Bacardi

This is a serriouss thread.

I use an Epson Perfection 2450 screecher er...scanner (never ever I will buy Epson again!!!), an Iiyama Visionmaster Pro454 crt and a Canon i950 printer (6 colours).
I use Silverfast as scanner software.

I managed to get everything profiled (I did look often at that pic you added to that famous chapter of Inside Photoshop, and worked my way through it). I mean: what I see on the scanner preview in Silverfast is what I get inside Photoshop and the print resembles quite closely what I see. (the printer needs a bit more Yellow, but that I know).

This works fine when I scan at 100%, 300ppi. But when I scan at 72 ppi and reduce the size, the final scan in Photoshop (still 7.01) is much, much darker.

Any idea what this might be?
Silverfast is great, but very complicated.
Too complicated in fact. Something like Photoshop.
But compared to it, the Epson software is something like Paint.

I would love some ind of discussion/exchange of views etc on calibration etc. Really cutting down to the marrow.
If there is interest for this subject of course.

Gametamer

It sounds like a profiling problem, Erik--although I have no idea why a 72ppi scan is so much darker than an equivalent 300ppi scan.

Try adjusting Adobe Gamma from Windows' Control Panel; now remember that this wonderful l'il utility only works in Adobe applications, and if you view work in two different apps, one of them being non-Adobe, you most likely will see a darker image.

Also, if you scanning hardware is calibrated for the Mac, and many scanners are, you might try View, Proof Setup, Macintosh RGB, to get a lighter view of your scan. To swap profiles, go to Edit, Convert to Profile, and then choose a setting (I ndo not use the sRGB color space, but this sadly is the default), but instead the Adobe RGB (1998), which is a wider space.

And if you're color-correcting/gamma correcting on a one-by-one basis, which would be my recommendation, try converting the image to 16 bits per channel (Image, Mode), then perform your Levels or Curves adjustments. Then pop the image back to 8 bits/channel, and save. The advantage to doing this is you lose less tones (editing in a "super" color space provides more tone reassignment area, less original image loss).

Does any of this help?

My Best,

Gare

Bacardi

Yes, it does. It confirms that I do what should be done.
I guess I'll have to ask over at Silverfast though as it looks like a problem in their software.

My monitor is rather well calibrated, following your instructions in Inside Photoshop. I read those theory chapters over and over again. And it's a rather good one also.

-I open Photoshop, import>Silverfast Epson.
-In Silverfast, I unchecked Preview Draft.
-I check with the original, knowing that the light colours of a monitor can never be exactly what I did with watercolour. But It comes quite close. I do a little curve tweaking in Silverfast untill my prescan matches as good as possible my original.
-I place the original back and scan at 100%, 300ppi
-The image opens in PS and when I look at it side by side with the review, they are as good as identical.
-I scan a second time, but this time for the web. Therefore I reduce the size to some 450 pixels wide.
-This scan opens in Photoshop and is much darker. When I compare the histograms in Silverfast and PS, the PS one contains much more threequarter greys. And imo, it should be identical.

Shouldn't it?

The only difference is that I changed some settings in Silverfast. So that's where, very probably, the problem lies. So I'll try to get an answer over there and keep you informed.

BTW: I always scan in 48 bit into Photoshop and immediately save this file. In case it's for eternity (cough), I scan in LAB mode. The reason I prefer to use curves etc in Silverfast is because it can work with the original material because the final scan comes after the corrections. For print, I do use AdobeRGB.

BTW: do you know how those six colours (CMYK, PhotoCyan and PhotoMagenta, lighter versions of-) work with the colorspaces?

I mean: it's the printer software that has to do the translation from RGB to CMYKpCpM and imo this will work correctly if and only if the software has been coded to work with a specific space. Probably sRGB or even worse, as the printer is sold to a general public.
Otoh: I could compare results with a big Epson (A1 size, 6 colours) at a printer's with whom I work from time to time. The Epson is calibrated for exact work for offset (cough...six inks, and the presses only have four...) and I must say that my own results really matched that pro printer's result. So I'm happy with it.

OK. I'll try again one last time, and then I'll contact Silverfast. If only their manual were readable! And their answers understandable...

Gametamer

Erik--

A very easy way to save for the Web is to scan away at high-res, the press Ctrl+Alt+Shift+W (File, Save for the Web), resize the image using Bicubic Sharper (this does not change your original, but instead resamples the Save for Web copy), choose png if you like (or jpeh High), and then export. You now have a smaller image for Webcasting that's the same gamma and colour space as the original---no need to scan twice.

BTW, and this is a PS problem, pfress F12 before you close the image, even though technically, you haven't changed it in any way. PS will nag you to save changes? if you don't...

My Best,

Gare

Bacardi

Yep. That's the way I do it, but I see/saw (you make me doubt) as a workaround. See, the original can be up to +3000 pixels large. This means that I have to dive down to some 20% or one pixel out of five.
I downsample in several steps (bicubic of course) and take care that PS must calculate and not simply discard one pixel out of two for example.
But even with a rather sophisticated sharpening (only edges) the result is often good, but sometimes not satisfactory. Sometimes it is visibly lacking detail.
Yes, perhaps I exaggerate. Hehe. You know me.

thanks for your time!
I would someone else would join in as there's a lot of interesting basic matter in here.