remove stepping from highlight gradient

[frailer]

This is a fairly common one for us in our prepress neck'o'th woods. CMYK gradient starting from bottom of page, fading to, say zero in top 3rd of page. The transition from around 7% to zero has noticeable steps. I know that you can fix it by adding (not too much) Noise, but I don't want to add any layers, since quite often it's preferable to~Advanced Edit Tool~and open the PhotoShop image from within the PDF. If I can avoid creating another layer when I fix it, I can save it straight back into the PDF, and get a "new,improved PDF". I recall maybe that you can Lasso the required area, feather the edges, add noise...am I on the right track? The contained PhSh image has been flattened, I think. It has the CMYK channels, but no layers.
It has a sharp digital picture of people sitting over the gradient. Need to exclude this. Suspect I may have to Pen Tool around this to isolate it. Maybe I will HAVE to Pen Tool select, in that case. Would love an experienced PhotoShopper's take on this. I want to keep it at the KISS level if I can. Sorry can't supply image due to proprietary/anonymity reasons. :rolleyes:

[hewligan]

If adding noise fixes it, why not just add the noise on the original layer rather than creating a new layer for the noise?

Editability's nice and all, but if you don't need it, or it gets in your way, why bother?

[frailer]

Sorry Hewligan, it's a bit of a saga, I know, as I've described it....I don't want add another layer, just add noise to the PhSh image/layer that I've got. Guess I'm fishin' more for a "how to", easiest/most efficient method, whatever. Don't want to add noise to the "people" pic sitting over the background; guess this means deep etching with the Pen Tool?
This is something we face quite often. You don't, (nor the designer), see the stepping until the file's been Rip'd and rendered, in our case Black Magic. We have an up-to-date Adobe CPSI RIP. It's just the nature of the file that it needs to be Noise'd so it will print acceptably.
Hope this makes some sense.

[hewligan]

Yeah, if it's a flat image, you're going to have to select out the gradient area from the image.

Still, for what you're doing, you don't need to do that very accurately. So I'm guessing what you really need here is some quick ways to make a selection.

Well, there's always the magic wand. It's not the most reliable, but if there's a clear colour difference it should work okay.

Also, since you said it had sharp edges, the magnetic lassoo tool might give you a decent result.

After that, it's time to look in your channels palette, and see if any of them look close to the mask you want. You can also combine them quite well using image -> adjustments -> calculations.

Also, you could try applying the noise to the whole image, and then using the history brush to paint it back to the previous state in the areas where you don't want the noise.

[frailer]

Great tips. At this stage it's a practice later in the week for "next time".
Thanks :)

[CamarotaDesign]

Use a high LPI screening processes like Staccato. That eliminates any banding (I think thats what you mean by steps) My staccato printer asks for files in 600dpi to print to an equivalent of 300lpi. I also work with a printer that prints a 200 lpi output and they have no problems with banding ever. I guess that's not going to totally help in your case since you arte probably limited by your output device, but I can tell you that higher resolution output devices solve a lot of problems that I commonly see people asking about here.

[Broacher]

Stochastic, or FM screening. One solution-- but it requires higher rez images and is not, by any stretch, a standard press setup (certainly not with ad work).

Actually, higher lpi itself, as the only variable, will increase banding, and lower lpi lower it-- again, all things being equal. What happens though is that when most people output to 200 lpi they use the highest ppi setting for their imagesetter, and the higher ppi-- the lower the banding. It's all in the math. You have more resolution to work with in making more dots, you'll have less bands. But, even then there's a limit-- most RIPs convert everything to the postscript standard 8 bits per channel, which means in some transitions, the steps will always be visible. There is Postscript 3's 'Smooth Shading' --but that only applies to vector work.

[CamarotaDesign]

yeah, definately need the high rez files. Working with 600 DPI photos is pretty demanding on resources when you are working with a full page. And yeah, certainly you won't be finding any magazine printers that are using staccato. So really this option is like an "under ideal circumstances" type of thing.

[Broacher]

Any chance of lowering your LPI and increasing your output PPI?

[frailer]

Our output is "locked in". We run a Fuji platesetter on a Celebra RIP. More than happy with conventional screening at 175 lpi. Stochastic's a whole other ball game; our pressroom isn't air conditioned, just a giant "shed".
I guess I really want to improve my PhSh technique and was fishing for tips on this kinda problem. Highlight gradients seem to be the ones that bug us most, because the stepping is "masked" until output; i.e. post-RIP proof, though it's created in the original raster file.

[frailer]

Revisiting this today; want to get a handle on it. You have answered something that has been bugging me and it's now obvious, Broacher. The 8-bits per channel limit of PostScript probably explains quite a bit on these things. If you were just printing out to a non-PostScript printer the problem may not occur or be less obvious, correct? The annoying thing being you are in a high quality print environment, but this limitation will bring it out.