Fernando Gonzalez Gentile wrote:
> ...
>
> Moose, I was waiting for you to come back so as to ask if you use Lab mode,
> or exclusively RGB.
>
Early in my learning about image editing, I came across someone singing
the praises of Lab mode for various things, including luminousity
sharpening. I tried it out a bit, couldn't see an difference/advantage,
and gave it up.
> In short, I found that when using plain USM or LCE, my .tiff became more noisy
I think it is more accurate to say that the existing noise becomes more
visible. A subtle distinction, perhaps, but more accurate than saying
that USM adds noise. In whatever form, USM tends to sharpen the noise as
well as image data, making it more noticeable.
I find it generally necessary to reduce noise in scanned film and higher
ISO digital files before applying USM and/or sharpening. That is, unless
I'm looking for grain/noise as a part of the final image.
> ; and reducing noise in Luminance channel after converting to Lab was pretty
> simple and effective
First, it's important to understand that, as wonderful as many of its
tools are, the noise reduction functions in PS are not very good. They
are far behind the performance of the various specialized NR apps. I
assume from what I read that others are equally good, but NeatImage is
what I use and know a bit about. Ni and PS NR aren't even in the same
universe, when it comes to flexibility and effectiveness. I tried the
improved NR in CS3 and found it useless by comparison.
NI works in a Lab space, or something very like it, with noise reduction
adjusted separately for luminosity and the two color channels. As it
works both as a plug-in and stand-alone, I assume their conversion
algorithms are different than those in PS. NI also has a resharpening
function, which also works in Lab space and, by default, sharpens only
the luminance channel. With certain image files, the sharpening works
like magic, with the result that the processed image has both less noise
and better visible detail than the original.
Second, I think you will find that most folks writing about these things
and showing examples are starting with digital camera files. I work with
both digital originated files and film scanned @ 4000dpi. It's my
experience that the two require quite different processing, especially
as regards NR and sharpening. Film grain is quite different in detail
from digital noise. As you are working with film, it might serve you to
regard opinions about processing from people working only with digifiles
with some skepticism.
I admit I often simply try things out when I'm not getting the results I
want. I was just working on some high ISO RAW images from the A650. NI
at full size did a pretty good job, but I couldn't get quite the amount
of NR I wanted without too much detail loss. So I tried finishing my
processing, leaving some blotchy color noise I didn't like, downsampling
to display size, then running NI on the smaller images. Magic! With the
right settings, the blotches disappeared and detail came up.
I quickly went back and applied this solution to the images already
processed before I tried it. I always keep an intermediate PSD file with
all processing applied and often more than one layer, so I can go
forward if I want to revise something, rather than starting over.
Sometimes, all I need is to change the opacity of a layer to make a
change I want after looking at a 'finished' image for a while. In this
case, that allowed me to apply NI to a reduced size, but still 16 bit &
aRGB, image before conversion to 8 bit sRGB and saving as a JPEG.
Whether this is only effective with certain size images with certain
noise characteristics. or may be more generally useful, I don't yet know.
> - started to think of doing nothing (apart from ICE ) in the scanner and do
> all post-processing in PS
Hurray, I thing he's got it! I use scanner software for only three image
processing functions, exposure, ISO profile application and dust reduction.
> , then, why not directly in Lab and the back to RGB.
If Lab does useful things for you, sure.
> Those conversions are lossless AFAIK.
>
As resolved in the LL forum thread you linked to before, 16 bit
conversions to Lab and back using the Adobe ACE engine, dithering off
and no BP conversion (I take care of BP & WP in the exposure part of the
scanning process.) has minimal loss.
About sharpening in general: I almost never use USM directly for
anything other than LCE. Generally, NI provides all the sharpening I
need at full pixel image size. I very occasionally use FM's
IntelliSharpen at full pixel size. Mostly, I use FM's IntelliSharpen II
at display size for web images.
It's obvious from the way they work that both versions use more subtle
and complex processes than USM alone. I can see the original version
using an edge selection step. And blending of layers happens. I don't
know the details, or much care. I do know that these tools are cheap and
do a better job than I was able to do before I started using them, and
with far less time and effort.
All of my sharpening is done in new layers and I click the top one on
and off to evaluate the effect. I then often adjust opacity and fairly
often apply masking to tune the sharpening. With NI I sometimes apply it
twice, once with and once without resharpening, then mask the layers to
get stronger NR in shadow and low detail areas and retain more detail
elsewhere. I'm just not seeing color shifts when I do this stuff, so
that's a non-issue for me.
Moose
--
_________________________________________________________________
Options: http://lists.thomasclausen.net/mailman/listinfo/olympus
Archives: http://lists.thomasclausen.net/mailman/private/olympus/
Themed Olympus Photo Exhibition: http://www.tope.nl/
|