While the method is often similar HOW any particular application does
its "thing" is determined by how well it's various Algorithms are
written.
True. The main point is, as Adam later says, "you can't un-ring a
bell", so neither can ever be as good as preserving the original
content, either "raw", or at least as a lossless generic image.
Raw image data is usually particular to whatever device it came from,
and has to be "cooked" to even see it.
For example, what comes off the scanner's pickup, or the camera's
sensor, is "raw", and the circuitry within the device bumps this down
to a generic image that can be directly viewed.
You can try to minimize the "editorialization" by not using in-device
manipulations, such as reduced resolution, sharpening, color
correction for light source, etc. but you still get a "cooked" result;
what was a scanner's 48-bit data stream is bumped down to 24-bit
TrueColor, and some processing has to be applied to a camera's raw
data to make any sense of it at all.
However, what modern cameras now do, is allow you to pull in raw data
and apply "camera logic" on the PC, which leaves the software applying
the logic in a controllable and vendor-improvable form.
Digital decisions you'd have made at the time of shooting, such as
color balancing for tungsten vs. phlorescent lighting, can now be
revisited when you process the raw data that predates whatever the
camera would have done to it. Optical decisions, such as optical zoom
or macro focusing, can't be revisited of course.
Lossless general image formats incluse those that are uncompressed,
such as .BMP, and those that may be losslessly compressed, such as
..TIF (lossless being the equivalent of zipping a .BMP).
You want to stay in these formats for the editing life of your image,
whether it be in Photoshop, GIMP or whatever. Perform the lossy
compression to .JPG (or the color reduction to .GIF) only at the final
output-for-purpose stage.
The way that JPEG works is inherently hostile to sharp, clear edges;
these get hammered into tramlines and blurring. For artwork that's
made up of sharp-edged stuff in few flat colors, as well as text and
screenshots, GIF may be better, even at the cost of shallower color.
Photoshop has been and remains the undisputed king of all photo
enhancement tools and it has been that for decades.
Yep. It's also damn expensive.
I've tried IrfranView and can't say that I was impressed. Remember
IrfranView is marketed as a VIEWER that can do conversions and make
some minor adjustments. It isn't a full blown graphics program.
Accepted. For example, when I was scanning photos and other damaged
paper artwork, I'd need semi-intelligent touch-up tools at the pixel
level, such as clone brush, selective sharpening etc. and I used Paint
Shop Pro for that, and still do. A particularly useful trick for
scanned artwork is to color-replace to the same color, which basically
reduces spurious color variation from the scanning process.
However, I find I rarely need pixel-level tools when working with
digital photos. I need what IView provides; a fast way to flip
through multiple shots, cropping, gamma adjustment and perhaps some
color balance over the whole pic, resampling, rotation and the ability
to save in different formats.
Saying it is "good" is ...
....more like saying the best racing bicycle is a good car. They are
inherently different tool sets - would you like to flip through images
in Photoshop, or have Photoshop drag itself off disk every time you
just wanted to look at a picture?
As far as quality, you can't unring a bell! If you start with a image
that is compressed, like a JPEG, a lot of the image's original bits
have been discarded in the initial compression process when the file
was originally saved. They're gone. Those bits no longer exist, these
is no way on God's green earth anything can put them back.
Yep - "information is that which cannot be re-created when lost".
It's interesting to crank up JPG compression to get a taste of what
"JPEGism" looks like - what I see is an image made up of tiles that
appear to be gradient-shaded according to corner pixel color values.
When that bangs into the tiling effect that is inherent in the matrix
mechanism of many sharpening, blurring etc. filtering tools, you can
predict you could be in for "beat noise" effects.
There's another possible reason for the subject line; problems that
arise when images are displayed in other than their natural
resolution, e.g. "full screen" or "to fit window". Image previewers
will nearly always work this way, as will IView if instructed to,
because screen res is lower than other output devices such as
printers, and decent cameras spawn images larger than the screen.
So an image that was well-resized (e.g. /2, /2, /2) to fit nicely when
(pre-)viewed as a "portrait", may mis-scale when rotated and viewed as
a "landscape". In addition, rotations of anything other than 90, 180
or 270 degrees are going to degrade the quality, because they require
a lot of edge-case pixel guessing.
--------------- ----- ---- --- -- - - -
To one who only has a hammer,
everything looks like a nail
