Colors while scanning a cross-processed film: why [not] so vivid?

Question

In this question I am trying to understand the nature of colors in scanning cross-processed films. I've thought that I'm okay with photoshop, with subtractive and additive colors etc etc etc, but somehow this is (at least :) one thing that I can't get. So,

First image is the result of the lab scanner (some kinda expensive Kodak film scanner). The colors are simply great: vivid and orthogonal. The contrast is so-so, that's why I've tried to re-scan the image with my home Epson V700 scanner and standard Epson Scan software. You can see the result on image N2. The details are finer, but the colors are way pale. I bet this is not the saturation only, since there is much less blue then on the lab scan. I've tried to recreate the lab results in photoshop (image N3), but with a very limited success: the blue is still not there.

I know that I use the simplest program, maybe some more sophisticated software would help me to get the results, but my question is more theoretical: what happens in the software of expensive scanners, what is that function that allows to get those way different colors? Why the lab scanner is capable to reproduce the colors so vividly? What can I do (in Epson Scan or in Photoshop) to get closer to the results from the lab?

Appreciate!

Update: No, simply adding the blue/cyan filter and darkening the exposure doesn't provide satisfying results. First, the brightest sky remain pale/white, second yellows will be dramatically affected.

Update N2: Tried to scan with different exposure setting. Still the colors are not even similar to the lab scan (image N4), even after photosho (image N5).

Answer

A great deal of what's going on here is really just color management/matching/profiling.

Different films have rather specific "looks", and to get the best out of each, you just about have to profile the scanner with each film. That was common/typical with the high-end professional scanners. If memory serves, some even scanned the bar code that's on the edge of most film so they could automatically identify the film and use the correct profile for it.

Most film scanners sold for consumer use were rather a different story -- quite a few didn't provide any profiles at all, leaving it up to the user to match colors as well as they could by eye. The few I saw that did include profiles were almost worse: they included one profile for the scanner in general, with no attempt at taking into account the film being scanned. Most of these did a halfway decent job on a few of the most popular mainstream films, but otherwise they were generally pretty bad (and the exceptions I saw to that seemed to be pretty bad at everything).

At least in theory, I suppose you could do a profile yourself, in much the same way you'd profile a digital camera: take a picture of something like a Macbeth color checker with the film in question, scan it through the scanner, and figure out the adjustments necessary to get the original colors.

I'm not at all sure how much good that'll do for cross-processing though -- here you're going for a particular effect that definitely does not include the colors being entirely accurate. From the look of your pictures, I'm guessing that you're running E6 film through C41 processing. If that's the case, a really good E6 profile may be sufficient to the job -- but I've done very little with scanning cross-processed film, so that's only a guess.

Check more discussion of this question.