Canon D30 Digital Better than Provia 100F !!!
bc1...@my-deja.com
contemplating a jump from film to digital. WARNING: do not believe
all the inflated numbers you hear about the image quality you can get
from film. Look at images and ignore numbers. http://www.luminous-
landscape.com/d30_vs_film.htm
Brian
Sent via Deja.com http://www.deja.com/
Before you buy.
Tony Spadaro
or film source, has been severely compromised.
--
Digital photo restoration in autumnal Chapel Hill NC
http://www.homeusers.prestel.co.uk/magor/tony
New - A Digital Workflow + Sharpness: Threat or Menace?
And Selecting Your First SLR - Java must be enabled
bc1...@my-deja.com
illustrate quality. A good 24-bit video card and a decent monitor is
capable of showing an awfully nice picture. It is certainly possible
to send images over the web without any degradation at all because they
are DIGITAL. Besides, if you are comparing two images side by side on
the same screen, both are "compromised" to the same degree. If one
looks better than the other on the web, do you really think the web
loser will become a print winner? Explain. I wonder if you even
looked at the web page that I referred to. It is extremely
interesting, even if you are a film diehard. The film was scanned with
a very high quality scanner, by the way, and the film scan has a much
larger file size than the digital image. Nevertheless it is difficult
to find anything about the film images that even approaches the digital
images. Having a bigger file size in and of itself is no advantage if
the final print image is no better - you just waste disk space.
Brian
In article <8t64el$pjo$1...@nnrp1.deja.com>,
Tony Spadaro
bc1...@my-deja.com wrote:
> The compromise becomes less when you magnify a portion of an image to
> illustrate quality. A good 24-bit video card and a decent monitor is
> capable of showing an awfully nice picture. It is certainly possible
> to send images over the web without any degradation at all because
they
> are DIGITAL. Besides, if you are comparing two images side by side on
> the same screen, both are "compromised" to the same degree.
around that little step. The only scanner currently on the market that
can possibly get to grain size for Provia F 100 is the one being
marketed by Polaroid and Microtek at 4000 dpi. Although it has the dpi,
it does not have the dynamic range to do any film justice.
If one
> looks better than the other on the web, do you really think the web
> loser will become a print winner? Explain.
web and print are different. The only way to compare prints is with
prints. Not web pictures. Not magazine reproductions. Prints. If you
are not aware of the difference I doubt anything anyone writes will
help. Go to a museum and look at a print. Then go home and try to make
a print. Do it digitally, or chemically, it does not matter.
I wonder if you even
> looked at the web page that I referred to.
Your link was dead.
It is extremely
> interesting, even if you are a film diehard.
Planning on a digital SLR in the next year. I've seen 3.3 megapixel
images, they are not bad - I'm waiting for 6.6, my wife would go with
3.3.
The film was scanned with
> a very high quality scanner, by the way, and the film scan has a much
> larger file size than the digital image. Nevertheless it is difficult
> to find anything about the film images that even approaches the
digital
> images. Having a bigger file size in and of itself is no advantage if
> the final print image is no better - you just waste disk space.
You are comparing apples to oranges at best. It's more like comparing
tomatos to complete applesauce. Have you printed either - and with
what, and how large?
I make a lot of 12x18 prints with my inkjet and the occasional
16.5x22. My 35mm films can handle going up that large because the
prints are meant to be viewed at 5 to 8 feet. Will your 3.3megapixel
digital files enlarge to 12x18 - Super B 13x19 paper? Try one. How far
back do you have to stand before it looks like a picture?
Put up a working link and I'll take a look.
bc1...@my-deja.com
Not sure why this link did not work on my first post. When I enter it
manually it works fine.
I have printed many 12x18s from both a Nikon990 and more recently a
D1. In the past I frequently made 11x14 prints from 35mm negatives
both from scans and using a darkroom. At 11x14 I was always dismayed
by the appearance of grain unless I was using Technical Pan Film. The
smoothness I can achieve with a D1 exceeds anything I've ever done or
seen with 35mm film. I have a strong personal bias against film grain,
and the current generation of digital SLRs provide the type of image
smoothness that I was formerly only able to get with 4x5 format
enlargements or even 11x14 contact prints. At 12x18 size, the detail
from a D1 or 990 leaves little to be desired compared to 35mm film. I
would be interested in hearing your impression of the digital vs film
comparison at the luminous-landscape website.
Brian
In article <8t672t$rk5$1...@nnrp1.deja.com>,
OneThumb
Jerry Coffin
> The compromise becomes less when you magnify a portion of an image to
> illustrate quality. A good 24-bit video card and a decent monitor is
> capable of showing an awfully nice picture. It is certainly possible
> to send images over the web without any degradation at all because they
> are DIGITAL. Besides, if you are comparing two images side by side on
> the same screen, both are "compromised" to the same degree.
This is simply false. Putting something on the web doesn't
compromise the picture by a fixed percentage or anything like that.
Instead, it places a ceiling on the quality that can be shown. A
digital camera is close to that ceiling, while film is FAR above it.
Both of them end up close to the same ceiling, but the degree of loss
for film is MUCH greater.
Note, in particular, that the range of colors that can be portrayed
on a monitor is _extremely_ limited. Most CCDs have roughly the same
limit, but good film has a MUCH greater capability.
> If one looks better than the other on the web, do you really think the
> web loser will become a print winner? Explain.
Yes, in many cases exactly that will happen. Scanning film and
getting good results is somewhere between a science and an art.
> I wonder if you even looked at the web page that I referred to.
> It is extremely interesting, even if you are a film diehard. The
> film was scanned with a very high quality scanner, by the way, and
> the film scan has a much larger file size than the digital image.
> Nevertheless it is difficult to find anything about the film images
> that even approaches the digital images. Having a bigger file size
> in and of itself is no advantage if the final print image is no
> better - you just waste disk space.
There are two problems. First of all, it's trivial to get bad
results from almost any medium. Second, even the very best scanners
lose quite a bit compared to the original. The very best scanners DO
have close to the resolution of film, but in reality to maintain the
full quality of the original, you need to scan at a minimum of twice
the resolution. Even if you do scan at that resolution, current CCDs
don't have anywhere close to the dynamic range of film either. In
theory, if somebody wanted to badly enough they could compensate for
this, but I'm not sure anybody tried to do so -- I am quite sure that
if they have, it's only in a few of the hideously expensive drum
scanners.
Note, however, that scanners are getting better all the time. If you
take a picture on film now, you can scan it ten years from now when
scanners are LOTS better. If you take it on a digital camera now,
it's NEVER going to get any better. Of course, if you shoot it on
film now, you can also use strictly photographic printing to get full
resolution now.
Of course, the vast majority of what people see out of digital
cameras has been processed heavily before they see it, even when they
think they're working with the raw data. If you apply the same kind
of processing to big scans, it's often trivial to get the results
that are dramatically better still.
If you don't believe what I'm saying about processing, keep in mind
that a digital camera normally has its pixels broken down as 50%
green, 25% red and 25% blue. With most cameras, you never see a file
that way though: it's already processed it enough that you're seeing
pixels that are each composed of all three colors. In the test on
the site you cited, he's almost certainly looking at things after
this level of processing has been done. If you accept that kind of
processing on digital images, there's no reason NOT to apply it to
scans from film as well, giving the same advantages, but starting
with more raw data to get better final results.
--
Later,
Jerry.
The Universe is a figment of its own imagination.
bc1...@my-deja.com
Michael Reichman's test was a very good real-world side-by-side
illustration of what can be expected from high quality digital vs film
in a computer darkroom environment.
A couple of points: 1) Darkroom processing probably loses at least as
much resolution and contrast as a high quality scan because the MTF of
the taking lens gets cascaded by the MTF of the enlarging lens -
scanners also use imaging lenses, but they tend to be extremely well
corrected, especially on a bona-fide 3200DPI high-end scanner; 2) If
the web presents a ceiling on observable quality of two different photo-
producing techniques, and if within that ceiling one of the techniques
clearly appears better than the other, then exactly how would one
arrive at an opposite conclusion by looking at actual prints? Wouldn't
the diffferences perceived on the web become even greater once the
ceiling is raised? 3) The cascaded resolution of film and camera lens
can definitely reach 100 lp/mm or perhaps a little more (IF you do
everything very carefully), BUT, the contrast of the details beyond 50
lp/mm or so becomes very low even in the best of circumstances - you
can detect that it is there but it doesn't make an overwhelming
impressing in a print. The inherent advantage of digital image capture
is that you don't have to go through any re-imaging steps in order to
produce a print. If you have 50% contrast at 30 cycles you don't wind
up cutting it in half or worse by projecting it or scanning it. The
resolution cutoff in digital SLRs is lower, but the image contrast up
to the cutoff is better, and that is one reason that digital images
look so good when carefully compared to film images. I will admit that
I have no interest in slide projection or slides as art objects unto
themselves without enlargement, but am instead primarily concerned
about high quality prints. The other problem with film is that even an
ultra-fine grained slide film like Provia has very disturbing
granularity compared to images from the current crop of digital SLRs.
The utter lack of grain is an overwhelming advantage of digital cameras
in my view.
Brian
In article <MPG.1460c6883...@news.mindspring.com>,
ji...@my-deja.com
bc1...@my-deja.com wrote:
> The following link will provide very interesting reading to anyone
> contemplating a jump from film to digital. WARNING: do not believe
> all the inflated numbers you hear about the image quality you can get
> from film. Look at images and ignore numbers. http://www.luminous-
> landscape.com/d30_vs_film.htm
I respect Michael Reichmann's opinions. He has quite a lot of experience
with digital image processing, and a great deal of experience with
photography in general. However, as he notes in the link you referenced,
he is a photographer, not a scientist; there may be a flaw in his
methadology.
One important aspect of Mr. Reichmann's comparison is that the scan of
the Provia F transparency was presented without any sharpening. It's my
understanding that scanning film in and of itself introduces
unsharpness, which is corrected by subsequent unsharp masking. It is not
clear that collecting data in the RAW mode similarly degrades the
image; instead it may (?) simply allow later control of sharpening.
It may be that the comparison is truly between data that CAN be
sharpened, and data that MUST be sharpened to restore the
on-film sharpness that was degraded by the scanning process. If that is
true -- and I'm a scientist, not a photographer -- the comparison is
biased to favor the D30.
Tony Spadaro
into consideration the many alterations he made to both images, there
cannot be a comparison. I've printed 3.3 megapixel images. They
certainly can compete with film for small prints. Even smaller digital
pics can be used for magazine reproduction. Anything with the rough
shape of the subject will do for newspapers. At 8x10 even my two
generation old printer shows that the digital image is not continuous.
When you read terms Unsharp Mask, Levels, Geniune Fractals, you know
these are not straight images. Straight images made of close subjects -
preferably with a range of colours, and shades of the colours, might be
more believable. I still would want to see the prints, however.
Digital is catching up with film, and it looks like the CMOS sensor
may very well be a viable alternative to CCD. This much is true. The
rest is applesauce topped with bull, sprinkled with hooey.
--
Digital photo restoration in autumnal Chapel Hill NC
http://www.homeusers.prestel.co.uk/magor/tony
New - A Digital Workflow + Sharpness: Threat or Menace?
And Selecting Your First SLR - Java must be enabled
Mark Spatny
wrote:
>Putting something on the web doesn't
>compromise the picture by a fixed percentage or anything like that.
>Instead, it places a ceiling on the quality that can be shown. A
>digital camera is close to that ceiling, while film is FAR above it.
>Both of them end up close to the same ceiling, but the degree of loss
>for film is MUCH greater.
>
>Note, in particular, that the range of colors that can be portrayed
>on a monitor is _extremely_ limited.
This is true, and a major factor I have to face in my job on a daily
basis. In visual effects for film, we have to work on computer
monitors that cannot display the dynamic range of color that the film
contains.
What a monitor can display at any one time is limited by 8 bits per
color channel, in a linear distrubution along the dynamic range. Film,
on the other hand, is better represented by a 10 bit logarithmic
scale for each channel of color.
There is simply no way you can compare film to digital, unless your
intended output is only on computer monitors.
Jerry Coffin
> Just LOOK at the sample photographs. Its that simple.
I have. I found them unconvincing at best, especially after reading
what he said about what he'd done and how they were created.
> I think that
> Michael Reichman's test was a very good real-world side-by-side
> illustration of what can be expected from high quality digital vs film
> in a computer darkroom environment.
I disagree. It's really that simple: he may have an expensive
scanner, but I think it's clear that he doesn't really know how to
use it. He also doesn't really know what processing is being done to
the pictures before he posts them, and it's clear that in one case a
GREAT deal more processing is being done than in the other. In the
end, most of what he's looking at is more a result of processing than
of the original photographs.
> A couple of points: 1) Darkroom processing probably loses at least as
> much resolution and contrast as a high quality scan because the MTF of
> the taking lens gets cascaded by the MTF of the enlarging lens -
> scanners also use imaging lenses, but they tend to be extremely well
> corrected, especially on a bona-fide 3200DPI high-end scanner;
I've compared the two, and the photographic enlargement is _clearly_
superior in every case I've seen. The only scanner I've seen that
even comes _close_ is a Howtek drum scanner with a resolution of some
thing like 8000 DPI. Even this doesn't _quite_ keep up with a really
photographic enlargement, but at least it's close enough that you
have to look very carefully to see that the scanner produces inferior
results.
Of course scanner lenses are well corrected, but so are good
enlarging lenses. I'd go so far as to say that if anybody knew a way
to make enlarging lenses substantially better than they are,
Rodenstock, Schneider, etc., would be VERY happy to make use of such
technology, so if somebody really knew how to make better lenses for
scanners, Schneider, Rodenstock, Nikon, etc., would apply the
technology to enlarger lenses as well.
> 2) If
> the web presents a ceiling on observable quality of two different photo-
> producing techniques, and if within that ceiling one of the techniques
> clearly appears better than the other, then exactly how would one
> arrive at an opposite conclusion by looking at actual prints? Wouldn't
> the diffferences perceived on the web become even greater once the
> ceiling is raised?
You're trying to put two things together that aren't really related,
and as a result coming to a completely erroneous conclusion.
If you start with a film camera, there are two things that govern the
quality of image you see on the web: one is the ultimate ceiling on
quality imposed by the medium of presentation. The other is the
deterioration in quality caused by the scanning process.
When you use a digital camera, one of these parts is eliminated.
This is why a good photographic print is typically a lot better than
anything you can get from a digital SLR, even though things may look
the opposite on the web.
> 3) The cascaded resolution of film and camera lens
> can definitely reach 100 lp/mm or perhaps a little more (IF you do
> everything very carefully), BUT, the contrast of the details beyond 50
> lp/mm or so becomes very low even in the best of circumstances - you
> can detect that it is there but it doesn't make an overwhelming
> impressing in a print.
This is a characteristic (primarily) of the lenses involved. Do you
honestly think that when you take an identical lens and put it on a
digital camera it's suddenly going to be a lot sharper and/or have
higher contrast than when the same lens is mounted on a camera that
uses film?
> The inherent advantage of digital image capture
> is that you don't have to go through any re-imaging steps in order to
> produce a print. If you have 50% contrast at 30 cycles you don't wind
> up cutting it in half or worse by projecting it or scanning it.
This is (more than) cancelled out by the fact that with a digital
camera you're starting with SO much worse quality that even if you
lose half the quality of what you had on film, it's still better than
the digital camera can manage, providing you start with properly
exposed film and scan it competently.
> The
> resolution cutoff in digital SLRs is lower, but the image contrast up
> to the cutoff is better, and that is one reason that digital images
> look so good when carefully compared to film images.
I _seriously_ doubt that this is even close to true. Good film can
record more information than most lenses can transmit, so for digital
cameras to record more information, the lenses would have to somehow
magically get better when mounted on digital cameras. It's true that
digital cameras use a smaller area at the center of the frame where
lenses are normally at their best, but this will only have a
relatively minor affect unless the lens _really_ falls off badly at
the corners.
> I will admit that
> I have no interest in slide projection or slides as art objects unto
> themselves without enlargement, but am instead primarily concerned
> about high quality prints. The other problem with film is that even an
> ultra-fine grained slide film like Provia has very disturbing
> granularity compared to images from the current crop of digital SLRs.
> The utter lack of grain is an overwhelming advantage of digital cameras
> in my view.
If you really want to get rid of grain from film, simply apply the
same kinds of image processing that's being done on the digital
pictures you see. Keep in mind that digital pictures start out as a
series of discrete pixels. Any time you see something that looks
reasonably "smooth", you're seeing the result of processing. Grain
in film also separates it into discrete pixels. The difference is
that if left unprocessed, the granularity of a digital image would be
SO bad nobody would even consider looking at it. The grain from slow
film is fine enough that it's objectionable under rather rare
circumstances, so most people think of it as "cheating" if they apply
the exact same techniques to it instead.
John Francis
Mark Spatny <vfxpr...@NOSPAMhotmail.com> wrote:
>
>What a monitor can display at any one time is limited by 8 bits per
>color channel, in a linear distrubution along the dynamic range.
Not necessarily. The graphics card in front of me has a 12 bits in,
10 bits out, gamma-corrected video stage. With a good enough monitor
I can get considerably better than a linear eight-bit response.
Anders Svensson
> A couple of points: 1) Darkroom processing probably loses at least as
> much resolution and contrast as a high quality scan because the MTF of
> the taking lens gets cascaded by the MTF of the enlarging lens -
> scanners also use imaging lenses, but they tend to be extremely well
> corrected, especially on a bona-fide 3200DPI high-end scanner;
You are here dismissing the fact that we are speaking of a
area of 24x36 mm vs far less. The notion that a enlarger
lens is cascading is wrong, and even the simple estimate
that it is added is not true either. We are speaking of
quality *loss*, not error adding.
> 2) If
> the web presents a ceiling on observable quality of two different photo-
> producing techniques, and if within that ceiling one of the techniques
> clearly appears better than the other, then exactly how would one
> arrive at an opposite conclusion by looking at actual prints?
Because a film negative, going thru' the process of printing
and digitizing is going down in quality. The intended end
result for a digital image is a digital (binary) image (say,
on a screen or a digital printer), the end result of a film
negative is a analogue image on a analogue media (a print).
If one of these media is lowered in quality until it meets
the other, and that happens at computer screen resolution, I
*know* which image is better quality to begin with.
Have you *ever* compared excellent, exhibition standard
prints with the best digital printouts from a digital camera
?
--
*************************
Anders Svensson
mail: anders.-.ei...@swipnet.se
*************************
Tony Spadaro
windows based or are we talking Cray?
Snorre A. Selmer
not be concidered as D30-pics, but GF-pics... This only shows that Genuine
Fractal is good... You _can_ make 2'x3' enlargements from 35mm film (yes
they will be grainy, but more than sharp enough)... Try that with the D30
(RAW, unprocessed D30 pics)... You can't...
--
Snorre A. Selmer
Helgeshaugen 25
3517 Hønefoss
NORWAY
Computer Consultant @ SRV in Oslo
http://setiathome.ssl.berkeley.edu/cgi-bin/cgi?email=sno...@statvoks.no&cmd=
user_stats_new
<bc1...@my-deja.com> skrev i melding news:8t63k0$p40$1...@nnrp1.deja.com...
> The following link will provide very interesting reading to anyone
> contemplating a jump from film to digital. WARNING: do not believe
> all the inflated numbers you hear about the image quality you can get
> from film. Look at images and ignore numbers. http://www.luminous-
> landscape.com/d30_vs_film.htm
>
> Brian
bc1...@my-deja.com
Anders Svensson <anders.-.ei...@swipnet.se> wrote:
> bc1...@my-deja.com skrev:
>
> > A couple of points: 1) Darkroom processing probably loses at least
as
> > much resolution and contrast as a high quality scan because the MTF
of
> > the taking lens gets cascaded by the MTF of the enlarging lens -
> > scanners also use imaging lenses, but they tend to be extremely well
> > corrected, especially on a bona-fide 3200DPI high-end scanner;
>
> area of 24x36 mm vs far less. The notion that a enlarger
> lens is cascading is wrong, and even the simple estimate
> that it is added is not true either. We are speaking of
> quality *loss*, not error adding.
>
not added (or better said, subtracted). IT IS MULTIPLIED. Using an
enlarging lens will reduce the contrast of any detail in the slide or
negative being enlarged. Assuming you use the enlarging lens at f8 and
that at f/8 it is diffraction-limited, the contrast of detail on film
at 100 cycles/mm will be cut in HALF. This is because the MTF of a
diffraction limited f/8 lens is 0.45 at 100lp/mm in green light.
Opening up to f/5.6 or f/4 will make things worse unless you have a
VERY atypical enlarging lens. If you can point me to an enlarging lens
that can exceed 50%contrast at 100lp/mm AT ANY APERTURE I would like to
know about it. The notion that the MTF of the enlarging lens must be
taken into account when calculating the final contrast of an image is
absolutely CORRECT. I did'nt even mention the MTF of enlarging paper
at 10 cycles/mm . . . . .
> > 2) If
> > the web presents a ceiling on observable quality of two different
photo-
> > producing techniques, and if within that ceiling one of the
techniques
> > clearly appears better than the other, then exactly how would one
> > arrive at an opposite conclusion by looking at actual prints?
>
> Because a film negative, going thru' the process of printing
> and digitizing is going down in quality. The intended end
> result for a digital image is a digital (binary) image (say,
> on a screen or a digital printer), the end result of a film
> negative is a analogue image on a analogue media (a print).
> If one of these media is lowered in quality until it meets
> the other, and that happens at computer screen resolution, I
> *know* which image is better quality to begin with.
I'm slightly confused about exactly what you are trying to say, so let
me make a statement and you can respond to it if you wish. If the
desired end product is an enlarged print, then digital cameras have a
built-in advantage over film. Digital prints are made using only ONE
optical step - namely the taking lens. Non-contact prints from film
require that the film must endure at least one additional image-
degrading optical step: namely, the film must either be enlarged using
an enlarging lens, or scanned with a film scanner. Film scanners use
lenses too, ya know, and there is also a drop in quality due to the MTF
of the linear CCD in the scanner. Now, this may seem unfair if you are
a film devotee, but it is the truth. The only way to eliminate the
second optical stage that I know of is to use a large format camera and
make contact prints, which I still occasionally do. Of course, there
is still the issue about the MTF of the printing paper . . .
>
> Have you *ever* compared excellent, exhibition standard
> prints with the best digital printouts from a digital camera
> ?
>
All of the exhibition standard prints that I have done in the past have
been from medium format, 4x5, 8x10 or 11x14 negatives. I do have a
number of high quality inkjet prints that I carefully prepared from
35mm slides and negatives using a 2700dpi scanner. With the exception
of prints made from technical pan film, all of these prints show
grain. Even Kodachrome 25. Prints I have made from Nikon D1 files
look better than my 35mm film scans because they show absolutely no
grain and have better shadow detail. The link given at the beginning
of this thread DOES provide a comparison of excellent digital and film
prints made by a very competent and well known photorapher. If you
haven't had a chance yet to study the link, you should - it would be
well worth your time. By the way, some of the best prints now made
from film are "digital printouts". Its getting really hard to justify
darkroom work anymore. Its also getting really hard to justify buying
film anymore - at least 35mm film.
Brian
> --
> *************************
> Anders Svensson
> mail: anders.-.ei...@swipnet.se
> *************************
>
>
bc1...@my-deja.com
want given the limitations of the printer, but you will be able to
clearly see the indidual square pixels in the image once the
enlargement becomes too great. What Genuine Fractals or any good
interpolation program does is to get rid of the stair-step effect when
you magnify a digital image too much. The effect is quite stunning
when you study it up close on a monitor, but it is clear that no
information is being added to the picture. Digital photographs that
have been processed through a good interpolation algorithm do not look
artificial to my eye. And they certainly don't insult me. By the way,
GF does not work nearly as well with blurry or grainy images. You have
to start with a good clean image with a minimum amount of noise. It
works much less well on images from small format digicams which contain
a noticeable amount of grain-like noise. For this reason, I suspect,
but haven't yet tried it, that GF would be significantly less effective
with film scans because of the large amount of noise (grain) in film.
In my opinion, the most interesting comparison between 35mm film and
digital would use the best possible technique in both cases. The
experiment done by Michael Reichman is one of the best comparisons I
have seen so far. I really don't think that eliminating the use of a
viable tool like Genuine Fractals would enhance such a comparison very
much. It would be like insisting that the film should be scanned with
a low-end scanner instead of a good one. Or enlarged with a really
cheap enlarging lens used wide open.
Brian
In article <8t8u76$78h$1...@snipp.uninett.no>,
John Francis
Tony Spadaro <tony_to...@my-deja.com> wrote:
>In article <8t7rfc$2qu3g$1...@fido.engr.sgi.com>,
> jfra...@dungeon.engr.sgi.com (John Francis) wrote:
>> In article <39f7627c...@news.earthlink.net>,
>> Mark Spatny <vfxpr...@NOSPAMhotmail.com> wrote:
>> >
>> >What a monitor can display at any one time is limited by 8 bits per
>> >color channel, in a linear distrubution along the dynamic range.
>>
>> Not necessarily. The graphics card in front of me has a 12 bits in,
>> 10 bits out, gamma-corrected video stage. With a good enough monitor
>> I can get considerably better than a linear eight-bit response.
>>
>>
> I want one! I'll bet I can't afford one though. Seriously - is this
>windows based or are we talking Cray?
Neither - we're talking a Silicon Graphics Octane2 desktop workstation.
(Crays never had particularly good graphics systems, anyway).
It's not really a cost-effective 2d graphics card, though.
It has a lot of very nice 3D OpenGL features which add
considerably to the price (although the support for filter
kernels means some simple image enhancement operations
go like a bat out of hell). The graphics chip itself is
about the same size (and complexity, and power use) as a
Pentium III. It costs us considerably more, though: we
don't attain the same economies of scale that Intel can.
Vadim Kouzmenko
scanning techniques. I would say poor techniques...
But what does it have to do with Provia 100F?
Anders Svensson
If you lose 50 % in step one (photography) and then lose 50% of that in the
enlarging process will mean that you lose 25 % in the whole process - that
is the correct way of adding up errors and I assume that you mean that when
you say multiply.
If you prefer to use the theoretical limit of diffraction as some kind of
benchmark, think about the effect of size of the active image area (all LF
photographers favourite argument against 35 mm's).
Secondly, don't dismiss that photography print making is a highly complex
process with dynamic image enhancing operations (chemical and physical) in
each stage.
Many of these are very similar to some Photoshop filters, obviously modeled
upon age old developing and copying methods. Just a memento and really a
point to make for those who believe photo is "clean" and digital
manipulation is "dirty". There are many similarities to digital enhancement
available to chemical photographers.
As for the rest of your deductions, I feel indifferent. I have analyzed the
quality bottlenecks in my feeble digital attempts to be firstly the the
print process and a distant second, the scan process.
I see it as completely reasonable that a digital camera has no grain.
BTW, I use pixel free film :-)
AlexZN
I don't understand where you got this purist attitude. I do belive that
the comparison has to be made between two images processed using the
standard reasonable workflow technique, which is exactly waht has been
done. It is the final result (i.e. the appearance of a final print)
that matters at this case. You don't "unsharp mask" your images that
you get froma film scanner? I don't believe you. Waht you are trying
to propose is like diong a comparison between an F5 and Pentax MX while
insisting that F5 should be kept in manual focus mode and use
centerweight metering. While "scientifically" pure, it is unrealistic
and doesn't tell you anything about the real world application. I
think this was so far the most illuminating comparison between digital
and film that I have read so far.
BTW, no one has commented on the fact that if you can get comparable
quality from a digital SLR, you immediately start saving large amounts
of money by eliminating film and processing cost. Michael has noted
that in his excellent review as well. I bet in a couple of years with
the cost of the digital bodies going down and the quality of the inkjet
printers and CMOS sensors going up, this might be a deciding factor.
Alex
In article <8t7lqi$4g1$1...@nnrp1.deja.com>,
Tony Spadaro <tony_to...@my-deja.com> wrote:
> I've read it. THe test is more than flawed, it is DOA. Without
taking
> into consideration the many alterations he made to both images, there
> cannot be a comparison. I've printed 3.3 megapixel images. They
> certainly can compete with film for small prints. Even smaller digital
> pics can be used for magazine reproduction. Anything with the rough
> shape of the subject will do for newspapers. At 8x10 even my two
> generation old printer shows that the digital image is not continuous.
> When you read terms Unsharp Mask, Levels, Geniune Fractals, you know
> these are not straight images. Straight images made of close
subjects -
> preferably with a range of colours, and shades of the colours, might
be
> more believable. I still would want to see the prints, however.
> Digital is catching up with film, and it looks like the CMOS sensor
> may very well be a viable alternative to CCD. This much is true. The
> rest is applesauce topped with bull, sprinkled with hooey.
>
> --
> Digital photo restoration in autumnal Chapel Hill NC
> http://www.homeusers.prestel.co.uk/magor/tony
> New - A Digital Workflow + Sharpness: Threat or Menace?
> And Selecting Your First SLR - Java must be enabled
>
jw&a
> been from medium format, 4x5, 8x10 or 11x14 negatives. I do have a
> number of high quality inkjet prints that I carefully prepared from
> 35mm slides and negatives using a 2700dpi scanner. With the exception
> of prints made from technical pan film, all of these prints show
> grain. Even Kodachrome 25. Prints I have made from Nikon D1 files
> look better than my 35mm film scans because they show absolutely no
> grain and have better shadow detail.
are you saying you can see the grain in all the 35mm prints, or in _all_ the
prints?
i made some prints from 120 plus-x and the one thing that stands out to me
is the complete lack of grain.
the better shadow detail part is interesting. is there a "characteristic
curve" for CCDs?
your extreem aversion to grain seem to be the deciding factor for you.
bob rogers
south carolina
jw&a
<ji...@my-deja.com> wrote in message news:8t7iei$1bm$1...@nnrp1.deja.com...
>
> One important aspect of Mr. Reichmann's comparison is that the scan of
> the Provia F transparency was presented without any sharpening. It's my
> understanding that scanning film in and of itself introduces
> unsharpness, which is corrected by subsequent unsharp masking. It is not
he also says he used the default settings on the scanner. i don't know about
that particular scanner, but with my scanner i get better or worse scans
with different things depending on the settings!
bob rogers
south carolina
Mark Spatny
Francis) wrote:
>Not necessarily. The graphics card in front of me has a 12 bits in,
>10 bits out, gamma-corrected video stage. With a good enough monitor
>I can get considerably better than a linear eight-bit response.
I see in your post that you are using an Octane 2 with a video card
I'm not familiar with. Is the vide card a recent development?
I've used Octanes and Onyxs with both SGI and Barco monitors, and
haven't seen any display like you've described.
smitbret
-Brett
> The following link will provide very interesting reading to anyone
> contemplating a jump from film to digital. WARNING: do not believe
> all the inflated numbers you hear about the image quality you can get
> from film. Look at images and ignore numbers. http://www.luminous-
> landscape.com/d30_vs_film.htm
>
> Brian
>
John Francis
Mark Spatny <vfxpr...@NOSPAMhotmail.com> wrote:
>On 25 Oct 2000 23:47:56 GMT, jfra...@dungeon.engr.sgi.com (John
>Francis) wrote:
>
>>Not necessarily. The graphics card in front of me has a 12 bits in,
>>10 bits out, gamma-corrected video stage. With a good enough monitor
>>I can get considerably better than a linear eight-bit response.
>
>I see in your post that you are using an Octane 2 with a video card
>I'm not familiar with. Is the vide card a recent development?
Yes - it's the VPro graphics card released with the Octane 2s.
(Not to be confused with the VPro graphics cards in the Intel-based
SGI workstations; SGI marketing definitely moves in mysterious ways).
It is (or will be) available as an upgrade in existing Octanes,
but you may need to upgraded the power supply at the same time.
It isn't available for anything but an Octane.
It's primarily an OpenGL card, but the frame buffer configurations
available include 48-bit RGBA. I've seen one hooked up to the
1600SW flat-panel display (1600x1024) - that looks pretty nice.
Tony Spadaro
start by crippling A and optimizing B - it ain't a test. Buy a digi-cam
if you wish. I intend to buy when I can get an EOS body with
6.6megapixels. I have printed from 1, 2 and 3 megapixel cameras. They
are not up to film quality - although in the case of 3.3 it can get
close as long as 8x10 is the largest you intend to print, and that is
not of a highly detailed image.
If you have read my past posts about image size and viewing distance,
you know I do not consider 18 inches to be the proper viewing distance
for 16x20. Some photographers put their nose on the image to view - I
take more of a painter's viewing distance. So when I say 8x10 is the
max for 3.3 it is not the same as saying 8x10 would be the biggest to
ever blow up 35mm film - as many med format people claim.
This character could even be right. 3.3, with a better contrast range
than I've see in digi-cams might actually look better than Provia F
100. I have serious doubts, but it could be. However. This "test" is
not proof of anything - anything at all. It's bull, topped with a
dollpo of applesauce, sprinkled with hooey and served on a bed of
bushwaa.
Tony Spadaro
out of my budget range, (although probably only a tenth the price of
the cray) but one can dream. I do have ambitions of working with 3d
someday though. I would love to create virtual tours of places that no
longer exist - using the photographs that do.
It would be fascinating to reconstruct Paris as seen by Marville,
which would be very much the same as described by Victor Hugo, and
then Atget's city, or San Francisco before and after the quake. I'd
even like to re-create the Western town sets that once existed on movie
lots. I went to the movies every Saturday while a kid, and probably
know those places as well as my home town.
bc1...@my-deja.com
between the current crop of 3.3MP digicams based on a 1/1.8" sensor and
the 2.6-3.3MP digital SLRs which have much larger sensors. I have both
a Nikon 990 and a D1, and even though the D1 has LESS resolution on
paper, in practice the images are much sharper and clearer. I am
saying this based on personal observation, of course. The small sensor
cameras are great, but they have a significant amount of noise that
looks a lot like film grain. They are also limited to 8bits per
color. The Nikon D1 and Canon D30 have essentially no noise at ASA200
equivalent and they can capture 12 bits of depth per color. I find the
extra bit depth makes a HUGE difference in my ability to extract
whatever shadow detail I desire from my RAW images. I used to scan
slides (Kodachrome25 and Kodachrome 64) with a 12bit per color 2700dpi
scanner in an attempt to do the same thing, but I was always
disappointed with the results.
Like you, I am also looking forward to a 6+MP digital SLR that will
accept all of my lenses. Based on my experience with a high quality
3MP camera, I don't think that such a camera is required to eliminate
the need for 35mm film - it has already happened! When 6MP arrives I
will have a hard time justifying the use of my medium format equipment.
Brian
In article <8taa0e$832$1...@nnrp1.deja.com>,
John Francis
Tony Spadaro <tony_to...@my-deja.com> wrote:
> Silicon Graphics is what I meant. I cross things up frequently. Still
>out of my budget range, (although probably only a tenth the price of
>the cray) but one can dream. I do have ambitions of working with 3d
>someday though. I would love to create virtual tours of places that no
>longer exist - using the photographs that do.
You would have loved SIGGRAPH this year, then. One of the papers
was on a project to do exactly that - reconstruct a virtual 3D
environment of a historical site. In their case, however, it was
an archaelogical site, so no contemporary photographs existed.
> It would be fascinating to reconstruct Paris as seen by Marville,
>which would be very much the same as described by Victor Hugo, and
>then Atget's city, or San Francisco before and after the quake. I'd
>even like to re-create the Western town sets that once existed on movie
>lots. I went to the movies every Saturday while a kid, and probably
>know those places as well as my home town.
It won't be too long before you'll be able to get a pretty good 3D
environment on your home PC - the games industry is driving the
price of 3D cards ever lower. But the modelling will still be
pretty expensive; man-hours aren't cheap unless you do it yourself.
OneThumb
>
> I've read it. THe test is more than flawed, it is DOA. Without taking
> into consideration the many alterations he made to both images, there
> cannot be a comparison. I've printed 3.3 megapixel images. They
> certainly can compete with film for small prints. Even smaller digital
> pics can be used for magazine reproduction. Anything with the rough
> shape of the subject will do for newspapers. At 8x10 even my two
> generation old printer shows that the digital image is not continuous.
> When you read terms Unsharp Mask, Levels, Geniune Fractals, you know
> these are not straight images. Straight images made of close subjects -
> preferably with a range of colours, and shades of the colours, might be
> more believable. I still would want to see the prints, however.
> Digital is catching up with film, and it looks like the CMOS sensor
> may very well be a viable alternative to CCD. This much is true. The
> rest is applesauce topped with bull, sprinkled with hooey.
I though Mr. Reichmann was fairly clear about what he did and the
results he produced. Given what he's done, which is reasonable and no
more than what most of us would do, I believe it. I don't understand
your point unless it's to question Mr. Reichmann's competence. Being
familiar with his work, I believe it when he says he is able to produce
stunning prints.
Tony Spadaro
I question his methodology. He may very well be able to make stunning
prints, but there is no way he can show those stunning prints on the
web. The pictures are of distant and fairly monochrome buildings,
without a lot of colour variation on the surfaces, and he used filters
on his images. This is not a comparison. There are too many variables
to make any valid conclusion from what he has posted.
I'm not familiar with his work. I never heard of him before. I have
no personal ax to grind with him, and he can use any method he likes to
make his pictures. He also has the right to make any statement he
wishes about the quality. That is all fine with me.
I too have rights however, and one of them is the right to point up
the flaws in his methods, his findings, the appropriatness of his
subject, and his recommendations.
Annika1980
>Newsgroups: rec.photo.equipment.35mm
> This "test" is
>not proof of anything - anything at all. It's bull, topped with a
>dollpo of applesauce, sprinkled with hooey and served on a bed of
>bushwaa.
I thought it was done about as well as it could be. What would you do
differently? You claim the test was bogus, yet you say some of the same things
the author says .... that at 8x10 or less, digital (the D30) is at least equal
to film. Of course, he claims that at 8x10 or less the D30 is superior.
I'll be interested in a comparison between the D30 and the Nikon D1. They say
the D30 wasn't designed to compete with the D1, but I say, "Why not?"
One reason I got my EOS-1V was that I wanted the best body I could get for what
would surely be my last film camera purchase. Like you, I'll be watching
closely until the 6+ MP digital cammys come out. Then, BAM!
Tony Spadaro
annik...@aol.com (Annika1980) wrote:
> >From: Tony Spadaro tony_to...@my-deja.com
> >Newsgroups: rec.photo.equipment.35mm
> I thought it was done about as well as it could be. What would you do
> differently?
Raw scan, saved and printed. Raw picture from the D30, saved and
printed - to 4x6, 8x10, 12x18 and if possible larger. No Unsharp
masking, no levels, no Genuine Fractals. THe subject should contain
areas of "busy" detail, a human face - fairly close, natural light.
There should be a minimum of 6 pictures one each shot in flat light,
low light, and very contrasty light with each camera.
I don't know a way to get around the lens magnification difference
without adding complication, so I would do it by using a fixed focal
length, and moving the camera to keep the main subject the same size on
the film/sensor. This means the background will change but exact
framing is not necessary.
For the subject, a blonde, a redhead, and a brunette - of either sex
would be good, especially if they were wearing print dresses - or
hawaiian shirts - here is the detail, the faces give skin tones. If
there were a Macbeth Colour Checker in the scene it would also help.
The background of at least one of the shots should be out of focus. A
nice fat tree with filtered light coming through would be nice but it's
getting a bit late in the year for that. It would be nice to see how
this type of highlight compares between the digital and the film. Some
would say this is a function of the lens only but the digital pictures
I've worked on have glaring white highlights, which indicated to me a
lower contrast range than slide film. I would like to know if the CMOS
sensor has a better response.
An enlargement of the darkest areas on the high contrast shot would
allow one to look at the noise level of the digital, and the scanner -
Fuji Provia F 100 does not have any noise, of course, but there will be
processing flaws, bits of debris, etc.
I have scanned a bit of Provia F and my scanner (LS2000) at 2700dpi
does not show any grain. The scanner he works with may or may not show
it. 4000 dpi scanners most likely will.
You claim the test was bogus, yet you say some of the same things
> the author says .... that at 8x10 or less, digital (the D30) is at
least equal
> to film. Of course, he claims that at 8x10 or less the D30 is
superior.
I said at LESS that 8x10, and "possibly" EVEN at 8x10 for a
3.3megapixel image.
>
> I'll be interested in a comparison between the D30 and the Nikon D1.
They say
> the D30 wasn't designed to compete with the D1, but I say, "Why not?"
The D30 is basically an Elan (EOS 50 or 30) body. It is not a
professional camera - which is neither here nor there. The image
quality of an Elan with a given lens will usually be the same as an EOS
1 with the same lens. It is in features that the I outstrips the Elan,
including the weather sealing.
bc1...@my-deja.com
I know what you mean about 120 Plus-X. I have made 16x20 prints from
some of my 6x6 Plus-X negatives that look as good as many similar sized
prints I have made from 4x5.
I think that CCDs have a fairly linear response curve throughout their
dynamic range, but I have not seen any measured data to confirm or deny
this. It does take a little extra work to extract the shadow detail
from 36-bit (12-bit per color) RAW images. What I do is open the file
twice, and adjust it in one case for hightlight detail and in the other
case for shadow detail. I then save both images as TIFF files. Then
in Photoshop I open both files and combine them using a layering and
masking technique. The results are almost always superior to images
that have not been combined in this way. With 36-bit files I am always
amazed by how much shadow detail is available. The challenge is to
reveal that detail in an aesthetically appropriate way. If you want a
detailed description of how I combine images just let me know.
You are right about my aversion to grain. It is the reason I switched
from 35mm to medium and large format 15 years ago, and is the reason I
am so excited about the results I am getting from my digital SLR. I
think that one overwhelming fact that gets lost in the film vs digital
debate is that photography is probably more intensely interesting now
than it has been since the introduction of the daguerrotype process.
Its truly a golden age.
Brian
In article <svh5lel...@corp.supernews.com>,
Glenn Woodell
>
>The following link will provide very interesting reading to anyone
>contemplating a jump from film to digital. WARNING: do not believe
>all the inflated numbers you hear about the image quality you can get
>from film. Look at images and ignore numbers.
I'd like to see both of them printed out!
Annika1980
>I think that CCDs have a fairly linear response curve throughout their
>dynamic range, but I have not seen any measured data to confirm or deny
>this. It does take a little extra work to extract the shadow detail
>from 36-bit (12-bit per color) RAW images. What I do is open the file
>twice, and adjust it in one case for hightlight detail and in the other
>case for shadow detail. I then save both images as TIFF files. Then
>in Photoshop I open both files and combine them using a layering and
>masking technique. The results are almost always superior to images
>that have not been combined in this way. With 36-bit files I am always
>amazed by how much shadow detail is available. The challenge is to
>reveal that detail in an aesthetically appropriate way. If you want a
>detailed description of how I combine images just let me know.
>
I'd like to know that!
Annika1980
>I'd like to see both of them printed out!
>
>
Are you implying that if one image looks better prior to printing, the other
image will end up looking superior? Don't think so.
I don't understand why some people have vehemently decried the guy's test or
test procedures. As far as I know, he isn't selling digital cameras. He is a
serious photographer. Almost everybody knows that digital will eventually
replace film for 99% of all photography. All his test shows is that that day
may be closer than we think.
Annika1980
Thanks for your detailed reply. Have you tried communicating your
suggestions to the author of the test? Perhaps he could incorporate your
suggestions in a future test.
I would question one thing, though, about what you proposed.
"No Unsharp masking, no levels, no Genuine Fractals."
This is certainly a "purist" view, but I don't think it reflects "real world
use" as someone else pointed out. Maybe if you work strictly with slides you
can avoid post-processing, but otherwise decisions will be made which will
affect the final output.
The goal for any format should be to get the best result you can get. If you
can end up with a better final output using digital then what's wrong with
that?
Also, his subject choices may differ from yours because he is using scenes he
is familiar with. Someone with a background in portraiture may choose
differently. I don't think this makes his test invalid.
Go with what you know.
bc1...@my-deja.com
I wrote the following for another forum a couple of weeks ago, and am
just pasting it in here. It describes how I extract more shadow detail
from my 12-bit D1 images using Photoshop and a specialized (but really
good and really cheap!) D1 program called Bibble. I use Bibble to open
up RAW files from my D1 and save them as 8 or 16 bit TIFF files.
Bibble has an exposure adjustment slider tool that allows me to bias
the image for either the highlights or the shadows. I assume that you
could do the same thing with a Canon D30, but I don't have any
experience with that camera or the software needed to interpret its RAW
files. Hopefully, the same guy who writes Bibble for the D1 will write
a similar program for the D30. You can adapt the same technique to
work with film by scanning the film twice using a different exposure
for each scan. I have heard of a similar technique used with two
separate camera exposures, but this would create a lot of registration
problems I would think. Anyway, here goes, and my apologies in advance
for any overabundant enthusiasm for the Nikon D1 - I know you are a
Canon fan!:
One of the things I really like about the D1 is that the RAW files have
12bits per color. This means that there is a lot of shadow detail that
can be taken advantage of. To get at it requires shooting in RAW mode
to get a NEF file and then using the following Photoshop/Bibble
technique: 1) Open the NEF file in Bibble and adjust the exposure
slider to give good shadow detail, and save this image as an 8-bit TIFF
file (call it ShadDetail.TIF for the sake of discussion). This image
will probably look a little too light; 2) Open the NEF file again in
Bibble and adjust the slider to give good highlight detail. The image
will probably look a little dark, or it may look reasonably normal
except that the shadow detail is blocked up. Save the file as an 8-bit
TIFF file and call it HighDetail.TIF for the sake of discussion.; 3)
Open both TIFF files in Photoshop (version 4 or higher) and make a
duplicate copy of ShadDetail; 4) Paste a copy of HighDetail into a
layer on top of ShadDetail; 5) In this two-layer file open up the
channels window and create a new channel and paste a copy of ShadDetail
into this new channel. It will show up as a black and white image. 6)
Load the new channel as a selection (use the bottom left hand button in
the channel window), then switch to layers in the same window, select
the top layer and then add a layer mask using the "reveal selection"
option; 7) flatten the image down to a single layer and make minor
adjustments using curves or levels as desired. The resulting image will
be a beautiful blend the shadow and highlight detail that cannot be
achieved by manipulating curves or levels alone. I find that almost
every image I take is enhanced by this technique.
I find that this techniqe works well as long as the two images are
separated by less than two stops of exposure in the Bibble slider. If
meaningful shadow and highlight detail are require a wider separation,
then I use a multi-step procedure in which I gradually add more and
more shadow detail. On one image taken of a doorway from inside a dark
room in a fortress I was able to use 4 steps to really bring out good
shadow detail while maintaining highlight detail.
I used to use this technique to bring out shadow detail when scanning
kodachrome slides with an LS-1000 scanner, and I learned about it in a
book called "Make Your Scanner a Great Design and Production Tool" by
Michael Sullivan, North Light Books, Cincinnati, 1998, pages 96-97.
You may find that the shadows contain noise. I frequently use the noise
filter in Bibble on the ShadDetail images to eliminate this.
Brian
In article <20001027110849...@ng-md1.aol.com>,
Vadim Kouzmenko
(Glenn Woodell) wrote:
And I'd like to see them both on a light box under fine 10x Schneider
loupe! No kidding! There are film recorders anyway. :)
Or lets compare projected slide to projected digital snap (LCD
projector comes to mind).
At least it would be more accurate than observing few tiny compressed
jpegs of questionable origin.
bc1...@my-deja.com
I think your comments are very well put, and I completely agree. One
thing to consider about portraits is that digital cameras are extremely
well suited for making them. I am speaking from personal experience as
well as from reading the reports of others. Many people might complain
that a portrait is not a fair basis of comparison because it would bias
the conclusions in favor of digital! For a good example of a high
quality digital portrait check out the full page (bigger than 8x10) D1
ad in the October (or perhaps September) issue of Photo District News.
Stunning. From everything I've heard, the D30 should be able to do at
least as well if not better.
Brian
In article <20001027113636...@ng-md1.aol.com>,
bc1...@my-deja.com
shows the MTF at f/8 for one of their Apo-Rodagon enlarging lenses.
Its a 105mm f/4, and is evidently pretty good or else they wouldn't be
bragging about it. It has an MTF of about 60% at 40lp/mm and about 80%
at 20lp/mm. The graph does not show anything higher than 40lp/mm, but
I would estimate that this lens might might image 100lp/mm at around
25%, and certainly much less than 50%.
Brian
In article <39F87495...@swipnet.se>,
Tony Spadaro
just as mad as I am. I buy the 3d magazines every time there is a
reconstruction of an archiological site \pictured - and once an issue
with Frank Llord Wright buildings that are gone. I'm big on
architecture.
Vadim Kouzmenko
>You are right about one thing. The loss due to the enlarging lens is
>not added (or better said, subtracted). IT IS MULTIPLIED. Using an
>enlarging lens will reduce the contrast of any detail in the slide or
>negative being enlarged. Assuming you use the enlarging lens at f8 and
>that at f/8 it is diffraction-limited, the contrast of detail on film
>at 100 cycles/mm will be cut in HALF. This is because the MTF of a
>diffraction limited f/8 lens is 0.45 at 100lp/mm in green light.
>Opening up to f/5.6 or f/4 will make things worse unless you have a
>VERY atypical enlarging lens.
Diffraction limit of f/5.6 is about 0.6 @ 100 lp/mm.
But what's your point? Why does 100 lp/mm disturb you? D30 MFT
response must be close to 0 at frequencies 20 lp/mm and above. Later
is equivalent to about 12 lp/mm in 35mm format.
At the same time RDPIII responds well up to 50-60 lp/mm.
>If you can point me to an enlarging lens
>that can exceed 50%contrast at 100lp/mm AT ANY APERTURE I would like to
>know about it. The notion that the MTF of the enlarging lens must be
>taken into account when calculating the final contrast of an image is
>absolutely CORRECT. I did'nt even mention the MTF of enlarging paper
>at 10 cycles/mm . . . . .
I'm confused about exactly what you are trying to say by mentioning
enlarging paper MTF. Does digital printing give better MTF on paper? I
doubt. Or maybe monitor CRT tube has better resolution (2 lp/mm)?
>I'm slightly confused about exactly what you are trying to say, so let
>me make a statement and you can respond to it if you wish. If the
>desired end product is an enlarged print, then digital cameras have a
>built-in advantage over film.
I can't agree with you. In my opinion:
- If the desired end product is print, negative film is still MUCH
BETTER in terms of dynamic range, gamut and resolution.
- If you wish to get the best you could get, then transparencies is
the way to go.
- If the desired end product is digital file, or if you can't justify
processing time (e.g. working for Reuters), then digital is the
answer.
>Digital prints are made using only ONE
>optical step - namely the taking lens.
I would rephrase. Digital prints are made using only ONE optical step
PLUS numerous digital transformations (digitization, interpolation,
color space transformations, adjustments, unsharpening, resampling,
dithering, compression, etc.).
>Non-contact prints from film
>require that the film must endure at least one additional image-
>degrading optical step: namely, the film must either be enlarged using
>an enlarging lens, or scanned with a film scanner. Film scanners use
>lenses too, ya know, and there is also a drop in quality due to the MTF
>of the linear CCD in the scanner. Now, this may seem unfair if you are
>a film devotee, but it is the truth. The only way to eliminate the
>second optical stage that I know of is to use a large format camera and
>make contact prints, which I still occasionally do. Of course, there
>is still the issue about the MTF of the printing paper . . .
So all we need is a decent enlarging lens. Huh?
Tony Spadaro
> >detailed description of how I combine images just let me know.
> >
>
> I'd like to know that!
>
or in some cases simply using the layers capability of most image
processors. At my website, the first article under "Photoshop Tips"
shows a method involving masks and layers. On the links page I have
links to other sites with more complex methods. The best reference on
it is probably the book "Photoshop Channel Chops" which goes into all
the various methods of combining images.
I recently did some experiments with expanding the contrast range of
Provia F 100 using two shots, and two scans combined. The idea was to
make a full range B/W image using colour slide film. The results were
okay, but it's not exactly a fast way to either shoot or print.
--
Digital photo restoration in autumnal Chapel Hill NC
http://www.homeusers.prestel.co.uk/magor/tony
New - A Digital Workflow + Sharpness: Threat or Menace?
And Selecting Your First SLR - Java must be enabled
Mark Spatny
Francis) wrote:
>Yes - it's the VPro graphics card released with the Octane 2s.
>(Not to be confused with the VPro graphics cards in the Intel-based
>SGI workstations; SGI marketing definitely moves in mysterious ways).
>
>It is (or will be) available as an upgrade in existing Octanes,
Thanks. I'll be looking into that.
jw&a
<bc1...@my-deja.com> wrote in message news:8tc9a4$qi7$1...@nnrp1.deja.com...
> Hello Annika1980:
> I wrote the following for another forum a couple of weeks ago, and am
> just pasting it in here. It describes how I extract more shadow detail
> from my 12-bit D1 images using Photoshop and a specialized (but really
i've done something sort of similar. i had an really dark print that my boss
took of a building at sundown. i put the print on the scanner and made two
scans in "billions of colors" mode (32 or 36 bits, i think). i adjusted the
input/output curve in the scanner driver. then i combined them in photoshop
as two different layers.
the final printed output amazed everyone in the office, because in the 4x6
chemical print, there were areas that were just plain solid black, but in my
printed output, there were colored details (green leaves, red bricks, etc).
my boss was particularly amazed "but you can't get detail when there is
*none* in the original"
bob rogers
south carolina
Tony Spadaro
"real world
> use" as someone else pointed out. Maybe if you work strictly with
slides you
> can avoid post-processing, but otherwise decisions will be made which
will
> affect the final output.
This is being presented as a comparison test. For a comparison test
you have to do a comparison right? If not, I can tell you that my
Subaru is capable of going faster than a Porsche - I was in the right
lane, he was caught behind a school bus. Ergo, Subaru is a better car
than Porsche.
> The goal for any format should be to get the best result you can
get. If you
> can end up with a better final output using digital then what's wrong
with
> that?
Nothing. But this is being presented as a review, and the conclusion
is being made public. The claim is not backed up by the research. Even
the author of the test points out that Genuine Fractals is needed to do
prints larger than 6x9 from his digital camera. Anyone who shoots
Provia F100 will tell you that Genuine Fractals is not needed for 12x18
- four times the area of a 6x9.
>
> Also, his subject choices may differ from yours because he is using
scenes he
> is familiar with. Someone with a background in portraiture may choose
> differently. I don't think this makes his test invalid.
> Go with what you know.
No one is talking portraits. Skin tones are important however, and a
human is a movable test subject.
bc1...@my-deja.com
detail out to 100lp/mm, the combination of film, taking lens and
enlarging lens pretty much reduces the visibility of that detail on the
final print to nearly nothing. First rate enlarging lenses are by no
means diffraction limited, even at f/8. A recent Rodenstock ad showing
the MTF curves of their 105mm f/4 Apo Rodagon indicates a contrast of
about 60% at 40 cycles when the lens is stopped to f/8. (see PHOTO
Techniques, July/August 2000 page 5) This of course is the theoretical
MTF of the lens design. I would hazard a guess that most enlarging
lenses fall short of this level of performance in the real world. I
personally feel that in 35mm format the only truly meaningful spatial
resolutions are those below 50 cycles per millimeter.
By the way, the D1 and the D30 digital cameras have a pixel size of
approximately 12 microns, which translates into about 40 cycles per
millimeter (corresponding roughly to 25cycles per millimeter on 35mm
format), not 20. I can detect a very slight amount of color aliasing
on my D1 for certain types of image detail, which indicates that the
MTF at 40 cycles due to the sensor+antialiasing filter is not quite
zero. I have photographed the PIMA/ISO resolution test chart with all
of my Nikkor lenses at all apertures on my D1, and the lines
corresponding to 20 cycles per millimeter are very clearly imaged in
all cases.
It is true that high quality film has more detail than current 3MP pro
digicams, BUT, that detail does not really reach out and grab you
because it is nearly invisible. You really have to look closely to see
it. The detail that is shown in good digital images is clearer,
particularly in shadow areas, and it is not detracted by grain. Thats
why digital images are able to look so good in comparison to film IMHO.
By the way, what is RDPIII?
Maybe we have really different eyes, but to me, prints from film are
definitely not MUCH better than prints from a good 3MP digicam. To me
they look quite similar except for the lack of grain in the digital
print.
And since you are so upset by digital processing, have you ever
considered the chemical complexity of what is happening in film and
enlarging paper? All that chemistry can cause physical harm to
darkroom workers and pollute the environment. You seem to have a
visceral dislike of digital imaging, but aren't the end results really
what count? As for myself, I like optics and photographs.
Brian
In article <39fca4c1...@enews.newsguy.com>,
Annika1980
>But this is being presented as a review, and the conclusion
>is being made public. The claim is not backed up by the research. Even
>the author of the test points out that Genuine Fractals is needed to do
>prints larger than 6x9 from his digital camera.
>Anyone who shoots
>Provia F100 will tell you that Genuine Fractals is not needed for 12x18
>- four times the area of a 6x9.
So then what?
You and the author are in agreement then .... at small sizes digital wins, at
large sizes film wins.
Annika1980
>> I'd like to know that!
>>
>> It's a common technique, and can be done using either multiple scans
>or in some cases simply using the layers capability of most image
>processors. At my website, the first article under "Photoshop Tips"
>shows a method involving masks and layers.
Thanks. The method you describe on your page is pretty much what I usually do.
The other poster mentioned using two identical layers and applying different
corrections to each one. OK, so far, so good. What I'd like to know is how he
gets them back together again. Does he simply mask out the unwanted areas of
the top layer?
Andrew Dacey
bc1...@my-deja.com wrote:
>
> The following link will provide very interesting reading to anyone
> contemplating a jump from film to digital. WARNING: do not believe
> all the inflated numbers you hear about the image quality you can get
> from film. Look at images and ignore numbers. http://www.luminous-
> landscape.com/d30_vs_film.htm
>
> Brian
>
> Sent via Deja.com http://www.deja.com/
> Before you buy.
I already brought up some discussions regarding his test methods in
rec.photo.digital, but I thought I'd do a quick summary here as well
just to foster discussion.
In my mind, you can't compare film and digital cameras without doing a
chemical print of the slide. The reason for this is because it's
currently accepted as the best method for making a print from a slide.
What I'd like to see done would be as follows:
1. Chemical print (preferrably Ilfochrome) of the slide.
2. Raw scan of the slide printed with basic printer settings.
3. Raw digital camera file printed with basic printer settings.
4. Scan of slide with adjustments made to attempt to maximize quality,
printed with settings with the same goal in mind.
5. Digital camera file printed with optimized quality in mind (this
would be the only print that I'd allow Genuine Fractals to be used with).
The reason that I include the need for the chemical print is because it
is supposed to be the best quality print possible. If you want to argue
that digital is a better process, you need to compare it with the best
non-digital process. Scanning the slide will likely cause a reduction in
quality. You can't then compare a print from that digital file and make
any conclussions regarding digital vs. film. You can only draw
conclussions regarding digital vs. scanned film.
The reason I want 2 prints of the digital files is because of the need
for optimization (usually). This is likely different between a digital
camera and a scanned slide. So if the digital camera needs less
adjustment but you don't perform any adjustments, then you'll list the
digital camera as winning, when it may not be an actual reflection of
the quality that can be achieved. For one, I find that every slide needs
the levels adjusted. Auto levels does a decent job sometimes, and
sometimes does a great job. But other times, it looks like crap. I'd be
much more appreciative of seeing what results could be achieved from a
properly adjusted scan.
He has also performed different adjustments to the digital camera's
image that were not applied to the slide. The main one would be the use
of genuine fractals. I don't object to it being used for an optimized
print, but when no such optimizations are made for the slide, then you
can't make a comparison.
I think my test suggestions covers all the concerns. You have what's
considered to produce the best possible print from a slide (the chemical
print) and you put it up against a number of different digital methods.
You can also gain a lot of different info. For instance, you can see
what the affect of genuine fractals has on the resulting images (because
you can compare a print with and without it being used). You can also
look at how much of a difference optimization makes. I suspect that you
may get truer colours in a raw file from the digital camera. If that's
the case, then it will definately look better than the scanned slide. It
would be interesting to see what the results would be. You can then
start making specific recommendations for which technique is the best
for different situations. For instance, if the digital camera looks
great without adjustments, then that's a good suggestion for people that
are unskilled in photoshop. If genuine fractals makes a huge difference
in the quality, then that's a good suggestion for people to use in order
to get bigger prints. And so on.
--
Andrew "Frugal" Dacey,
fru...@tildefrugal.net
http://www.tildefrugal.net/
Tony Spadaro
images and most techniques. There is also combining layers, masks and
layer masks using the Layers palette modes (Normal, Multiply, etc). In
Photoshop, the Image menu commands (I don't know which other image
editors have them) Apply Image, and Calculations, plus the newest one
Channel Mixer have modes not available from the palette, and can
combine percentages in a way not possible using layers. These are
explained in the book "Photoshop Channel Chops" by a trio of men who
were using the program before it was owned by Adobe. "Real World
Photoshop" has some information on this too.
Most other books pass over the channels quite quickly and with little
information not available in the Photoshop manual. A man who was
working on a chapter on re-touching with channels for "Inside Photoshop
6" was at one point gathering information about this, but he has not
been in touch in about a year and I have no idea if any of the material
he collected will actually be added to the book.
Vadim Kouzmenko
>My point was that even though really good film is capable of recording
>detail out to 100lp/mm, the combination of film, taking lens and
>enlarging lens pretty much reduces the visibility of that detail on the
>final print to nearly nothing. First rate enlarging lenses are by no
>means diffraction limited, even at f/8. A recent Rodenstock ad showing
>the MTF curves of their 105mm f/4 Apo Rodagon indicates a contrast of
>about 60% at 40 cycles when the lens is stopped to f/8. (see PHOTO
>Techniques, July/August 2000 page 5)
60% @ 40 lp/mm is VERY GOOD resolution indeed.
>By the way, the D1 and the D30 digital cameras have a pixel size of
>approximately 12 microns, which translates into about 40 cycles per
>millimeter (corresponding roughly to 25cycles per millimeter on 35mm
>format), not 20.
Really?
A) How many sensors does it take to make 1 white pixel in D1?
The answer is 4: 2 sensors sensitive to Green, 1 Red sensor and 1 Blue
sensor. To your knowledge, the highest uninterpolated resolution of D1
is 1000 x 656, not 2000 x 1312.
B) How many pixels does it take to resolve 1 line pair? The answer is
about 4 pixels per line pair.
So its 11 micron * 2 (see section A) * 4 (see section B) = 88 micron -
is the maximum width of projected line pair that D1 sensor resolves
well.
What we get is: 1 mm / 88 micron = 11 lp/mm
At 20 lp/mm you'll get 0 response.
> I can detect a very slight amount of color aliasing
>on my D1 for certain types of image detail, which indicates that the
>MTF at 40 cycles due to the sensor+antialiasing filter is not quite
>zero.
That's not the problem of antialiasing filter. All area array sensors
are prone to color aliasing at spatial frequencies close to maximum
resolution.
The only real cure for that is 3CCD.
>I have photographed the PIMA/ISO resolution test chart with all
>of my Nikkor lenses at all apertures on my D1, and the lines
>corresponding to 20 cycles per millimeter are very clearly imaged in
>all cases.
Can you show that results to us?
>It is true that high quality film has more detail than current 3MP pro
>digicams, BUT, that detail does not really reach out and grab you
>because it is nearly invisible. You really have to look closely to see
>it. The detail that is shown in good digital images is clearer,
>particularly in shadow areas, and it is not detracted by grain. Thats
>why digital images are able to look so good in comparison to film IMHO.
>
>By the way, what is RDPIII?
RDPIII is Provia 100F.
>Maybe we have really different eyes, but to me, prints from film are
>definitely not MUCH better than prints from a good 3MP digicam. To me
>they look quite similar except for the lack of grain in the digital
>print.
Well, grain was never an issue to me. 8x10 from 35mm film looks very
good (better than prints from digital sources). And if finer grain is
needed, you can always choose appropriate emulsion and appropriate
camera for the job.
>And since you are so upset by digital processing, have you ever
>considered the chemical complexity of what is happening in film and
>enlarging paper? All that chemistry can cause physical harm to
>darkroom workers and pollute the environment.
:)
I'm not specialist in this area, but guess that CCD production, inks
manufacturing, utilization of camera batteries harm environment as
well.
>You seem to have a
>visceral dislike of digital imaging, but aren't the end results really
>what count? As for myself, I like optics and photographs.
No, I like digital very much. I think scanning backs for 4x5 are
great. What I dislike is overrating of digital cameras.
C.L.Zeni
>
> The following link will provide very interesting reading to anyone
> contemplating a jump from film to digital. WARNING: do not believe
> all the inflated numbers you hear about the image quality you can get
> from film. Look at images and ignore numbers. http://www.luminous-
> landscape.com/d30_vs_film.htm
If you're such a digital fan why are you even pharting around here on a
film newsgroup? You're wasting your breath.
I'll not drag out the interminable arguements of one vs the other here -
been done way too many times. But I will say that digital has its
place.
But my kit is NOT one of those places.
--
Craig Zeni - REPLY TO -->> clzeni at mindspring dot com
http://www.mindspring.com/~clzeni/index.html
http://www.trainweb.org/zeniphotos/zenihome.html
As God is my witness . . . I thought turkeys could fly. - Carlson
C.L.Zeni
99%??? I don't agree. Kodak, Fuji et all have way too much money
invested in their emulsion based manufacturing etc to allow digital to
displace it to that degree anytime soon.
Annika1980
>Almost everybody knows that digital will eventually
>> replace film for 99% of all photography. All his test shows is that that
>day
>> may be closer than we think.
>
>99%??? I don't agree. Kodak, Fuji et all have way too much money
>invested in their emulsion based manufacturing etc to allow digital to
>displace it to that degree anytime soon.
>--
They won't have a choice. You notice who's making many of the digital cameras
now, dontcha? Kodak & Fuji. They know where the market is going.
bc1...@my-deja.com
mentioned about film. Implies camera bodies and lenses to me. Guess
what - digital SLRs like the D1 and D30 use 35mm lenses, and the
digital bodies have many features that overlap those of film bodies.
This seems like a very good forum to discuss digital SLRs. Some people
reading the posts in this group may actually have an open mind
regarding film vs digital, and they might appreciate information to
avoid spending their money in the wrong way. For example, I think it
is a very good idea to invest in Canon or Nikon 35mm optics because
excellent digital bodies are now available, and even better ones are
likely to appear within a year or so. On the other hand, some people
might think twice about investing in an expensive film body if they
were aware that digital cameras can provide equal or even better
results. I have absolutely no regrets about buying 35mm film equipment
over the past couple of decades because I can use 90% of what I bought
(the lenses) even if film becomes difficult to obtain in the future.
Brian
In article <39FADC...@mindspring.com>,
Terence Danks
<Spiney...@hedge.hog> wrote:
>
>If you're such a digital fan why are you even pharting around here on a
>film newsgroup? You're wasting your breath.
>
>I'll not drag out the interminable arguements of one vs the other here -
>been done way too many times. But I will say that digital has its
>place.
>
>But my kit is NOT one of those places.
The clear implication that the subject thread is inappropriate to the
group is absolutely wrong! Of course it is appropriate!
"Digital fan"??! Most of us are photographers. If digital can do it
better, at less cost, I am all for it. I have E6 processing costs of
about $500 CDN this month. I would love to go digital if it is ready.
No, I am not sure if it is ready . . . yet. Of course, I have not had
the opportunity of using a D30 or a D1 yet either. Have you? (Thought
not.)
Eventually I am sure it will be ready. These attacks on the
methodology of the tests have gone too far in my opinion. Digital is
closing in fast. Soon, likely not with the D30 or the D1, but perhaps
with the next models (6 megapixels?), I will be moving to digital.
There is nothing in the in this NG that suggests digital 35mm
photography does not have a place here.
Terence A. Danks
Nova Scotia, Canada
Wildlife and Nature Photography
http://www3.ns.sympatico.ca/danksta/home.htm
Tony Spadaro
Terence Danks <dan...@ns.sympatico.ca> wrote:
>
> Eventually I am sure it will be ready. These attacks on the
> methodology of the tests have gone too far in my opinion.
In what way? The test is flawed, and this should be pointed out, as
it has been.
Digital is
> closing in fast. Soon, likely not with the D30 or the D1, but perhaps
> with the next models (6 megapixels?), I will be moving to digital.
That's where I figure I will go digital.
>
> There is nothing in the in this NG that suggests digital 35mm
> photography does not have a place here.
Agreed 100%. Digital IS photography. The actual medium the image is
recorded upon is not important. This forum is titled 35mm, but will no
doubt evolve over the next few years. The number of posts on digital
SLRs in production, announced, and rumoured is proof of that.
Still it is doing a dis-service to publish falsified tests. I have
seen similar "proof" that 35mm produces a picture better than medium
format because the lenses are better. And there was Kodak out to
convince the punters that APS, or 110, or the infamous disc produces
pictures as good as 35mm.
Digital photo restoration in autumnal Chapel Hill NC
http://www.homeusers.prestel.co.uk/magor/tony
New - A Digital Workflow + Sharpness: Threat or Menace?
And Selecting Your First SLR - Java must be enabled
Annika1980
>>These attacks on the
>> methodology of the tests have gone too far in my opinion.
>
> In what way? The test is flawed, and this should be pointed out, as
>it has been.
I don't think the case against the test methodology has been sufficiently made.
If the purpose of the test was to match film vs. digital then many of your
arguments would be valid. However, that wasn't the purpose. The purpose, as I
understood it was to compare the printed output that you could get from the D30
to the printed output that you might expect from film. I mean, the author
chose the EOS-1V for his comparison. You gotta give him props for that!
Tony Spadaro
annik...@aol.com (Annika1980) wrote:
> >From: Tony Spadaro tony_to...@my-deja.com
>
> >>These attacks on the
> >> methodology of the tests have gone too far in my opinion.
> >
> > In what way? The test is flawed, and this should be pointed out, as
> >it has been.
>
> I don't think the case against the test methodology has been
sufficiently made.
> If the purpose of the test was to match film vs. digital then many
of your
> arguments would be valid. However, that wasn't the purpose.
That was exactly the stated purpose.
The purpose, as I
> understood it was to compare the printed output that you could get
from the D30
> to the printed output that you might expect from film.
I saw no printed output from either, and doubt anyone else did.
--
Andrew Dacey
Tony Spadaro wrote:
> I saw no printed output from either, and doubt anyone else did.
I his description of his testing, I do believe that he states that he
compared printed results. Obviously, he can't display those online, so
he uses the scans instead. Given that he mentions what loupe he used, I
doubt that he used a loupe for examining screen display.
Andrew Dacey
Tony Spadaro wrote:
>
> In article <4u8mvsoepkoveddst...@4ax.com>,
> Terence Danks <dan...@ns.sympatico.ca> wrote:
>
> >
> > Eventually I am sure it will be ready. These attacks on the
> > methodology of the tests have gone too far in my opinion.
>
> In what way? The test is flawed, and this should be pointed out, as
> it has been.
He even states as much in the revised page. For anyone who read the page
before the editorial was added, I'd say re-read the page before making
further comments. In the editorial, he completely retracts his comments
that digital is better than film. He revises his conclussions to simply
be that digital provides a better result for the way that he works.
I emailed him personally and got a very nice reply from him. He told me
flat out that he didn't compare a Cibachrome/Ilfochrome because he
doesn't work with them. His concern is with a digital workflow, so
that's what he tested. He admited that the test was not really a
scientific test but really a test of which produces better results based
on the way he works. This was basically a summary of what he discusses
in the editorial.
In the end, all we're left with is an author saying that the D30 is the
tool for him. For his workflow, it provides better results than film.
This could mean anything from the scanner he uses sucks (I doubt it, the
Flextight is a sweet piece of gear) to he doesn't have much skill in
photoshop (quite possible considering he only applied auto levels for
adjusting the colour balance). Or, it could just be that he doesn't want
to have to spend a lot of time working an image in photoshop. Whatever
the reason, he's decided that the D30 gives the results he wants. That's
a big difference from his initial "the D30 beats film" conclussion.
Tony Spadaro
annik...@aol.com (Annika1980) wrote:
>
> So then what?
> You and the author are in agreement then .... at small sizes digital
wins, at
> large sizes film wins.
>
>
Not exactly. The luminous guy says digital wins >>period. I say
digital has it's value, especially when speed is necessary, but the
luminous guy's test is garbage.
Tony Spadaro
As you have pointed out the ammendment to the test is a good deal
more realistic than the original "conclusions".
A year or two ago some Epson ads in magazines like Publish contained
an actual inkjet print (of a leopard as I remember). I've forgotten
which printer they were advertising at that point. I was able to
conclude that inkjet was not as good as photo enlargement. I also
concluded that it was time to buy an inkjet printer.
A portrait photographer who sometimes post to the scanner and printer
forums has been using digital for over a year. Last time I was in touch
with him he was using a 2.2 megapixel cameral.
The digital era has arrived. Digital will grow and chemical will
shrink - both will improve.
Anders Svensson
> Whatever
> the reason, he's decided that the D30 gives the results he wants. That's
> a big difference from his initial "the D30 beats film" conclussion.
Obviously, and to his credit, he retracted the statements that made Tony
(and others, like me) think that he was comparing the general quality of
film (methods) vs digital (methods). Perhaps that happened because he
realized that they were too broad ?
Given that so many of us (I have just started myself) use hybrid methods for
digital imaging (like film + scanning), I don't think anyone is served by
over enthuastic statements about what gives best quality.
>From what I have seen and understood, film in the camera, a scan and digital
processing the rest of the way to a printed image also give excellent
results as long as *prints* are compared.
Systemically, a 300 dollar body and a 500 dollar scanner will be as good,
more versatile and lower cost than any of the top SLR digital cameras. But
there are weaknesses with each concept. The film-slide system has a longer
time to image (developing comes in between, and scanning must take place)
and the digital SLR needs (expensive) supporting equipment as soon as the
number of images in storage (unsent, unstored on other media) is maxed out
(currently, that's low).
There is no *way* a 35 mm slide produced by a digital camera can compete
with a slide from a film using system, yet. This proves very little but to
counter a "digital is always better" argument. Unless, like me, slides is a
preferred way of looking at large images.
If we look at such parameters as absence of grain, color saturation, color
hues, edge definition and other "giveaway" indicators of quality, it is
likely that a highly post processed image will "look better". It can remove
pixellation synthetically (fractal programs), it can adjust colors and hues
to favorably match the print process, and it can synthetically adjust for
lack of such important visual "quality clues" as edge sharpness.
When such a super processed image (some of this is done in-camera) is
compared to a non processed digital scan of a slide, I would *expect* the
digital to look better on a computer screen or on a print from a digital
printer.
I would not like to compare that image to a properly made, high quality
print, made by ordinary photographic methods (that also can "manipulate"),
yet.
Digital image processing is in it's infancy, current photo technology is
probably at or near it's peak :-) Hand colored B&W photos can also look
great, BTW :-)
> --
> Andrew "Frugal" Dacey,
> fru...@tildefrugal.net
> http://www.tildefrugal.net/
Terence Danks
touch upon a concern of mine. I have no experience with digital
cameras period!
However I do have considerable experience with computers and have a
definite love/hate relationship with them. My tolerance for buggy
software and hardware installation difficulties (installing the
Polaroid SS4000 on my system a month or two back was a very trying
experience) is at a very low ebb these days.
I have "electronicised" cameras (F90X and EOS3). They have been
trouble free and easy to use once the instruction manual has been
mastered. I sense the new digital cameras as being more a "computer
peripheral" than a camera in the traditional sense. How susceptible
are these cameras to computer-like glitches?
So few of us have actually tried machines like Nikon's D1 and Canon's
D30 that there is no body of experience for judging the dependability
of these expensive tools when used in the field.
I am already too tied to this quirky and persnickety computer on which
I am "keyboarding" this post. It drives me to distraction at times
(since installing the SS4000, I am unable to get my flatbed to work at
all . . .typical!)
A digital camera, which seems all but inevitable in the quickly
approaching future, will bind me even more totally to these infernal
boxes that seem so plagued with incompatibilities and whose
specifications and capabilities.change on an almost daily basis.
While I look forward to digital capabilities, I dread the downside
that will likely be the price that will have to be paid.
Don Marcotte
one of my PCs. The Agfa 1280 uses a DB9 serial port. The software is a
snap to install and use. As one who owns 3 scanners (2 SCSI, one
parallel port) and two parallel port printers, I can sympathize with
your problems. Installing scanners and multiple parport printers is an
adventure. It took me days to get everything sorted out and I have an
extensive PC background (but growing intolerance!). My latest challenge
is installing proxy software so my second PC can access a cable
modem....... I'm at 2 days and counting.
Don
Annika1980
>There is no *way* a 35 mm slide produced by a digital camera can compete
>with a slide from a film using system, yet. This proves very little but to
>counter a "digital is always better" argument. Unless, like me, slides is a
>preferred way of looking at large images.
That would depend on which digital camera you are using (as well as which film
system). It may be true for the D30 but not for the Foveon camera or a
Hasselblad with a digital back.
James Switzer
Newspaper, magazine, and freelance photogs, around the world, are changing
over to digital cameras. The Kodak modified Nikon and Canon bodies as well
as the D1 make up the bulk of the deployed cameras.
There is enough reliability in these systems for these operations to depend
upon them and photogs to earn a living.
Kodak has a feature you may be able to make benefit of: wireless
transmission. Imagine you are in your hide but are running out of disk
space: it has been a productive day. Just connect your camera to your cell
phone and transmit your images back to HQ.
At the National Association Of Photographers yearly meeting in Buffalo
earlier this year, I watched Kodak demo the system. It failed the first
time out but worked like a charm from then on.
FWIW.
Cheers
Jim
"Terence Danks" <dan...@ns.sympatico.ca> wrote in message
news:kfcovscl5lrvtjn3v...@4ax.com...
Annika1980
>> If the purpose of the test was to match film vs. digital then many
>of your
>> arguments would be valid. However, that wasn't the purpose.
>
> That was exactly the stated purpose.
Did we read the same article?
If the purpose was to match film vs. digital then the author wouldn't have
chose the D30. He'd have gone with something like the $50K Foveon studio
camera or maybe a large format with a digital back. Now I'd like to see some
tests done with those babies!
Anders Svensson
Of course, 35 mm slides are very tricky to produce digitally. That is because the
machinery to make them is so intricate and expensive that they often are made by
photographing larger originals with a 35 mm camera :-)
Vadim Kouzmenko
And from my personal experience - Provia matches colors pretty well.
What you see in the report is a scanned image. If scan is not
accurate, then THERE IS SOMETHING WRONG WITH SCANNING ITSELF.
I don't want to speculate about what exactly went wrong, but there are
several concerns and unanswered questions:
1) How scanner was calibrated? Such test is nonsense without proper
hardware calibration with respectable tools.
2) In what color space image was captured? Was there any
profile-to-profile conversion? Poor color spaces, like sRGB,
deteriorate colors.
3) What color management settings were used in Photoshop? Improperly
configured Photoshop could be a problem.
4) Auto Levels was done on scanned image. I doubt that applying Auto
Levels is the best technique. And keep in mind that levels adjustments
shift both contrast and colors.
There is no way you can judge Provia or even D30 capabilities from
that report. There is nothing you can judge except author's attitude
towards comparing digital vs. film.
P.S.
Blue colors on D30 image looks unnaturally over saturated (my monitor
is hardware calibrated on a regular basis).
On Sun, 29 Oct 2000 07:08:02 GMT, Marcio Watanabe <mar...@nospam.com>
wrote:
>I'm amazed nobody made comments on the color differences between the
>two shots of the buildings. The author claims that the colors of the
>D30 shot is closer to reality than the Provia scan. Since I'm looking
>at the pictures on a computer monitor, I have no way of telling which
>one is more accurate, but I can say this: if there is such a thing as
>a color accurate image, there is no way that there can be this much
>difference between it and a Provia shot. Look at the blue colors on
>the warehouse and the EDS sign (if this EDS is Ross Perot's EDS, the
>blue does not look anything like reality in my monitor, but then it
>could be my monitor, so I only want to comment on color differences).
>How can Provia be this far off from reality?
bc1...@my-deja.com
4x5 film, then the following link will provide some interesting reading
for you:
http://www.phaseone.com/whyp1/quality/quality_index.html
See ya,
Brian
In article <20001029115159...@ng-fi1.aol.com>,
Tony Spadaro
Terence Danks <dan...@ns.sympatico.ca> wrote:
>
> I have "electronicised" cameras (F90X and EOS3). They have been
> trouble free and easy to use once the instruction manual has been
> mastered. I sense the new digital cameras as being more a "computer
> peripheral" than a camera in the traditional sense. How susceptible
> are these cameras to computer-like glitches?
One thing I suspect is there will be early glitches in models - Like
the Canon exposure problem of a year or so ago, or some problems that
have showed up in F100s, etc. But unlike a computer, which it a
conglomerate of parts assembled to a purpose, and then added to for new
purposes, a camera is a pretty discreet item.
You will not be adding software or hardware that will alter the way a
camera runs - I might be being optimistic. My computer occasionally
throws a snit and "loses" either the SCSI chain, or my graphic tablet,
but neither of those items (and the SCSI chain is actually three pieces
of hardware running off one card) was part of the computer I bought.
Camera add-ons simply don't change the basics of the system. They are
just add-ons. The problems of some 3rd party lenses and new camera
models is rather troubling though. I doubt I'll ever buy anything but a
Canon lens in future.
--
Jerry Coffin
mar...@nospam.com says...
> I'm amazed nobody made comments on the color differences between the
> two shots of the buildings. The author claims that the colors of the
> D30 shot is closer to reality than the Provia scan. Since I'm looking
> at the pictures on a computer monitor, I have no way of telling which
> one is more accurate, but I can say this: if there is such a thing as
> a color accurate image, there is no way that there can be this much
> difference between it and a Provia shot.
Quite the contrary. There's no way there's this much difference
between reality and _properly exposed_ Provia, but incorrect exposure
can screw things up terribly.
Worse, if he hasn't calibrated the color on his equipment quite
carefully, we're probably just seeing the effect of how lousy the
color profiles are when you get them from most manufacturers. Color
perception depends on too many factors for a manufacturer to ship a
profile that means much of anything WRT absolute color accuracy.
OTOH, using careful calibration, things change considerably. I've
taken a shot (yes, on Provia) scanned it in, printed out the result,
and done a colorimeter comparison to the original subject. With care
I can match the two FAR more accurately than I can perceive with my
eyes.
> Look at the blue colors on
> the warehouse and the EDS sign (if this EDS is Ross Perot's EDS, the
> blue does not look anything like reality in my monitor, but then it
> could be my monitor, so I only want to comment on color differences).
> How can Provia be this far off from reality?
It undoubtedly isn't. Unfortunately, you have no idea how he's
calibrated his monitor, and he has no idea how you've calibrated
your's.
There may be some room for argument about whether grain looks better
or worse than pixelization, but there's NO room for argument about
the subject of color accuracy -- even starting with a film that's
provably NOT very accurate (e.g. Velvia) it's still possible to
calibrate the rest of the digital workflow to compensate. I know
because I've done it -- I can take a shot with Velvia, scan it in,
print it out, and produce results that match the original subject
with a matter of a few percent (by measurement, NOT the rather
undependable perceptions of my eyes).
The main problem is that this WON'T EVER work simply by doing
something like installing color profiles from a manufacturer. To get
good accuracy, you've GOT to calibrate every piece of your equipment
yourself. Profiles from manufacturers (especially on monitors)
simply can't take enough into account to mean much. Monitors
deteriorate with age and your color perception is influenced by
ambient light, and they can't even TRY to take these into account.
Being fair I have to give some companies credit: just for example,
the color profiles ALPS ships for their printers are _amazingly_
accurate, at least in my experience -- though my tweaks produced
output that was measurably more accurate, I'm not at all sure I could
depend on telling the difference visually.
--
Later,
Jerry.
The Universe is a figment of its own imagination.
bc1...@my-deja.com
It takes one row of pixels to image a line and two rows of pixels to
image a line pair. The resolution of the D1 really is 2000 x 1312
pixels. This is shown in actual test photographs as I mentioned. You
won't take my word for it even though I own a D1, have been a
photographer for over 20 years, have a Ph.D. in optics, am a
professional lens designer, and have carefully tested my D1 with over
15 different lenses at all apertures. However, there is a very well
known website that you should become familiar with, and which backs me
up. Check out this page to actually see for yourself what the
resolution of the D1, the Canon D30, the Fuji S1 and the Nikon 990:
http://www.dpreview.com/reviews/canond30/page18.asp
The numbers on the chart refer to lines per picture height divided by
100. To get line pairs, just divide by two.
Brian
In article <39fa3581...@enews.newsguy.com>,
PBurian
small from Provia 100F would be even sharper and more colorful. And in much
larger prints, the film would win hands-down.
I don't think any manufacturer is saying their digital cameras can match film
in quality yet. Yes, the day will come, but not yet.
Still, the 8x10" prints were very nice, esp. when Unsharp Mask was used in
Photoshop before printing.
Peter Burian
Ryan Shaner
The true test would be to view the Provia transparency with a good
loupe and compare it with a high-res on screen image of the D30 image.
I've never seen ANY transparency look that desaturated: the problem
has to be with the scanning. Something is just not right with the
image comparisons.
--
Ryan Shaner
E-mail: rxsh...@home.net
*Remove NOSPAM to reply via e-mail
bc1...@my-deja.com
comparison, backed with some decent blowups to illustrate your points
would validate your opinion. Do a test prints made from both digital
and film sources, do the best job possible on each print - that is,
don't intentionally choose subject matter or test conditions that might
bias one in favor of the other - and then present your case.
Brian
In article <20001030111955...@ng-cr1.aol.com>,
Vadim Kouzmenko
>image a line pair.
Yes, but only in the best scenario, when lines and sensors grid are
aligned precisely. The worst case would be when one line lights half
of one sensor and half of another.
So averaging must be done. I said "about" 4 pixels per line pair. The
exact figure must be between 2 and 4.
>he resolution of the D1 really is 2000 x 1312 pixels.
Well, Brian, that's not exeactly uninterpolated resolution. There are
only 656,000 red sensors, 656,000 blue sensors and 1,312,000 green
sensors per CCD chip. 1000 x 656 is it's true resolution. Other than
that is resampling.
>This is shown in actual test photographs as I mentioned. You
>won't take my word for it even though I own a D1, have been a
>photographer for over 20 years, have a Ph.D. in optics, am a
>professional lens designer, and have carefully tested my D1 with over
>15 different lenses at all apertures. However, there is a very well
>known website that you should become familiar with, and which backs me
>up. Check out this page to actually see for yourself what the
>resolution of the D1, the Canon D30, the Fuji S1 and the Nikon 990:
>
>http://www.dpreview.com/reviews/canond30/page18.asp
>
>The numbers on the chart refer to lines per picture height divided by
>100. To get line pairs, just divide by two.
Interesting. Thanks. So what we've got here?
D1 MTF response is 0 at about 1300 LPH, or 650 cycles per picture
height, or 40 lp/mm. Using Photoshop, I tried to estimate D1 MTF:
------------------
LPH lp/mm response
------------------
2 6 100%
4 13 95%
6 19 75%
8 26 50%
10 32 35%
12 38 20%
------------------
According to datasheet, Provia MFT response is about 33% at 50 lp/mm.
On the other hand, I've estimated 35% @ 32 lp/mm for D1, which is
equivalent of 21 lp/mm on 35mm frame. Thus, 35mm + Provia IIIF is
capable of producing 2.5 times sharper results than D1/D30 does.
s...@randomc.com
>But you didn't actually make a direct visual comparison. Such a
>comparison, backed with some decent blowups to illustrate your points
>would validate your opinion. Do a test prints made from both digital
>and film sources, do the best job possible on each print - that is,
>don't intentionally choose subject matter or test conditions that might
>bias one in favor of the other - and then present your case.
No, that's the wrong approach. You want to use processes and subject matter
that does favour the strength of each medium to get its best. Its important
to know what favours each medium so you know how to get the best out of it.
Then we will know where each has its place.
Steve
>Brian
Keith Curtis
image projected on a digital projector side by side. Keep going back and see
how they compare a sthey get bigger,
Just a thought.
Anyway I can,t wait to get mine, I never was comparing it with film as such,
images are images like paintings. I don,t tend to treat them as an absolute
technical exercise but what is in the picture. The bronica is so sharp that
half the time you have to soften it. Just wanted a tool that I could use my
Canon lenses and not be the high price of the Kodak family of press cameras.
This seems to have it, loads of my work seems to end up for peoples web
sites and press releases. Should be fine and save me loads of time and money
running back a forth to labs and worrying if they are going to knacker yet
another roll of film.
Keith Curtis
www.photo-art.co.uk
bc1...@my-deja.com
digital camera. Don't forget, when you expose the film you have to use
a lens which will in the best of circumstances provide 50% contrast at
50lp/mm and then if you make a chemical print the enlarging lens will
reduce the contrast by another 50% at 50lp/mm. So the final on-print
contrast from 50lp/mm detail is 0.33 (from the film) x 0.5 (from the
taking lens) x 0.5 (from the enlarging lens = 0.0825. This level of
contrast will just barely be visible to the eye. The important detail
that actually contributes to perceived image quality is at much lower
spatial frequencies: 10 to 30 lp/mm. If you were looking at
resolution images from a D1 or D30 on Photoshop and using those images
to estimate MTF (a clever idea by the way), then you will be able to
get a similar result on a print assuming you print large enough to
overcome limitations of the printer. This is equivalent to ignoring
the MTF of the enlarging paper of a chemical print - if the print is
large enough then it can be neglected as unimportant.
I think that it would be interesting to photograph the ISO chart with a
variety of film emulsions and compare the final printed charts with
similar printed charts from a digital camera. Such a comparision could
be demonstrated accurately on a computer screen, also. I would expect
fine-grained film to have an edge in resolution, but that digital would
appear cleaner. I DO still have a regrigerator full of film, so . . .
let me know if you have any suggestions for such a test.
As I write this I am printing a series of 12x18s from some of my D1
files, and I am amazed by how well certain images respond to
sharpening. "Organic" images like portraits as well as abstract hard-
edged images seem to do best with digital. Some prints look definitely
have a large-format look to them, with stunning color and clarity.
Images with lots of small lettering do the worst because it is possible
to detect aliasing in large prints.
Brian
In article <39fda211...@enews.newsguy.com>,
Meryl Arbing
that there must be something wrong with the process that led him to those
conclusions. I think that the author was trying to fix things and so
introduced errors into the experiment.
He thought that by using the same Canon lens that he would remove that
variable but he did not. Since the digital camera has a "lens mutiplier"
factor of 1.6 when he chose a Canon 100-400 zoom for the film camera, it
became a 160-640 zoom on the digital. To compensate he zoomed the Film
camera to 160mm and left the digital at the lowest setting. Even though it
was the same lens, it is safe to assume that lenses do NOT improve in
performance as they zoom.
We know that Provia F 100 is definitiely ISO100 but with digitals (even when
you set the camera to one ISO) vary the actual ISO rating as part of their
auto exposure process. So the digital can only be assumed to be ISO 100. I
doubt that it would reach ISO200 but it could be higher (or lower) than
ISO100 which effects the appearance of CCD "noise" (the digital equivalent
of film grain). Some digital cameras at high ISO levels may as well be
shooting in a blizzard!
He uses an excellent scanner but no where does he indicate the resolution of
the scan. The scanner is capable of 3200dpi and I know my own, smaller,
2400dpi film/slide scanner produces an image of 3120x2080 pixels compared to
the maximum 2160x1440 that the D30 can produce. The scan he would have
produced should have been at least double the digital "resolution" of the
digital camera. He states that he used the "default settings" of the scanner
but never states what thaey are. Perhaps they were not optimal.
He says that he turned off any in camera sharpening and applied
autosharpening from a piece of software on both images. This also is fraught
with problems since the software could have (and most likely did) choose
different settings for each image depending upon the dpi resolution. He
would have been better served if he use the sharpening tool in Photoshop and
made sure it was set to the same degree for each. He decided NOT to do that
because he was afraid he would slant the results. I think his proposed
solution was more problematic than just setting it himself. Now we will
never know how much or how litle sharpening was applied and we certainly
can't duplicate the effort.
The same problem comes when he applies another Automatic change.. the
Autolevels in Photoshop. Again, both images are treated differently
according to the algorithm in the software. It does not mean that they are
given equal treatment.
When he came to make prints of the images he made 8x10 and 13x19 and
examined them on the screen. This is quite a subjective process...as the
conclusions indicate but, understanding this, he gets one of his friends to
evaluate the shots as well. We have no idea how involved this friend was in
the process. It would have been better to have a completely uninterested and
unaware 3rd party who did not know which print was digital...of course 5
seconds with a loupe would settle the matter for most people.
I have several digital cameras as well as film cameras. I appreciate the
nice features of digital and it is not important to me whether they exceed
what my film cameras will do.
"PBurian" <pbu...@aol.com> wrote in message
news:20001030111955...@ng-cr1.aol.com...
Billy R
color information is interpolated. Color interpolation can only avoided by
using multiple image sensors, e.g., one for each primary color.
"Vadim Kouzmenko" <vkouz...@onebox.com> wrote in message
news:39fda211...@enews.newsguy.com...
Tony Polson
> "Vadim Kouzmenko" <vkouz...@onebox.com> wrote in message
> news:39fda211...@enews.newsguy.com...
> > Well, Brian, that's not exeactly uninterpolated resolution. There are
> > only 656,000 red sensors, 656,000 blue sensors and 1,312,000 green
> > sensors per CCD chip. 1000 x 656 is it's true resolution. Other than
> > that is resampling.
>
> Not true! The D1 delivers a full 2000 x 1312 resolution image. Only the
> color information is interpolated. Color interpolation can only avoided by
> using multiple image sensors, e.g., one for each primary color.
On the contrary. What Vadim says is true. The D1 has 2.6 million
single colour pixels. It outputs a 2.6 million pixel file, with each
pixel multi-coloured. This can only be done in post-processing, because
there is not enough data collected by the CCD when the picture is taken.
I think it's a matter of concern when a 2.6 million pixel output file
has two of the three primary colours defined by only one quarter of
those 2.6 million pixels. I suppose we should be thankful that Nikon
don't claim an output quality of a greater number of pixels than the CCD
actually has, like another manufacturer we could mention whose digital
SLR uses the Nikon mount.
--
Tony Polson, North Yorkshire, UK
PBurian
>invested in their emulsion based manufacturing etc to allow digital to
>displace it to that degree anytime soon.>>>
Both will continue side by side for decades. Newspaper photogs have mostly
switched to digital already. When a 10 megapixel digital SLR is made, a lot of
magazine etc. photogs will switch too.
When will photo enthusiasts switch? Some, never. Future shock. I'll buy one
when they make a true 10MP model for under $3500. But will keep my current
camera too.
Peter Burian
PBurian
>They won't have a choice. You notice who's making many of the digital
>cameras
>now, dontcha? Kodak & Fuji. They know where the market is going.>>>
Oh, I'm sure they would be happier if digital were never invented, but the
market for film was not growing in leaps and bounds. They came up with APS
format to help that, but it did not do much.
I think they are hoping that people with digital cameras will still want prints
from a lab. Paper is a major source of income. (They missed the boat by letting
Epson etc. take over the market for affordable photo-realistic printers.)
All new mini-lab systems include slots for SmartMedia, Compact Flash, etc.,
cards so photo finishers will be set up to make prints from memory cards. A lot
of people will still want that; they take the card in to the mall and for $7.
they get prints the next day. Like always.
A lot of comfort level with that process. No hooking up the camera to a
computer, playing with software, a mess of cables to a photo-realistic printer,
etc. etc.
Don't undersetimate the above esp. with people over 40. They want comfort food
and comfort pictures.
Peter Burian
bc1...@my-deja.com
to do any visual comparison at all. Perhaps a series of different
subjects ranging from resolution test charts to portraits to
landscapes, etc., but in each case identical for both digital and
film. Agreed that you should use processing to bring out the best in
each medium - thats what I meant, although it might be nice to have a
series of test images that are sharpened vs. unsharpened, etc.
Brian
In article <8tkioe$7cn$1...@bcarh8ab.ca.nortel.com>,
bc1...@my-deja.com
I suspect (hope?) that you will by buying your $3500 10MP camera sooner
than you think.
Brian
In article <20001031081012...@ng-mg1.aol.com>,
bc1...@my-deja.com
big prism to combine them, and a new set of lenses to deal with the
prism. Color interpolation works quite well and is the intelligent way
to go. The artifacts are measureable and such measurements are widely
available on the net for most digital cameras currently on the market.
You don't lose as much as you fear. On most of my recent digital
images I am hard-pressed to find artifacts due to color interpolation.
Brian
In article <le7tvss7fum451cm5...@4ax.com>,
Tony Polson
> Tony Polson wrote:
> > On the contrary. What Vadim says is true. The D1 has 2.6 million
> > single colour pixels. It outputs a 2.6 million pixel file, with each
> > pixel multi-coloured. This can only be done in post-processing,
> because
> > there is not enough data collected by the CCD when the picture is
> taken.
> >
> > I think it's a matter of concern when a 2.6 million pixel output file
> > has two of the three primary colours defined by only one quarter of
> > those 2.6 million pixels. I suppose we should be thankful that Nikon
> > don't claim an output quality of a greater number of pixels than the
> CCD
> > actually has, like another manufacturer we could mention whose digital
> > SLR uses the Nikon mount.
>
>
> I'd be more concerned if I had to pay for 3 large CCDs instead of 1, a
> big prism to combine them, and a new set of lenses to deal with the
> prism. Color interpolation works quite well and is the intelligent way
> to go.
It might be "intelligent" but it makes assumptions that may or may not
agree with reality. It's quite alarming that even a Nikon D1 makes up
2.6 million red and blue "decisions" on the basis of only 650,000 red
and 650,000 blue pixels.
> The artifacts are measureable and such measurements are widely
> available on the net for most digital cameras currently on the market.
> You don't lose as much as you fear.
You mean you don't lose data that you haven't gathered in the first
place. I don't disagree with that. The real question is whether 2.6
million blue (or red) values can realistically be inferred from 650,000
pixels of information. Extrapolating by a factor of 4 takes some doing.
> On most of my recent digital
> images I am hard-pressed to find artifacts due to color interpolation.
The software may be clever enough to fool you into thinking that the
output is a true representation of the scene originally photographed.
No amount of looking at a digital image will find where the colour
interpolation went wrong.
The better the software, the more credible the result will be. But just
because its credible doesn't mean it's right. The software may be no
more than a credible liar.
One thing that digital enthusiasts seem to share is an unquestioning
desire to believe in the technology, regardless of what is really on
offer. They believe, whilst carefully ensuring that they never consider
the facts, and what those facts really mean for imaging.
What the facts mean is that digital has a long, long way to go before it
can gather information of the quality that can be recorded on 35mm film.
Until then we will have to be content with output that is largely
synthesised on the basis of inadequate image data.
Karen Nakamura
Hmm...
If you had a white sky with a single telephone wire going across the
horizon, the wire being a hairline just 1 pixel wide.
It would hit the:
RGBGRGBGRGBGRGBGRGBGRGBGRGBG
line of pixels. You'd have to have a pretty smart algorithm to make sure
that line wasn't color aliased in a strange way. It becomes harder if
the line itself has color, or the background is colored, etc. Or if you
have two hairlines going parallel to each other, just one pixel apart.
That's why the "polar bear" digicam shots usually have weird color
aliasing in the bear's fur.
Karen
In article <8tlt9s$vt6$1...@slb2.atl.mindspring.net>, "Billy R"
Bjorn Rorslett
Neither Tony nor Vadim are entirely correct in their assumptions. There
are indeed 2000 x 1312 individual pixels on the CCD of a D1. All are
capable of storing image information as data on image luminance, and
hence all of them contribute to the total spatial resolution of the
captured image. Pixel-by-pixel colour variation is not random, on the
contrary, there is a very high degree of spatial autocorrelation
involved so "interpolating" the colour of a pixel from its neighbours
can be done with minimal loss of spatial resolution. The exception
being, of course, along sharp transitions from one colour to another
where some "bleeding" or moire will occur. However, clever
post-processing can largely take care of that as well.
The fact remains that printing magazine front covers or double-page
spreads from D1 files is entirely possible, and with an excellent result
indeed (I have had this done many times this year). For a professional
photographer this is the answer which matters, not that of the doubting
and pixel-counting people. Digital can, and of course will, be better
than today's technology, but even the current crop of digital cameras
earn good money in practical picture-taking and image sales.
Regards
Bjorn Rorslett
Visit http://foto.no/nikon/ for UV & IR Colour Photography and other
Adventures in Nature and Digital Photography
jw&a
> >
> All new mini-lab systems include slots for SmartMedia, Compact Flash,
etc.,
> cards so photo finishers will be set up to make prints from memory cards.
A lot
> of people will still want that; they take the card in to the mall and for
$7.
> they get prints the next day. Like always.
>
> A lot of comfort level with that process. No hooking up the camera to a
> computer, playing with software, a mess of cables to a photo-realistic
printer,
> etc. etc.
>
> Don't undersetimate the above esp. with people over 40. They want comfort
food
> and comfort pictures.
>
that, and you get photos from a really, really, good printer, without
needing to capitalize it, maintain it, or feed it.
bob rogers
Vadim Kouzmenko
>You are comparing film MTF to a final ready-to-print image from a
>digital camera. Don't forget, when you expose the film you have to use
>a lens which will in the best of circumstances provide 50% contrast at
>50lp/mm and then if you make a chemical print the enlarging lens will
>reduce the contrast by another 50% at 50lp/mm. So the final on-print
>contrast from 50lp/mm detail is 0.33 (from the film) x 0.5 (from the
>taking lens) x 0.5 (from the enlarging lens = 0.0825. This level of
>contrast will just barely be visible to the eye. The important detail
>that actually contributes to perceived image quality is at much lower
>spatial frequencies: 10 to 30 lp/mm. If you were looking at
>resolution images from a D1 or D30 on Photoshop and using those images
>to estimate MTF (a clever idea by the way), then you will be able to
>get a similar result on a print assuming you print large enough to
>overcome limitations of the printer. This is equivalent to ignoring
>the MTF of the enlarging paper of a chemical print - if the print is
>large enough then it can be neglected as unimportant.
OK, from what I saw, I'm convinced that the gap between:
a) 35mm camera + camera lens + fine grained color film + either
enlarging lens, or projector lens or loupe
and
b) 3MP hi-end digital camera
might be less than 2.5 times.
Resolution is not all that counts here, and I believe that 35mm slides
are still the best of what you can get. But I do admit that the
difference between digital and 35mm film is smaller that I thought
before.
>I think that it would be interesting to photograph the ISO chart with a
>variety of film emulsions and compare the final printed charts with
>similar printed charts from a digital camera. Such a comparision could
>be demonstrated accurately on a computer screen, also. I would expect
>fine-grained film to have an edge in resolution, but that digital would
>appear cleaner. I DO still have a regrigerator full of film, so . . .
>let me know if you have any suggestions for such a test.
If you do a test, just do it well :)