Previous: Resolution and resolving power< > Next: Color

Index of this page

• Test results
  • Per channel scan performance and noise

Per channel scan performance and noise

One aspect I didn't explore in the first public release of this test is the per-channel scan performance. Now with the new scans available of the overhauled Imacon 848, I did take an even closer look at the scans and discovered something new I hadn't be aware of before. The images below are crops of the full 8000 ppi resolution scans. No sharpening has been applied. Each series of 4 pictures per film type represent a crop of the original scans, and separated by color channel (Red, Green, Blue).

Kodak Ektar 100
Original	Red channel	Green channel	Blue channel

Especially noteworthy is the unexpected bad scan result in the blue channel of this scan, that is also visible in the Kodak Portra 160 VC scan series below, but not in the Fuji Velvia scan. The blue channel, disregarding the overal darker appearance (caused by the simple fact that we are looking a crop of skin tone, and skin does not have much "blue", hence low values in the blue channel which are rendered dark by Photoshop), clearly has a much more "grainy" and "patchy" look than the other two color channels.

Kodak Portra 160VC
Original	Red channel	Green channel	Blue channel

Similar results to Kodak Ektar 100, however even more grainy, since the film itself has more grain. Again, note the blue channel and the patchy grain therein. To make sure this observation isn't based on a misinterpretation caused by the differences in overall brightness and contrast of the channels, I have made a quick "match" of the Green and Blue channels by changing their brightness and contrast to approximately match each other. So I darkened up the Green channel to match the darker Blue channel, and lightened up the Blue channel to match the lighter Green channel. You can see the result below next to the original channel results:

Kodak Portra 160VC
Original Green channel	Blue channel matched to Green channel brightness	Green channel matched to Blue channel brightness	Original Blue channel

Notice that, although the differences appear slightly smaller now, there still is a readily perceptible more grainy appearance of the Blue channel, so it does indeed seem that this observation is justifiable. The Blue channel does perform worse.

Fuji Velvia 100
Original	Red channel	Green channel	Blue channel

Notice that, besides the overall finer grain of the Fuji Velvia film, that especially the Blue channel has much less grain in it. If you disregard the difference in overall brightness, it is hardly more grainy than the Green channel. Maybe still a bit, but nowhere near the extent to which the color negative film results are affected.

So, what is actually causing this issue? Why is the blue channel performance for the color negative films Ektar 100 and Portra 160VC so much worse than that of the Velvia 100 slide film?

Well, the answer might again lie in the nature of the color negative film. Not only has color negative film a limited contrast range that needs to be expanded during the scanning or wet printing process, but there is something else to take notice of. Look again at the difference between the color negative and slide film:

	Kodak Ektar 100 negative Limited range of colors in a color negative, only orange / magenta type.
	Fuji Velvia 100 slide Full range of colors in slide film

Notice the color of the color negative, well, what is it? Something like orange / pink / magenta? Hey, doesn't that remember me of something? It looks suspiciously like a filter needed to use tungsten film (a film designed to be used under artificial light conditions) under daylight conditions. Such a filter is an effective blue light blocking filter to compensate for the extraneous blue light sensitivity of the tungsten film...

That means that any blue light emitted by the scanners light source, is to some part effectively blocked by the color negative film. Now that small amount of blue light that manages to get through the darkest part of the negative, probably must be boosted electronically considerably to give any useful reading, leading to a more grainy scan result in this channel. Considering the very patchy result of the Blue channel, it seems likely that especially one of the dyes of the color layers in the color negative film is blocking the light.

Now of course, the extreme high density blacks of shadow areas in a slide / color positive film like Velvia are just as effective in blocking blue (or any other wavelength of) light. In fact, they probably block it even more... But there is one crucial difference with color negative film: in slide film, the areas of highest dye densities are shadow areas, so they are being translated in deep blacks and shadows in the scan. Any "grainy" look caused by excessive noise because of a low electronic reading will hardly be visible in a slide film scan, as it is hidden in the dark parts of the scan.

Now contrast this with the situation of a color negative film: the areas of mid to high dye density, effectively blocking blue light transmission, are coding for light to midtone values in the scan!

Well, a final look at the Alpha 900:

Sony Alpha 900
Original	Red channel	Green channel	Blue channel

Noteworthy is the fact that, contrary to the scans, all channels seem to be performing about equal, even the Blue channel has very fine detail. Especially noteworthy is the cheekbone. Do you notice the barely visible slightly darker and patchy, vertical line pattern there? I am referring to the lower right 1/4 corner. At first, I was thinking that there might be some sensor / camera issue here, causing the uneven result... but than I had a closer look at the original testchart, and YES, it was there! A barely visible, very faint black dotted offset screen printing pattern, that is amazingly registered by the Alpha 900 in it's Blue channel! Phew... who would expect that? Again a prove of how much detail is captured by the Alpha 900.

Previous: Resolution and resolving power< > Next: Color