Canon Scanner Desaturating Hot Pink?

[Makosuke]

Thought I'd throw this out just in case anyone had a suggestion. I've been scanning a bunch of CD case art obscure enough that I can't find a good one on the web, to tidy up my iTunes library. I have a Canon MP620 wireless all-in-one.

The scans have mostly come out very nice--I usually need a slight hue adjustment (+5) in Photoshop to get the color to match perfectly, but otherwise saturation and hue is fine across the spectrum.

...except hot pink. The couple of things I've scanned that have bright neon pink on the cover (flat areas of artwork) come out with the pink looking way desaturated. It's bad enough that it's impossible to correct for with simple hue/saturation controls.

Anybody have any idea why this might be? Is it a hardware issue, something in the scanner driver, or do I have something set wrong? (I've tried most combinations of auto/manual color adjust, and it doesn't make any difference, though manual usually comes out better.)

I'm viewing on a new 27" iMac screen that I've done a ColorSync calibration on (no hardware calibrator, though), so I'm assuming it's not an issue with my monitor or calibration on that end. The results are the same on a 17" MBP as well.

 

[Dreamail]

Two reasons I could think of:

a) The cover uses a special neon pink color ink

In this case the actual color only seems bright pink to the eye as the color ink itself converts ultraviolet light to the visual spectrum seemingly drawing more light in than is visible. That's how all bright 'neon colors' work.
The scanner light itself doesn't emit as much ultraviolet light so the pink color cannot draw from that source and what you end up with is only the visual spectrum reflection which is a lot duller.

If you have a digital camera you can test this:
Completely darken the room, open the scanner and use its light source to light the image - and take a picture with your camera. Is it the same dull pink? If so, the scanner is not to blame (well its light is as it doesn't emit enough ultraviolet light, or its sensor doesn't pick up this ultraviolet-shifted-to-visual-spectrum color).


b) Monitor spectrum incapable of displaying the color

Not all printable colors are equally well representable on a computer monitor. Typically the monitor color spectrum is smaller than the print color spectrum.
Certain pinks and certain intensive blues are often very hard to get 'right' onscreen compared to the printed image. Your pink might fall into that range.

 

[Makosuke]

Thanks for the suggestions. I did some follow-up testing based on them, and neither panned out, unfortunately:

A) I'm aware of inks (and apparently detergents--hence the "Whiter than white" tagline) that convert non-visible light into the visible spectrum. I'm not seeing how this would be any different than viewing the same image with the naked eye under lights that don't put out any UV--which to the best of my knowledge the CFLs I'm using don't--but even so I did test.

The booklet in question (and a similar-color pen) looked normal when illuminated by the scanner's light in an otherwise dark room, both to the naked eye and my camera. Similarly, the pink in question photographs just fine, under artificial light, with my camera as well as the built-in iMac iSight. It looks a little peaked with the 4-year-old iSight in my MBP, but then so does everything, and it's still much better than the scanner output.


B) Also something I'd thought of, but again the iMac's IPS screen is quite capable of displaying the same color in the pictures from my camera, as is my 17" MPB's screen as well as the camera's built-in screen and EVF. I can also reproduce it effectively enough in Photoshop with some work. So it's definitely not an inability of the monitor to reproduce the color.


Which leaves me back at the scanner. Based on other colors in the images with this particular hot pink, it appears to be confusing the scanner's built-in color management, causing it to turn red levels WAY down. What's odd is that it seems to do this even if I tell it not to do any automatic adjustment...

 

[Dreamail]

Thanks for doing the tests!

I still wonder whether it is a light frequency that the LED scanner elements don't see.

Or the opposite, as it is drawing so much more light intensity into the visible light spectrum perhaps some built-in gain control reduces intensity to not have it overexposed.


I wonder if this is related:
I usually have this behavior (way too dark colors) when trying to scan silver or gold foil text. Those usually come out black even though you would assume them to reflect all the light the scanner hits them with - i.e. they should scan almost white.
Either the light emitter and light detector are at an angle where it misses direct reflections. Or the sensor element reduces any super bright light so much that it becomes dark. This would be a built-in sensor hardware feature not a color correction software feature.

 

[Makosuke]

I'd noticed the reflective-is-black thing myself, but the reason is definitely because of the light-to-sensor angle; I tested to confirm by putting a round reflective object on the scanner, and the white stripe from the reflective light indeed shows up a little off center (looks like the detector is on the trailing edge of the moving bar). This is presumably necessary to avoid reflections when scanning somewhat reflective media, like glossy photos.

As for the hot pink problem, some further testing shows that cranking the saturation up in the +60-90 range gets it looking close (though of course that screws up all the other colors on the scan).

Weird. I'll have to try it on my dad's Epson and Brother scanners, to see if they do the same thing...

 

[Makosuke]

Ok, I did an additional test with a couple of other scanners:

On a new-ish lower-midrange Brother multifunction printer, the results were exactly the same as on my Canon multifunction--severely desaturated hot pink.

However, on a much older, upper-midrange Epson scanner (1240, I think), the pink looked perfect.

Now, I'm pretty sure the Epson uses a CFL backlight, while the Canon I assume uses RGB LEDs like the LIDE series, and I'm guessing (for no reason other than that it's lower-end and a rather thin unit) that the Brother does as well. Logically, I'd infer that LED-based scanners have problems with hot pink for some reason.

What's weird, though, is that obviously the source of illumination is capable of producing the color--the fact that both my eye and camera could see (and photograph) the pink correctly when illuminated only by the scanner's light proves it. Which makes me think that it's either a limitation of the sensor in these scanners or some sort of weird interaction between the RGB LEDs and the sensor(s). Or maybe Brother is buying their scanner parts from Canon or something, and there's a firmware oddity, though a quick Google search doesn't turn up any comments along those lines.

Man, that's weird. Not that it matters, but this is one of those times I really want to know what exactly is going on here.