Re: Decent results with Gutenprint - the poor man's RIP.
Re: Decent results with Gutenprint - the poor man's RIP.
- Subject: Re: Decent results with Gutenprint - the poor man's RIP.
- From: "Roberto Michelena" <email@hidden>
- Date: Wed, 21 Jun 2006 08:18:00 -0500
I've been wondering about that one lately. It might account for a
particular kind of "rainbowing" I experience in the graybalance every
now and then, at least in the lower to medium end printers. The
rainbowing appears as a slight purple cast in one particular area of
a gray wedge.
I guess we've all seen those, in a variety of color casts; not only magenta.
In addition, the linearisation curves (specifically
magenta) show a kind of bump in the road, which I associated with the
dithering threshold.
But consider the fact that eventually you can get a good gray ramp
from that printer with that same driver, using either a different
linearization or different profiling program or patch count. In fact,
usually by using a low patch count in your profiling, you achieve a
smooth gray ramp while losing numerical accuracy.
So I'd think the "rainbowing" comes from giving the bump you mention
a higher weight than it should have (during the profile construction),
or maybe partically by measuring error.
Obviously, if a print exhibits such a response, it will be hard to
linearise, since it is not particularly useful to measure every
discrete value to get a good representation of the bump. If you
misrepresent the bump in both the linearisation and the 3d lut, then
it may well result in a rainbowing effect as described above.
I've always wondered why no one uses a "lots of points" linearization.
Linearization is so important that doing it with 21 points (5% steps),
then have one point with measuring error, or just happening to hit a
small local artifact such as the dithering treshold you mention, and
bum! you get bad results.
With nowadays fast spectros, I'd rather have a 256-point linearization
target, then average values to smooth out the curve.
Or at least, 60 values, just the same as the 21-point curve but adding
a -1% and +1% to each point. So something like 0,
(4,5,6),(9,10,11),(14,15,16)...and so on up to (94,95,96)(99,100).
Then consider the x% point to be a weighted average of its reading
plus the two neighbours.
That should get rid of measurement errors and also give local
disturbances a more proper weight.
I haven't looked at all the options yet, so I don't know if the
following is available already, but I would very much like to see a
driver that allows one to build a custom graybalance, yet have it act
as a normal RGB driver.
I believe some of Epson's standard RGB drivers have a "gray balance
tuning" option (via a separate piece of software?); don't know if what
they do is tweak the profile, or the internal linearization curves, or
what.
The question becomes: how can one
"methodically" create a custom graybalance, while bypassing the
normal printing pipeline. This Gutenprint could well do it, no?
Any RIP or driver that allows you to build your own linearization
curves, not from measurement but from numerical input, could be used
for this purpose, by the G7 method. (www.gracol.org).
Oscar Rysdyk
sounds familiar... ex Visu?
-- Roberto Michelena
Infinitek
Lima, Peru
_______________________________________________
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