From 16-bit RGB to 8-bit CMYK (gradient on large area)

[Ilario5E19]

Hi guys, newbie here, as always 😉

I'm working on a photobook project with Blurb. The project contains various enlarged photos (double-page, around A3, horizontal) each one with a subtle gradient on a large area.

I work in 16-bit RGB to avoid banding when grading etc. Once finished I convert to 8-bit RGB, and after that I convert to CMYK via Blurb profile. This gives me also the advantage of the dither option.

The problem is that after this conversion, I like to make a sort of “secondary” grading with Curves properly in CMYK, and at this point I see that 8-bits are few because the gradients start banding.

I could convert from 16-bit RGB to 16-bit CMYK, in order to avoid banding during the “secondary” grading, but in this way I loose the usefull dither option (in fact if I convert 16-RGB to 16-CMYK the dither option is disabled).

So my question: which is the best method to convert from RGB to CMYK remaining at 16-bit, and then changing to 8-bit at the very end WITH dither option?

I don't know if it's correct, but what I'd do is the following: 1st grading in 16-bit RGB, then converting to Blurb's CMYK always at 16-bit, then 2nd grading, then returning to RGB without conversion (simply Mode > RGB), and finally converting again to CMYK 8-bit with dither option.

Does it make sense? Or is there an easier and more elegant method to do the same?

Thanks in advance.

[Ilario5E19]

Perhaps I solved: when I'm in CMYK 16-bit and I simply do Mode > 8-bit, that should introduce dither  since it's specified into the preferences. Is that right?

[Stephen_A_Marsh]

The LSB dither is very subtle. Sometimes stronger noise is needed, still subtle though. You usually don't wan't to introduce noise into the extreme shadows/highlights though. I created an action to avoid that:

https://www.dropbox.com/s/5rmz0b6c7a40h7y/Smart%20Noise%20CS3.atn?dl=0

[Guest]

You don't wan't to introduce noise into the extreme shadows/highlights though.

By @Stephen_A_Marsh

Hi Stephen. Doubt here: if the CMYK image is intended for printing, I believed a very small amount of noise would be better in shadows. I believed that because I noticed that recommended TIC for graphic background colours (that are perfectly uniform colours) is around 260% (example is rich black). So, in a real life photo (with a TIC around 300% or more) I suppose it's better to use a small amount of noise in the shadows. Is it not correct?

[Stephen_A_Marsh]

A real life photo should have enough natural variation that noise isn't required for technical reasons, only artistic. Adding noise is generally required for synthetic elements such as gradient fills or gradated masks. Adding noise will naturally affect extreme highlights and shadows with both light and dark pixels. This may or may not be desired in all cases.

Edit: this is not related to total ink limits for rich black panels.

[D Fosse]

Just add a little bit of noise, and work in 16 bit throughout. Not so much that you see it - just enough to break up the banding.

Dither is noise and this is the standard way to deal with 8 bit banding.