Re: DMAX - Total Ink selector for Photoshop

I remember these arguments from my daze working in Service Bureaus and prepress. Thank God I don't do much of that anymore. As a freelance production artist a lot of my job is preventing technical fingerpointing from reaching my client. My client is a marketing manager - how the job gets printed is of zero importance to him. When he starts getting emails saying the ads aren't passing density issues with long descriptions in pseudo-technical language which the operators don't understand then I'm in trouble.

Mike found a terrific workaround for identifying density - making a target mask and isolating DMAX violations using Threshold. When I ran the script on my image it came up accurate +/- 5%. There were pixels just under my target of 300. Probably due to some of the rounding, ie. 191.25 -> 191, etc.

Now all I have to do is run this script, run a curve on the mask, and keep my nice dark shadows and detail which my designer is so fond of. And the nasty emails from production stop.

BTW - the ad I ran it on as a test had been kicked back. The DMAX violations as indicated on the mask and verified by taking numeric readings can best be described as a smattering of fly*** on a bed***. Density is a printing issue involving large contiguous areas (1 sq. in. or greater) of dense ink which affect drying time in milliseconds, and minor saturation effects on some papers.

Maybe someone reading this long-winded description can add to or correct this opinion. But 100 scattered pixels out of 10 million are not going to create a saturation issue.

Anyway - Mike is a genius in my book - and now I have a 2 minute fix I can run on any file I create. And my client and all the media buyers in between no longer get those stupid emails. Thanks Mike.

Charlie Dvorak
Graphic Designer for
Speed Channel




Mike Russell wrote:
From: "Lee Blevins" <leeb@xxxxxxxxxxxxxxxxxxx>
[re 0 to 266 internal range]

I've wondered for some time if this is really true. There are 256 binary
values in an 8 bit byte so that would yield numbers from 0 to 255 in
decimal math.


Yes, this is deep in the bones of Photoshop. Internally, CMYK is like RGB : 0=black and 255=white


This also means that there are two or three binary numbers for each
percent dot.


Yes - and these extra values are used internally.


We could argue that there are values between full percent mearsures but
can halftone rips really show them?


I don't think so. The formula for calculating gray levels for a given
resolution producting halftone dots escapes me at the moment. But if I
remember correctly it was something like the resolution divided by the
screen ruling plus one or something like that.


An individual halftone dot is limited by the number of imagesetter points that comprise it. A 1200 dpi imagesetter set to a 120 line screen would have 100 points per halftone dot. In effect this means you are limited to 100 gray values per dot, plus one for no dot at all. Techniques such as error diffusion help a little with this, so you can do tricks like add a point randomly to every 10th halftone dot, and get a slight darkening of an area.


So in reality could a full scale image (having all 100 percent dot
values) be expressed in a 7 bit value? Or could the 0 or 255 byte be
reserved for other meanings?


7 bits per channel turns out to be fine for printed output, which is why 8 bits per channel remains such a good foundation for image editing. It's unlikely that the additonal bits will ever be used for anything, but there are new image formats, such as HDR, with 23 significant bits and tremendous dynamic range.


Can a binary image contain the NULL character?


Zero percent transparency can be thought of as an image NULL value, since it defines areas of an image that are neither white nor black.


I stay up late at night and wonder about this stuff.
I need a life.


LOL. I resemble that remark :-)
---
Mike Russell
www.curvemeister.com


.