Re: Pre-Linearization or Iterative Profiling (WAS Re: Printer profile mis-shape?)
Re: Pre-Linearization or Iterative Profiling (WAS Re: Printer profile mis-shape?)
- Subject: Re: Pre-Linearization or Iterative Profiling (WAS Re: Printer profile mis-shape?)
- From: Rick Gordon <email@hidden>
- Date: Mon, 28 Aug 2006 16:53:28 -0700
So, then going back to the consideration of how to hack this behavior into ProfileMaker (until, hopefully, it's incorporated for real), might there be a way calculate an optimized reference file, and then build a new testchart in MeasureTool from the modified reference file?
Rick Gordon
------------------
On 8/28/06 at 7:41 PM -0400, Marc Levine wrote in a message entitled
"Re: Pre-Linearization or Iterative Profiling (WAS Re: Print":
>Marco,
>
>> The way I understand it (my apologies if my language turns out not to be
>> completely appropriate), this "pre-linearization" (PL) of the target does
>> nothing more than just change what color numbers are being sent to the
>> printer when one prints the target.
>
>[ML] Exactly correct. Which is exactly what the linearization curves in your
>RIP do. Imagine the numbers for you digital CMYK patches on the other side
>of a RIPs's calibration curves....different. With MonacoPROFILER, you are
>getting a new target with values from the "other side if the curve". The
>curve itself is kept in MonacoPROFILER and the resulting profile tabels are
>welded together with the curves.
>>
>> Say that patch 3 in row one of your RGB target reads R 170, G 170, B 0
>> before PL. After PL, the patch may R 194, G 165, B 0 instead. A majority of
>> the other patches in the target will also change values, to some extent.
>
>Everything will change, dependent on the curves.
>>
>> But this doesn't "linearize" your device in any way: it only sends numbers
>> (this is the way I understand it) that will output colors on that device
>> that, once measured, will be closer to the nodes in the profile's tables
>> describing its output. The point is to facilitate the math inside the
>> profile, so that the nodes and the interpolations are more precise and more
>> reflective of the actual behavior of the device, producing results that are
>> both smoother and more neutral.
>
>Um...no, and yes. What you are doing is linearizing the behavior of the
>device by a different metric that should affect a more uniform distribution
>of color sampling - which will ultimately help the profile do it's math. It
>is not because the linearization aligns the target path values to the final
>grid values in the rendering tables, but to simply produce a better behavior
>to sample. Again, you can think of the lin curves in Profiler (that you
>can't see) as a supplementary calibration curve that is combined with the
>standard lin curve to produce a different behavior. Because the curve is
>kept in MP, you need to apply the curve values to the target file before in
>is delivered to the RIP where it goes through the "second" lin curve".
>>
>> To recap, PL generates a new set of color numbers in lieu of the original
>> set of color numbers in the target. This target is then printed on the
>> device, and hopefully the measurements made from the modified target will
>> make the profiling software work more efficiently to describe the behavior
>> of the device.
>
>Yes.
>
>> But in no way this "linearizes" anything.
>
>Well....yes it does. It affects a linear state on the printer to sample the
>"original" target values that we save into an interim "curved" state for
>delivery to the RIP. Putting it a different way, you could take the lin data
>from MP, blend it together with the standard lin data, and effect the same
>type of linear condition that MP targets - only 100% in-RIP. You might need
>a copy of Excel and a fancy calculator, but you could do it. You could then
>run the standard target (un-lin'd) and you should get the same results.
>
>> I don't see how
>> this "pre-linearized" set of target colors would interfere with the
>> linearization from a RIP, though I would think it advisable to linearize the
>> RIP first, *then* build the "pre-linearized" target from that RIP's
>> internally linearized output.
>
>It doesn't interfere and it is advisable to calibrate in the RIP first. You
>are correct that the MP lin should be performed after the RIP lin. If you
>get this, then you get the gist. The only remaining piece is accepting that
>the MP lin is actually a linearization - a point of decision that is up to
>you.
>>
>> Being that I'm neither a mathematician nor a scientist, I can't explain it
>> better than that. And if I made any errors in my description, I would kindly
>> ask to be corrected.
>>
>I think you did fine. The key point is that you can linearize to a given
>metric, and then linearize again to a different metric. In both cases, the
>printer behavior is changed. When using the MP lin, you effectively align
>the printer behavior that is more linear through lab space. It's not to say
>that the device was linear in the first place, just probably to a different
>metric such as density. When the behavior in linearized against a Lab
>metric, it typically has the effect of creating a more even distribution of
>color sampling, using grids of color combinations, throughout the profile
>space.
>
>Hope this helps,
>Marc
--
___________________________________________________
RICK GORDON
EMERALD VALLEY GRAPHICS AND CONSULTING
___________________________________________________
WWW: http://www.shelterpub.com
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Colorsync-users mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden