Chris B. wrote:
> I don't normally examine shots like that, so it was a bit of a
> test, especially at this time of the morning. However, I
> notice that the centre square is a little softer at the
> extremes of aperture (largest and smallest) than the middle
> settings. It looks pretty good to me; what else should I
> notice? :-)
What? You want me to provide the technical analysis too? ;)
The purpose of the test was purely for personal reasons--I
wanted to benchmark my imaging workflow using the IS-3 as the
source. I was wanting to use the IS-3 for shooting film on Isle
Royale instead of using the OM system. I set up the test using
full-stop apertures and common focal lengths.
There are several factors in play here that all have an effect
on the image quality:
1. Lens. I always knew this lens was a dog wide open, but never
knew that it kept getting sharper the more you closed it down.
Defraction limiting does not seem to be an issue with this lens.
I was always afraid of using F16 and F22. I know that the IS-1
with the 35-135mm lens was much sharper in the mid apertures.
What really floored me was how gut-wrench aweful 100mm was--this
must be the cross-over point in the zoom lens design.
2. Film. XP2 gets sharper the more you overexpose it. (per
Ilford). This test was at the rated speed of ISO 400. In a
fringe shot on the roll I overexposed by a stop and the apparant
sharpness increased by about one line pair. Good to know. Grain
noise also decreased a little bit due to contrast gain.
3. Processing. The film was processed in the one-hour lab at
Wal-Mart. I've noticed a slight improvement in grain when the
film is developed dip-n-dunk. But in this test roll there is
little evidence of grain migration which provides a form of USM
sharpening.
4. Scanner. All of these images were scanned with auto-exposure,
auto-focus on the Nikon Coolscan V. The 5X overscan image was an
exception where I manually adjusted everything. Scanner focus is
always an issue, but as a general rule (maybe with the exception
of a couple of the images), the scanner focus was not the
limiting factor in this test. However, when viewing the
diagonal resolution test lines, you will see a marked decrease
in resolution in the diagonal than the horizontal and vertical.
Where you have lines visible up to 20 (2000 vertical lines of
resolution) in the horizontal test bars, the diagonal test bars
have mushed out by 18. I can only assume that this is a scanner
(4000dpi) issue as the lines look clearer with a loupe. But
there is no aliasing!
5. Software. Scanning was done using the very latest/greatest
version of Viewscan. Vuescan and the Coolscan V have a little
exposure issue, and the latest version has addressed better, but
it still exists. I did use the Ilford XP2 profile. Vuescan was
used because I'm able to disable all sharpening and
noise-reduction, and it does batch scanning a whole lot better
and faster. It also lets me do overscanning. Based on what I've
seen here, a 5X overscan really did help define the grain better
and gives it a more natural, pure optical enlargement look. 5X
was even noticably better than 3X.
6. In the natural world, the viewfinder coverage isn't that big
of a deal, but when shooting test charts it is. I lost a lot of
comparable resolution as a result. As compared to the tests on
DPReview, this setup tops out at about 2000 LPH. But, since I
lost about 15% of the actual film surface due to viewfinder
coverage, the total film scan is closer to 2400 LPH. In reality,
though, because of the Coolscan's inability to sharply scan the
edges of a negative anyway, the truely usable film coverage is
back closer to the reduced area anyway.
I provided the tests as is with no modification so you can apply
your own sharpening and noise-reduction techniques to see what
improvements are possible. I have experimented with some of my
common USM settings and with a couple of images was able to
achieve about a 10% increase in apparant resolution. (no
haloing).
AG
__________________________________________________
Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around
http://mail.yahoo.com
==============================================
List usage info: http://www.zuikoholic.com
List nannies: olympusadmin@xxxxxxxxxx
==============================================
|