Conversion to grayscale Tiff deepens shadows, but not for jpg2000

[Guy Burns]

I have a multi-layered Tiff file coming from a B&W negative scanned into 16-bit RGB. I want to create two archives from this file:

• a flattened jpg2000 file (for further use)
• a tiff file (just in case I want to edit the original in future)

To save space, I want to use only the green channel (the sharpest of the three), and then convert to 8-bit grayscale. For the jpg2000 I save as jpg2000, close, open, and then convert to 8-bit grayscale. The result is perfect!

When I save to 8-bit grayscale tiff, the shadows are deeper.

I don't understand why this is the case. PS does warn me that when I change modes the image may change. However, how to explain that jpg2000 files suffer no change, but tiff does?

The jpg2000 and tiff files, BC193 (18MB), can be downloaded here.

[c.pfaffenbichler]

»When I save to 8-bit grayscale tiff, the shadows are deeper.«

Did you flatten when the image when converting the tif to 8bit grayscale? 

If not a change is likely unavoidable because not all Adjustment Layers can exist in RGB and grayscale. 

Basically I think it would not seem like much of an exaggeration to say you started out wrong for what you seem to want to achieve. 

[D Fosse]

You can't make any color space conversion (color or grayscale) on a layered file. You have to flatten to maintain appearance. An adjustment is always relative to the color space.

However, it should be possible to "extract" one channel and use only that. But there's something you need to know first: a single channel is represented according to your working gray. If your working gray has a different tone response curve (gamma) than the original RGB color space, you get a tonal shift.

So if the original file is Adobe RGB, you need to set working gray to gamma 2.2. If ProPhoto you need gamma 1.8. And if sRGB, you need to set working gray to sGray.

[Guy Burns]

Thanks to D_Fosse, I think I understand where the problem arises. The scans come in as Epson RGB, but the individual channels are represented in my working gray – and the gammas might differ. I noticed that there are two Epson Gray colour spaces (1.8 and 2.2) under colour settings, which must have been installed with the scanner software. If I choose the right one, maybe this slight difference in shadows will disappear.

I'll try it.

[Guy Burns]

After trying it, the gammas of Epson RGB and Epson Gray (both varieties) must be slightly different as I can't get them to match.

But if I convert to Adobe RGB and use Gray 2.2, the shadow difference disappears. Same if I choose sRGB and sGray. I didn't realise converting to grayscale would be so involved.

This thing about channels using "working gray" explains another anomoly I came across before I started this thread. During testing, I copied the Green channel and pasted it above the Color Mixer layer (which was set to the Green channel). I assumed that the pasted Green channel and the Colour Mixer layer would give identical results. They didn't. When I alternated between the two, there was an obvious difference in brightness. Now I know the cause: one was in Epson RGB and other in my working gray.

To sum up: the difference I was seeing was not due to changing modes, but due to the difference in gamma between Epson RGB and my destination gray space.

Thanks Mr Fosse.