Gamma equalising?

"Ken" wrote in message

Please bear with me – I may not have the various terminologies correct, but hopefully you will be able to understand what I am trying to communicate.

I have Photoshop 7.0 on one of my computers and CS2 on another.
I am involved in the preparation of a newsletter in which we usually include images. These images may be scans of text documents, scans of photos or photos directly from digital cameras.

Although I work in colour, the printed output is in black & white and is produced by a commercial printer.

The printer requests that I send him the 16 pages for each newsletter as a PDF file. To this end he has provided me with the PDF profile for his printing machine to use when saving the PDF.

I believe this is a color profile. Install it on your system in the approrpriate folder, and use Photoshop’s convert to profile command to convert your photographs to this profile.

I have found that on some occasions two photos that might look quite good when viewed separately on the monitor appear unsatisfactory when printed.
Mostly this seems to happen when the photos are on two pages that are printed on the same face of the paper e.g. page 2 and page 15.

This is a dot gain issue. I’ll guess the images were too dark. Use of the pdf profile to convert your images should help with this, though there will be variation in dot gain depending on the side of the sheet.

My suspicions are the overall gammas for each of the images varies too much from one another for the printer to cope with. As a result, they either both appear substandard or one page will have good quality and the other page printed on the same piece of paper will be bad.

If you’re not converting your images using the profile, they will likely be too dark.

I know how to use the info feature to establish the average gamma (K) of a 5×5 pixel area. I would imagine that if I could select a larger (variable size) area to average, I would be able to adjust the various photos to be closer to each other in gamma. I can’t find a way to do this.

Curves. Find a midtone, and set it to a consistent value. Also be caureful that your shadows and highlights – areas with detail – are set to 90 percent and 10 percent, respectively. You may also have pure black and white areas – for example your text scans.

In summary. Am I correct in my diagnosis of the cause of the problem; is my approach to solving it correct; can you help me with the procedure for achieving the desired result?

Converting your images to the pdf profile is probably the biggest missing piece. Other thoughts: get a book by Dan Margulis, which deals well with printing issues in Photoshop, and check out the comp.publish.prepress group, where issues like this one are discussed routinely by people with press experience.
—

Mike Russell
www.curvemeister.com/forum/

Mike,

Thanks for responding.

"Mike Russell" wrote in message

The printer requests that I send him the 16 pages for each newsletter as a PDF file. To this end he has provided me with the PDF profile for his printing machine to use when saving the PDF.

I believe this is a color profile. Install it on your system in the approrpriate folder, and use Photoshop’s convert to profile command to convert your photographs to this profile.

It is a ‘joboptions’ file. My understanding is that this gives Adobe the parameters to use when the file is being converted to the PDF. I have not received an actual "printer profile" for his printer. I’ll ask him for one.

Curves. Find a midtone, and set it to a consistent value.

That’s one of my big challenges. I find it extremely difficult to pick the exact spot of 5×5 pixels to use as the midtone.

I find that the 5×5 area is too small. In a face the skin tones can vary quite a lot across the face and I find it really difficult to knwo which 5×5 pixels area to use. Is there any way that one can set a larger area for the ‘colour picker’ so that one could get a reading for the ‘whole’ face?

Also be caureful that your shadows and highlights – areas with detail – are set to 90 percent and 10 percent, respectively. You may also have pure black and white areas – for example your text scans.

The 10% margin makes sense.

To set the 10% and 90%, would I be correct in assuming that this is done by setting the shadow end of the curve at 25 and the highlight end at 225 on teh output side?

Ken

"Ken" wrote in message
[re PDF profile]

It is a ‘joboptions’ file. My understanding is that this gives Adobe the parameters to use when the file is being converted to the PDF. I have not received an actual "printer profile" for his printer. I’ll ask him for one.

Talking to the printer is a good idea. What you care about is the dot gain value. If you get no better answer, set up a 17 percent dot gain profile for your ink. This will print a little light on some presses, but this is less of a problem than printing too dark.

To set this up, go to Color Settings, click on "Gray" under "Working Spaces", and select the "Custom Dot Gain" item in the list. You’ll get a chart – type 67 in the entry for 50 percent, for a dot gain of 17 percent. Click OK, and save your custom profile under a name you will remember. As you scan each image, use Photoshop to convert the image to this profile name.

That’s one of my big challenges. I find it extremely difficult to pick the exact spot of 5×5 pixels to use as the midtone.

For overall light and dark appearance, rely on the screen. Then convert to your 17 percent profile if necessary and verify that your shadow and highlight are at 90 and 10.

I find that the 5×5 area is too small. In a face the skin tones can vary quite a lot across the face and I find it really difficult to knwo which 5×5 pixels area to use. Is there any way that one can set a larger area for the ‘colour picker’ so that one could get a reading for the ‘whole’ face?

This is not necessary, though you should calibrate your monitor carefully with Adobe Gamma, then adjust for a good appearance on your monitor. If you did need to average a larger area, Curvemeister will allow an artbitrary sized sample area, but this is not your problem.

The 10% margin makes sense.

The 10 percent allowance for highlight and shadow is a bit conservative, but about right for what you are doing.

To set the 10% and 90%, would I be correct in assuming that this is done by setting the shadow end of the curve at 25 and the highlight end at 225 on teh output side?

That’s approximately correct, but to be clear, this will be the final value of the shadow and highlight, not the position of the curve in the control. For example, if the darkest part of your image with significant detail has an initial value of 75% (this is in ink values, using your 17 percent gamma profile), place an info point there, and then move the endpoint of your curve so that it reads 95 percent. There can be areas of your image that are 100 percent black, provided they have no detail at all.

For example, a black background to an ad would be 100 percent.

These suggestions are reasonably accurate, but they are not absolute, and you should attempt to fine tune them bases on your experience. After the next job prints, if the overall look is too light, go with a smaller dot gain value. If your highlight and shadow are too conservative, back them off a few percent, say to 93 and 7.

You mentioned scanning type before. As you may have discovered, this will give you fuzzy edges. If you have any choice in the matter, use an OCR program or type it in directly as text.

Again – Professional Photoshop by Dan Margulis is probably the most relevant book for these issues. No one else discusses dot gain, etc, like he does. The Pocket Pal, from International Paper, is another useful handbook that costs about 15 dollars. www.ippocketpal.com .
—

Mike Russell
www.curvemeister.com/forum/

Oops.

"move the endpoint of your
curve so that it reads 95 percent"

should have said 90 percent instead of 95
—

Mike Russell
www.curvemeister.com/forum/