I don't think it will mater, assuming the final image is going to be flattened before printing.
Not necessarily. Photoshop can tend to do gradients better than the vector programs, but flattened or unflattened may not have much to do with it. There is an equation that Adobe put forth many years ago for figuring out whether or not one would see banding. I don't have a link for you, but you can google something like "Adobe gradient banding equation" and I'm sure you'll find it.
I think some of this depends on rip functionality too. As I understand, some rips have ways of smoothing gradients to mitigate any banding (such as subdividing shades of color channels to fractions of a point, e.g. instead of 0, 1, 2, 3 all the way to 100 for a given color channel, it might be 0, 0.1, 0.2, 0.3, etc.).
Great info!! Thanks. Now for the noise would I create a new layer and add, then apply a multiply, or just add it right to the gradient? Thanks
Noise: I've gotten the best results with Gaussian noise at 1.5-3.5, Monochromatic. Depends on the colors used in the gradient as to what value to use. Better to have an adjustable layer so that you don't have to rebuild the gradient AND the noise if you don't like how it looks initially. So, create a new layer on top of the gradient layer, fill the layer with white so that PS will have something to build the noise on, then use the filter, then change blending mode to multiply and adjust layer opacity as needed. For opacity, I recommend a happy medium between visibility of the noise and reduction of the banding to the eye. Again, that will depend on the colors in the gradient as well as how PS has constructed the gradient, and upon the values you use in the noise filter. Also sometimes helps to increase contrast after creating the noise but before adjusting it's layer opacity.
BTW, be careful using the noise filter with gradients that fade to white. If you do that, be sure to fade the noise to white also. Otherwise the white end of the gradient will actually have noise in it instead of being absolute white.
If you are making a gradient go from a color to black, use a mix of 100% black and some % of the other color. This will make your gradient look much smoother.
I've seen the best results using 100% of the other color all the way across, and then fade only the Black from starting to ending values (e.g. 100c 0k fades to 100c 100k, so that Cyan stays the same value across the gradient and only the Black gradates). This will limit any potential banding to only one color, in this case, Black.
----------
There's something called "the 7-inch rule" a single-ink gradient (at typical linescreen) can make a 100% change in 7" and not band (if it goes from 50% to 0%, it can do that in 3.5" and not band).
multi-ink gradients usually don't band due to the different screen angles of the halftone (though each plate can appear severely banded). 3-ink and 4-ink gradients almost never band due to the 3 or 4 screen angles involved.
(the noise trick in Photoshop is the best thing you can do with those 1-ink gradients that must be longer than 7 inches)
I believe the 7-inch rule and Adobe's equation are related IIRC.
----------
If you have a background color other than white, say red for example, and have a black gradient that is going to go from black to red, fill the entire background with the red color and then, on a separate layer, have the black gradient set up to go from 100% black to 0%.
Correct. Though the black doesn't necessarily have to be a dedicated layer - important thing in my experience is that the gradating color be isolated to as few channels as possible (YMMV).
