Dan Margulis Applied Color Theory - Moiré in a Subject

Date: Sat, 16 Nov 2002 09:05:37 -0500
From: John Romano
Subject: Moire removal

Hi all
i have a job that is scanned originals of bags, gym bags duffel bags and so on . if i play around with unsharp masking on the scanner i can reduce them slightly before things start getting too soft. but these bags have alot of texture and the customer wants them as sharp as possible, so blurring,or scanning as a descreen is not acceptable. does anyone have any tricks or tips that could eliminate some of this ? i have tried different mounting angles and different scanners from a hell drum scanner to a flatbed. just curious if their is any other way to do this.

thanks
john


Date: Fri, 15 Nov 2002 20:54:16 -0600
From: "Dan Tesch"
Subject: Re: Moire removal

Does the printer that will be printing this have the ability to use stochastic screening? If they do and can demonstrate past success with it this will solve your problem.


Date: Sat, 16 Nov 2002 10:11:51 -0500
From: John Romano
Subject: RE: Moire removal

Dan

thats what we were trying to do and we ran a press test, stochastic 20 microns i believe and it didnt help the moire at all,in fact it actually made it worse.i thought it was going to help but no dice.so im faced with correcting / rescanning a bunch of files.

john


Date: Sat, 16 Nov 2002 20:05:12 -0500
From:"Michael Demyan
Subject: RE: Moire removal

Hi John & crew:

Moire removal is possible by converting to lab.

Use the lasso or appropriate tool to select the moire area.

=====================================================================

Method #1

It's not a big problem - - very easy to fix. I automate it with a simple photoshop action.

- convert to lab color.

- select B chanel.

- apply a slight amount of Gaussian Blur.

- select lab channel to preview the result.

- if it's not enough, select A channel, and apply slight Gaussian Blur.

- select lab channel to preview the result.

- if still not enough, repeat the process.

- when done, convert back to RGB.

- when you determine how much it USUALLY requires, record an actiion with those settings and you'll never have to worry about it again.

Be ware:

- apply only as much Gaussian Blur as is necessary, as too much can effect color saturation.

http://www.dpreview.com/forums/read.asp?forum=1020&message=2981550

========================================================================

Method #2

You can accomplish the same thing (as far as I can tell) by doing the following:

- Duplicate layer

- Slight Gaussian Blur

- Change Layer Blending to COLOR

- Flatten image

The cool thing about this method is that between the time you change the Layer Blending to "COLOR" and the time you "Flatten Immage" you can use the eraser to erase portions of the "Blurred" layer that loose details or saturation that you wanted there (like teeth, eyes, etc.). You could even erase everything but the part that has the moire in it.

http://www.dpreview.com/forums/read.asp?forum=1020&message=2983474

========================================================================

The above methods are compliments of contributers on www.dpreview.com

Fuji SLR camera forum.

Mike Demyan

http://home.ptd.net/~mdemyan


Date: Sun, 17 Nov 2002 02:02:40 -0000
From: Stephen Marsh
Subject: Re: Moire removal

John writes:

> does anyone have any tricks or tips that could eliminate some of this ?
> i have tried different mounting angles and different scanners from a hell
> drum scanner to a flatbed.
> just curious if their is any other way to do this

John, perhaps try these links for more ideas or techniques:

http://www.pixelfoundry.com/Tips/KPT10/KPT10.html

http://www.camerabits.com/QM2.html

http://www.camerabits.com/techtalk.html

You may also want to seriously look into the resolution and physical size of the output. If resizing in layout to another magnification which is not 1:1 100% then this can make the problem worse.

At the same time, resizing down in Photoshop can add to the problem too - but at least with a 100% sized output file you know exactly what you are getting when it comes to pattern reproduction. You can try the single downsample step, 66% or 33% steps then a final step down to the final size or incremental steps in 10% or so until you reach the desired size in an effort to keep/reduce patterns...all of this can be done with or without small USM in the steps. There is a lot of room for experimentation when it comes to this, depending on the pattern and output.

Regards,

Stephen Marsh.


Date: Sun, 17 Nov 2002 13:39:33 EST
From: Dan Margulis
Subject: Re: Moire removal

John Romano writes,

>>i have a job that is scanned originals of bags, gym bags duffel bags and so on . if i play around with unsharp masking on the scanner i can reduce them slightly before things start getting too soft. but these bags have alot of texture and the customer wants them as sharp as possible,so blurring,or scanning as a descreen is not acceptable.>>

Moire is an interference artifact between two patterns. It comes in three varieties.

1) Between the object being photographed and the scan or digital capture sampling method.

2) Between the digital file and the phosphors of the viewer's monitor.

3) Between the digital file and the screening pattern of the printed product.

A lot of people worry needlessly about moiré when it's only one of type 2, which can be ignored. (To find out, check the picture at different magnifications on screen. If you can't see the moiré at every magnification, it isn't there.)

A lot of people also confuse type 1 with type 3. In John's case we know that type 3 is impossible, because he's already run it with a stochastic screen. So, the moiré is in the original digital file.

This is hard to take out, but it can be done. The suggestion was made by Mike Demyan to convert to LAB and blur the A and B channels. This is a good suggestion but it isn't quite enough. The moiré will affect color and contrast both, whereas blurring the A and B will affect only color.

So, before proceeding, one must check the individual channels to see where the moiré is. Commonly, it's in only the cyan or only the magenta. Occasionally the black. Then, convert a copy to LAB, blur the A and B as Mike describes, but when returning to RGB or CMYK replace *only* the moiré channel, as the blur may have adversely affected one of the others.

Somehow, though, the contrast moiré has to be affected. This can be done by attacking the L channel while still in LAB. I prefer to return to RGB or CMYK afterwards, set up a layer, and blend an unaffected channel into the moiré channel, then set the layer mode to opacity. This should eliminate the contrast moiré and the AB blur will have killed the color moire.

Dan Margulis

P.S. When I was on vacation, Ron Scratch posted a description of how he handled a similar situation by working with the L channel. His post is as follows:

********************************
I had recently run into a pickup image of a dress that had produced a moiré pattern when it was picked up and resized for another book. The moiré was so bad, that normal softening of the offending channels in cmyk did me no good. Rescanning the image was not an option because the transparency was not available. Turning to a page in Dan Margulis's book Professional Photoshop 6./chapter 15. I modified the moiré recipe and produced very good results.The size of the image to be used was sized @ 40% which produced the moiré . Fixing the image using the recipe at the new size did not work, however applying the following steps to the original larger image worked very well. Take the original image and mode change to Lab. Make two copies of the "L" channel, name one channel "X" and the other channel "Y". Display the "X" channel, apply a gaussian blur to the "X" channel - the book advocates a radius of 1, however for this particular shot I used a radius of 2. Now- display the original "L" channel and choose the "Apply Image" function and choose the blending option "darken", Apply the "X" as the source to target the original "L" channel @100%. Now you can discard the "X" channel. Next- display the "Y" channel, use curves to make the highlight a 20% k dot then apply the "dust & scratches" filter - the book advocates a radius of 1 with no mention of a threshold value but I used the default of zero. Now- display the original "L" channel and choose the "Apply Image" function and choose the blending option "lighten", Apply the "Y" as the source to Target the original "L" channel @100%. Now you can discard the "Y" channel. Your image is now ready to be converted back to CMYK. Since this garment was shot on figure, I retained a copy of the original CMYK image to preserve the balance of the flesh tones, background etc. And placed the new garment onto the original file. When the image was resized, no moiré was prevalent. Success!

Ron Scratch


Date: Sun, 17 Nov 2002 14:41:40 -0500
From: John Romano
Subject: Re: Moire removal

thanks to all who replied,

i have been using the convert to lab method and blurring the a+b channels and then some channel mixing for the k channel when i get back to cmyk and cut and paste back to my original file in the areas that i needed.looks pretty good on screen but wont know till i see some proofs.
thanks again

john


Date: Mon, 18 Nov 2002 11:58:21 EST
From: Harald Johnson
Subject: Re: Moire removal

Dan: "Moire is an interference artifact between two patterns. It comes in three varieties. .."

Some good tips offered!

Here's another idea for inkjet ouput (related to Dan's #3): Experiment with the image/file resolution. For example, I had a sweater with a tight weave. The moire at 300 ppi was evident, but when I changed it to 299 ppi, it disappeared.

Harald Johnson
DP&I.com ( http://www.dpandi.com )


Date: Mon, 18 Nov 2002 13:11:44 EST
From: Jeff Tytel
Subject: Re: Moire removal

This is always an interesting subject. The first question is: where are you seeing the Moire? On Screen? At Proofing ? (what type of proof and dot structure).

I have seen moire on screen and it disappears at proofing and reappears on press. I have also seen moire on screen, at proof and then disappears on press.

I would love to hear Dan's feedback and any prepress people on this.

Jeff Tytel
Tytel Images Unl. Inc.


Date: Mon, 18 Nov 2002 16:33:42 -0200
From: pajuaba
Subject: Re: Moire removal :

Besides method#1 (which I use most) and #2, there is a third one that is good to complement #1: after converting to Lab, use the rubber stamp tool to clone a+b channels from good areas to the offending ones.
--
Rodolpho Pajuaba
www.pajuaba.com.br


Date: Mon, 18 Nov 2002 18:00:32 -0500
From: Jerry L. P'Simer
Subject: Re: Re: Moire removal

Jeff Tytel wrote:

> I have seen moire on screen and it disappears at proofing and reappears on
> press. I have also seen moire on screen, at proof and then disappears on
> press.

In my experience, if the moire can be seen in the image data at 100% or higher magnification in any channel then the moire exists in the file. If not, than it is caused by the proofing or screening methods used to make the proof or film. Certainly the measures used to proof or output the file can make things even worse when the moire exists in the file itself.


Adobe Photoshop training classes are taught in the US by Sterling Ledet & Associates, Inc.