View Full Version : Banding in ink optimized gradients

12-21-2010, 05:03 PM
I was running some tests using ink optimizing on raster gradients. I noticed that after optimizing, some of the gradients had some banding.

For example, a gradient that goes from 0,0,0,0 to 90,80,70,10. After ink opt, it goes from 0,0,0,0 to 50,40,30,60. (maybe not the actual numbers)

However, in say the black channel, there might be an area where the shade goes, 6,7,8,9,8,9,10 or 14,15,16,18,19,20. I've seen this in all the channels, not just black.

Has anyone noticed this in actual jobs and has it been visible on the press sheet?

12-21-2010, 07:41 PM
Hi Jamie,
What color response were you optimizing? What were the settings for dvl creation?


12-21-2010, 08:49 PM
Gracol PCO, Ink Opt is relative, 320, 100, 80. I also tried 95 and 100 and it showed up there too. I haven't tried anything lower than 80. I also exported the profile and used it with convert to profile in Photoshop and I got the almost same result.

After more looking, here's what I found out: In the "Convert to profile" dialog in Photoshop, there is a checkbox that says "Use Dither". Using a threshold layer, I can see that turning this on makes the vignettes smoother by dithering the transitions from one level to the next. By using a threshold layer on the file that was ink optimized by CF I can see that CF does not do any dithering so the level transitions appear more abrupt.

I should point out that this test file was a synthetic gradient with no noise. An actual photograph with natural noise may not behave the same way. However, people do like to make gradients in Photoshop, so we can't pretend we won't see this in real jobs.

01-07-2011, 07:47 PM
You'll find that even before you apply the profile in photoshop, this phenomena occurs. My tests are showing that this "tint jumping" occurs mostly as a result of how the gradient is rendered in photoshop. Using dithering didn't change the behaviour for me, however.

I will look into this further.