the other scan had identical adjustments made in Photoshop. These adjustments involve setting the white point to 50 and the gamma value to 3.0. The images are then resized and saved out as high quality (Q = 90) jpgs in Photoshop to make them viewable on the web. This system is not perfect due to the compression of the images and other factors but I believe it accurately shows the effects of making levels adjustments in Photoshop vs Nikon Scan.

Surprisingly, these test images show the image adjusted in Nikon Scan to have lower contrast, more posterization, and greater noise than the image adjusted in Photoshop. Wondering if this could be attributed to a mismatch between the gamma and white point scales in Nikon Scan and Photoshop, the experiment was repeated with the white points and gamma at different levels in Nikon Scan. However, no settings in Nikon Scan could be found to achieve an image that matched the quality of the image adjusted in Photoshop. Another guess was that the differences may have been due to Photoshop and Nikon Scan applying the curves at different resolutions (16 bit vs 14 bit) but an email from Scott Robertson, suggested a more likely answer: The difference was a result of Photoshop "slope-limiting" the gamma adjustment while Nikon Scan applied a normal gamma curve. "Slope-limiting" is a feature used by Photoshop to limit the gaining of the values (and thus noise) in the darkest tones. It is done by applying a curve that has a small linear segment (with a limited value) near D-Max. There is no documentation that could be found to indicate that Nikon Scan uses "slope-limiting" so assuming it does not, the extreme adjustments made by applying a 3.0 gamma curve on the darkest tones of the Nikon Scan image would explain the differences we see between the two methods.

CONCLUSION: Contrary to expectations, making adjustments in Nikon Scan was shown to provide worse results than scanning images at maximum bit depth and adjusting them in Photoshop. Also, there was no evidence to suggest that making levels adjustments in the scanning software affects the scanner in the analog domain. It would have to be concluded from these results that the advice commonly circulated by people and even websites set up to offer scanning tutorials, is incorrect. These results and conclusions are also consistent with statements made by Ed Hamrick who developed Vuescan.

It should be noted though, that because this is such an extreme example, most images will not show the differences (if any) to be as dramatic as those seen here. For the sake of working from a scan with the least amount of post-processing applied to it though, it is still recommended that all levels adjustments be made in Photoshop.