Startling News!

[Truelight]

I'm probably getting in over my head here in such a "techy" discussion, but I hafta ask this one... If we are truely comparing film v.s. digital technolgy for quality here, what is the point in -scanning- the film?? Show me a print made from a digital camera and printed using digital equipment and then show me the same print made from the film negative and processed traditionally with chemical methods. You don't doubt that the film-based print will still look better?

All bets are off when we scan (digitize) the negative. What you see then is the quality of the scanner, not what can be done when working with traditional film methods.

Just my 2-bits (bytes?)

Rick Ohnsman
[email protected]

The D# pictures don't show out of focus effects or camera movement. They
were scanned by a Nikon LS-2000 if I'm reading the caption right. I've
never encountered the beast but it does seem to resolve the grain. Look
at the blue sky, for instance.

The grain of the 200 is larger and clumpier than the 100 and both show
well.

That I know to be a common misconception. You're not seing the grain.
You're seing the interference pattern between the grain and the "sampling
matrix" of the scanner (LS200). On a Polaroid SprintScan 4000 (4000 dpi)
you would see a different pattern. The ASA100 grain is really resolved at
6000 ... 8000 dpi.

I think there wasn't a lot more in those 35mm camera images to be
brought out.

Well you have looked at the URL I mentioned which shows what scans look
like at really high dpi. To put two and two together, those images are
not really representative of what film can do.

Clark's stuff is very interesting. It sure proves that you can get more
detail from a good scan, but my own experience with PhotoCD's is more
like the D# results than like Clark's. We all can't own a drum scanner
so the practical equivalent is more like the $1000 film scanner.

While of course we all don't own a drum scanner (we wish! , it is not
legitimate to drag the today's scanner limitations to the judgement of
film. In a few years, those same frames could be scanned on Polaroid or
Nikon or Minolta scanners with much, much higher resolution and dynamic
film, which will bring more of the detail that the film frames captured
would keep. That's the film quality headroom advantage - while one may
not need or be able to afford to capture the full information on film, as
scanners get better, better and better images could be retrieved. By
contrast when one takes a digital image, not much can be done. FRactals
and other interpolative algorithms can do some, but, looking at the URL I
mentioned, I doubt they'll do as much as what a higher resolution scan
can retrieve (up to say 10000 dpi).

If that affordable gizmo won't make our film efforts much better than a 3
megapixel digital camera, why bother? No processing flaws, emulsion
dings, scratches, fingerprints, dust or spit on a digital picture (don't
ya hate it when you were just trying to blow off that eyelash and...).

I do, yes I have to buy those compressed air cans which is an expense.

Well one can screw up any image. Many photographers don't hold their
digital or film cameras steady enough to obtain the sharpest picture
possible. Many other types of mistakes are made. However, incompetence
does not affect what the medium is capable of when used properly.
Depending what one wants to do with the images, digital cameras may
eliminate some steps and make others more streamlined and more
convenient. However there are nonetheless tradeoffs and limitations.

Clark makes reference to what I call the 300 ppi myth. Simply stated it
implies that you need 300 pixels per inch on the paper for a picture to
look photographic.

We may talk about 300 ppi on 8x10, etc, but I feel "quality headroom" is
of important significance in an image capturing technology. SOmeone had
mentioned a batch of WWII frames having been discovered; blown up to
great enlargement and examined up close at teh exhibition. So maybe just
making a "satisfyingly sharp looking if examined at proper distance"
8x10" or 11x14" is not ultimately enough. Some images are timeless and
priceless. I have a few I consider very dear to me. I sure am happy I
took them with a 35mm camera with a high quality lens and ASA25 film,
rather than a circa few years back digital camera. I wish I could have
been able to afford a medium or perhaps even large format camera to take
them with. Unfortunately that was out of my range. As the masses switch
to digital cameras will (no longer mass-targeting) 35mm become gradually
out of range of a typical user (pricing aand availability wise as well as
advertisement / education), BEFORE prosumer digital cameras truly match
the quality of 35mm film process? Think about it.

FJM

 

[Peter iNova]

The D# pictures don't show out of focus effects or camera movement. They
were scanned by a Nikon LS-2000 if I'm reading the caption right. I've
never encountered the beast but it does seem to resolve the grain. Look
at the blue sky, for instance.

The grain of the 200 is larger and clumpier than the 100 and both show
well.

That I know to be a common misconception. You're not seing the grain.
You're seing the interference pattern between the grain and the "sampling
matrix" of the scanner (LS200). On a Polaroid SprintScan 4000 (4000 dpi)
you would see a different pattern. The ASA100 grain is really resolved at
6000 ... 8000 dpi

FJM

Common misconception or not, that pattern of disruptive clumpy sky isn't anywhere near as smooth as it is from both of the digital cameras on either scanner. Especially compared to the D1.

The format of illumination in various scanners brings out different sorts of detail, too. Some admit from the start that they're fighting scratches and imperfections. (I think D# dropped a zero from the EPSON GT-9600 scan specs. It looks a lot more like a 4000 dpi scan.)

I've actually placed a number of my slides in a Leica microscope to see what's really going on. Some of what I pointed to shows up there much better than on these D# quick scans. What is seen at 40x and up through the microscope bears out my less than sanguine opinion of transparency film and reinforces my opinion that the films D# is showing don't have a lot more to give.

Some of that clump in the blue sky is structures of irregularity common to most film but the irregularities are made from grain that, while not resolved here, isn't much below this effect. With double the resolving power in a scanner you'd be looking at the grain directly, color fleck for color fleck but the picture wouldn't be yielding much in the way of added useful detail.

The 6000 and 8000 dpi scans of 35mm images that clark puts forward... how much smaller can you shrink them without reducing the actual, practical information to the level seen from the 4 x 5's? At least twice in my Photoshop. Try it with the Epson 4000 dpi scan and the detail lost in the process only begins to retreat from the image below a 60% reduction. And for comparison to a digital image, the inherent noise of the transfer process must be counted. If one were preparing this image for reproduction, one would have to make about twenty points of retouch on the Visa/Master Card sign at the upper right. Extrapolate that to the whole image and it becomes hundreds of spurious emulsion dings, dust flecks, hairs, fibers and "gradue" you couldn't possibly let get by if the shot were being reproduced in Nat Geo.

The D# images are already showing phenomenon I've seen many times in the scope. The slight color fringe in the Kenwood lettering is a pretty familiar artifact, for instance.

In most cases if a digital camera gives you detail, it really was there and isn't an artifact of the carrier (except for antialiasing and moiré effects, sigh.)

One thing troubles me about the D1 shot of the same Kenwood sign area. The fine lattice under the sign (part of an AC unit the sign is covering?) is very hard-edged. Stair stepped. Bad antialiasing. I wonder if the photographer didn't put the in-camera sharpening on "stun"?

The convergence of digital sensing versus film won't be a dead issue in favor of digital for a while, but you have to ask yourself if the current pace of development continues until theoretical limits on circuit refinements are reached, at which point was film rendered irrelevant? A hundred years from now? Five? It WILL happen and my money is favoring the "sooner than you think" camp. For many applications, it's pretty obvious that the point film becomes obsolete is already in the "ago" category.

-iNova

 

[Peter iNova]

I'm probably getting in over my head here in such a "techy" discussion,
but I hafta ask this one... If we are truely comparing film v.s. digital
technolgy for quality here, what is the point in -scanning- the film??
Show me a print made from a digital camera and printed using digital
equipment and then show me the same print made from the film negative and
processed traditionally with chemical methods. You don't doubt that the
film-based print will still look better?

All bets are off when we scan (digitize) the negative. What you see then
is the quality of the scanner, not what can be done when working with
traditional film methods.

Just my 2-bits (bytes?)

Rick Ohnsman
[email protected]

Good question. For the photographic print, the idea you present sets up a direct apples to apples comparison. Now look around. The real world shows you thousands of printed color images every day. Not many are photographic. Nearly zero of them were NOT digitized at some point. Every magazine illustration, news photo, advertising shot, billboard, postage stamp, graphic, letter on a page... all have fallen into the computer.

Film is a phenomenon that had to exist before digital imaging became practical. It started with photographic processes discovered in the 1830's -- it's that old. In the last about twenty years the entire process has flipped from camera to page, most of it in the last ten.

Affordable digital cameras with images above 4 x 6 print quality are only three years old and we are now at three megapixels, by my measure, over 8 x 10 quality.

Your question is rapidly becoming, "Film? Why?"

-iNova

 

[Daniel]

The 20 March issue of Forbes magazine has an interview with Hiroshi Ono
of Sanyo. Much big news in a small article:

1. Unbeknownst to most of us, Sanyo has been making digicams for Nikon,
Olympus and others who put their brand name on the camera. Sanyo is now
the world leader in digicam production with 40% of the market.
2. Current market is about $3 billion.
3. Digicam market is growing at a rate of 60% per year.
4. 5.4 million digicams sold in 1999, 7.5 to be sold in 2000, 10 million
to be sold in 2001.
5. All digicams are made in Japan except for Polaroid which is made in
Taiwan.
6. Digicams to outsell film cameras by 2001 in Japan and by 2002 in
North America.
7. “By 2002 or 2003, digital cameras will overtake film in
resolution.” Hiroshi Ono
8. This fall Sanyo will introduce a magneto-optical storage drive that
will cost only 1/6th as much as current flash memories and have ten times
the capacity. Each 730MB disk to cost $20.
9. “Film cameras are likely to disappear by 2005 or 2010 at the
latest.” Hiroshi Ono

Looks like film cameras will become a market oddity even sooner than most
digicam owners have been predicting, and A LOT SOONER than the film
diehards have been stating.

Rodger

 

[Robert Leigh]

Item 5 should have said ALMOST all digicams are made in Japan....

 

[FJM]

The D# pictures don't show out of focus effects or camera movement. They
were scanned by a Nikon LS-2000 if I'm reading the caption right. I've
never encountered the beast but it does seem to resolve the grain. Look
at the blue sky, for instance.

The grain of the 200 is larger and clumpier than the 100 and both show
well.

That I know to be a common misconception. You're not seing the grain.
You're seing the interference pattern between the grain and the "sampling
matrix" of the scanner (LS200). On a Polaroid SprintScan 4000 (4000 dpi)
you would see a different pattern. The ASA100 grain is really resolved at
6000 ... 8000 dpi

Common misconception or not, that pattern of disruptive clumpy sky isn't
anywhere near as smooth as it is from both of the digital cameras on
either scanner. Especially compared to the D1.

I do agree, Peter. But it is the scanner issue. A bad scan can diminish the apparent quality of film just like really poor digicam software could extract a less than optimal image from a CCD - and perhaps even more so.

With double the resolving
power in a scanner you'd be looking at the grain directly, color fleck
for color fleck but the picture wouldn't be yielding much in the way of
added useful detail.

I cannot agree with this. One would have to actually compare the results of the high res vs. low res scans in the actual practical use environment to make it a conclusive argument. Furthermore the URL I mentioned provides a bit of evidence to the contrary of what you assume here.

The 6000 and 8000 dpi scans of 35mm images that clark puts forward... how
much smaller can you shrink them without reducing the actual, practical
information to the level seen from the 4 x 5's? At least twice in my
Photoshop.

I would not fully agree that this point could be made to apply to film in general. Do not forget that the image was taken with the consumer lens. This is the low end, not the high end. With the higher end glass, under many conditions, the film is what limits the resolution, and we're talking about what film can do, not cheap lenses.

The convergence of digital sensing versus film won't be a dead issue in
favor of digital for a while, but you have to ask yourself if the current
pace of development continues until theoretical limits on circuit
refinements are reached, at which point was film rendered irrelevant? A
hundred years from now? Five? It WILL happen and my money is favoring
the "sooner than you think" camp.

1. I think if the consumer market were driving the development of digital imaging technology at the present pace without topping off until full quality parity with film were obtained (along such criteria as resolution and dynamic range and tonality (including B&W images), and low light noise performance, and sustained frames per second at full resolution / max. quality, and per-frame cost of long term storage at full resolution / max. quality, I would say it would be 10...20 years before digital would be able to overtake film.

2. However I feel that the industry will succeed to lead the consumers to believe that digital cameras have achieved if not surpassed "35mm film quality and performance level" by the time the manufactures start claiming 6...10 megapixel (interpolated) imaging capability on their products, which will be within 2...4 years.

I.e. like I said we'll be led to believe digital will have reached the film in quality before it actually does, just like we believed 35mm just about delivered medium format roll film's quality when in fact it did not. If you skip a generation and compare today's pro digital camera images to yesteryear's (to say nothing of contemporary) medium format images you'll see the trend. Now, the medium format cameras and film and processing ARE still available - but not necessarily practically accessible to the average "prosumer" as far as availability and price point as it is no longer within the mass-market mainstream with its economies of scale and revenues to further push R&D at anything other than glacial pace. I have NOTHING against digital cameras. I love digital cameras. I would just hate to have a big chunk of availability / affordability / further R&D taken away from 35mm film well before (if ever) the prosumer range digital camera technology matures to deliver, in competent hands, the full extent of the quality that 35mm film can deliver in competent hands.

FJM

 

[Will Perlis]

I'm probably getting in over my head here in such a "techy" discussion,
but I hafta ask this one... If we are truely comparing film v.s. digital
technolgy for quality here, what is the point in -scanning- the film??

Rick,

Maybe because of the underlying assumption that images are being produced for commercial and hopefully profit-making purposes. Once the quality of the end product created using digital imaging at the front end is rated "good enough" by the standards of the people who okay the checks there is no point using film.

From some experimentation with the CP900, IMX most "civilians" take a look at 8x10 prints and say "Wow, that's fantastic!". If it weren't for the super-lousy ergonomics and the various maddening delays built into that beastie, I'd still be using it for general around town shooting.

Arguments about resolution and dynamic range aside, two things 35mm has going for it are speed and ease of ...um... initial image capture , IMO. (And to preserve Peter's fingers: Yes, I'm well aware that advantage will soon disappear too. )

Will

 

[Peter iNova]

The D# pictures don't show out of focus effects or camera movement. They
were scanned by a Nikon LS-2000 if I'm reading the caption right. I've
never encountered the beast but it does seem to resolve the grain. Look
at the blue sky, for instance.

The grain of the 200 is larger and clumpier than the 100 and both show
well.

That I know to be a common misconception. You're not seing the grain.
You're seing the interference pattern between the grain and the "sampling
matrix" of the scanner (LS200). On a Polaroid SprintScan 4000 (4000 dpi)
you would see a different pattern. The ASA100 grain is really resolved at
6000 ... 8000 dpi

Common misconception or not, that pattern of disruptive clumpy sky isn't
anywhere near as smooth as it is from both of the digital cameras on
either scanner. Especially compared to the D1.

I do agree, Peter. But it is the scanner issue. A bad scan can diminish
the apparent quality of film just like really poor digicam software could
extract a less than optimal image from a CCD - and perhaps even more so.

With double the resolving
power in a scanner you'd be looking at the grain directly, color fleck
for color fleck but the picture wouldn't be yielding much in the way of
added useful detail.

I cannot agree with this. One would have to actually compare the results
of the high res vs. low res scans in the actual practical use environment
to make it a conclusive argument. Furthermore the URL I mentioned
provides a bit of evidence to the contrary of what you assume here.

If you go to the original D# page and load the full resolution image from the largest image (the one I believe mislabeled as 400 dpi) you find an image that is clearly flawed. But if you search around in the dark areas you find an amazing thing. Detail that is as sharp as pixels can resolve. Tiny, microscopic even, flecks of dust and fiber.

Now if that scanner can resolve dust at the one pixel level, how come the image behind the dust is so poor? Depth of field? No, not everywhere to the exact same extent as the dust, which has examples nearly everywhere, too. We can speculate a lot about the quality of the original shot, but the quality of the scan is really much higher than I thought when getting into this discussion.

If that original film image were as sharp as the dust on it, it would have come through much better. This image is just under 4000 pixels wide and the 35mm image here can be reduced to 60% of this size without eliminating film borne data. Try a bicubic reduction and you will still be able to read everything the film original had resolved.

I guess we'll never know some of the conditions about this shot and I'm sure that other shots could enter this particular scanner and create a better picture but the fact that an instrument this inexpensive can resolve microscopic detail so well only fuels my conviction that the days left for film superiority are vanishing rapidly.

The 6000 and 8000 dpi scans of 35mm images that clark puts forward... how
much smaller can you shrink them without reducing the actual, practical
information to the level seen from the 4 x 5's? At least twice in my
Photoshop.

I would not fully agree that this point could be made to apply to film in
general. Do not forget that the image was taken with the consumer lens.
This is the low end, not the high end. With the higher end glass, under
many conditions, the film is what limits the resolution, and we're
talking about what film can do, not cheap lenses.

I would suggest that what we are comparing is the thousand dollar solution as being the basis for an apples vs. apples sort of comparison. Lord, I hope a Leica prime is mightier.

The convergence of digital sensing versus film won't be a dead issue in
favor of digital for a while, but you have to ask yourself if the current
pace of development continues until theoretical limits on circuit
refinements are reached, at which point was film rendered irrelevant? A
hundred years from now? Five? It WILL happen and my money is favoring
the "sooner than you think" camp.

1. I think if the consumer market were driving the development of digital
imaging technology at the present pace without topping off until full
quality parity with film were obtained (along such criteria as resolution
and dynamic range and tonality (including B&W images), and low light
noise performance, and sustained frames per second at full resolution /
max. quality, and per-frame cost of long term storage at full resolution
/ max. quality, I would say it would be 10...20 years before digital
would be able to overtake film.

2. However I feel that the industry will succeed to lead the consumers to
believe that digital cameras have achieved if not surpassed "35mm film
quality and performance level" by the time the manufactures start
claiming 6...10 megapixel (interpolated) imaging capability on their
products, which will be within 2...4 years.

I.e. like I said we'll be led to believe digital will have reached the
film in quality before it actually does, just like we believed 35mm just
about delivered medium format roll film's quality when in fact it did
not. If you skip a generation and compare today's pro digital camera
images to yesteryear's (to say nothing of contemporary) medium format
images you'll see the trend. Now, the medium format cameras and film and
processing ARE still available - but not necessarily practically
accessible to the average "prosumer" as far as availability and price
point as it is no longer within the mass-market mainstream with its
economies of scale and revenues to further push R&D at anything other
than glacial pace. I have NOTHING against digital cameras. I love digital
cameras. I would just hate to have a big chunk of availability /
affordability / further R&D taken away from 35mm film well before (if
ever) the prosumer range digital camera technology matures to deliver, in
competent hands, the full extent of the quality that 35mm film can
deliver in competent hands.

FJM

I'm certain this will appear argumentative, but it is really just something I observed after the previous post when I wanted to see how much I was trading off by so wholeheartedly moving away from film.

I took the large scan from D# and started looking for flaws. Dust, poorly clumped film grain, fibers, processing dings... anything I might feel was fair game for retouching in preparation for large size publication of this image. Not that I would want to, but just to see how much touch-up this frame represented. I was merciless.

I surrounded each flaw I would have touched with a little color box. The reduced image is so small that you can't possibly see the flaws at this scale, but if you load the original you could use this as a map to point you to each spot. The exercise here has to do with the physicality of film and how it collects real world noise. The jpeg here is quite crude to save bandwidth.



While doing this, the vertical banners left of center caught my eye. I had used them in a comparison illustration a few days ago and it seemed to me that the Nikon 990 version was more "readable" than this film image scan. I had to reduce the portion of the film image to 60% of its scan size to make it pixel comparable to the native 990 slice, but this is what that produces. Of course, the film scan is gaining massive pixel-perfection from the oversampled original and the 990 slice is right off the file with no extra manipulation. Notable is the fact that the banners aren't even the same ones from shot to shot. But looking to the readability issue, I think there is a slightly clear winner.

I see artifacts in both, but the test shows my eye that a 990 holds better detail than this particular film (Fuji 200) shot, has less noise (lower grain), trades some qualities for others and compares real favorably to a 35mm film original. At least in this apples/watermelon test.



-iNova

 

[Derek Clarke]

I'm afraid not. This is a 2" disc AFAICR, so while it is much smaller than the floppy that Mavicas are built around, it still will be more suitable for bigger cameras.

And it certainly wont squeeze into a CF card.

Any idea whether this new 720mb memory card would fit into a compact ll
card slot? Reason being is cause I haven't bought my digital camera yet
and this new cheap memory card can weigh heavily on what camera to buy!!
Sure hope someone can answer this question.

 

[FJM]

Peter,

The flaws that you point out in the digital scans from the film are valid and very clearly show the incompetence and / or the lack of care in the way the 35mm film technology was used for the sake of that test. You keep insisting on comparing these results. Fine by me. If these results are all that a given user is capable of obtaining from 35mm film in those shooting conditions, that user indeed could be better off even with today's digicam. Some of us photographers do aim higher though.

 

[FJM #17938]

Maybe because of the underlying assumption that images are being produced
for commercial and hopefully profit-making purposes. Once the quality of
the end product created using digital imaging at the front end is rated
"good enough" by the standards of the people who okay the checks there is
no point using film.

I think Will makes a very valid point here. In a commercial environment, spending extra dollars, effort, or time to obtain quality beyond "good enough" to fit a specific application of images or output medium is obtained is in many cases simply wasteful (unless strong customer expectation demands a specific format or medium).

E.g. Nikon D1 and perhaps a few other better digicams as well, make perfect sense for many magazine and newspaper illustration use.

By the same token most wedding photographers stick to medium format.

However there are also applications / tasks for advanced amateur or professional photography where no "reasonable quality cap" exists - i.e. the higher the quality, the merrier. E.g. picturetaking for historic purposes, to "freeze in time" a moment in the life of a person, a building, an environment, etc. These have large (in some cases unlimited) historical scope and will sometimes be scrutinized up close - not just looked at "normal" distances to get the overall impression.

In an overly commercialized society, don't forget about those!

 

[Peter iNova]

Peter,

The flaws that you point out in the digital scans from the film are valid
and very clearly show the incompetence and / or the lack of care in the
way the 35mm film technology was used for the sake of that test. You keep
insisting on comparing these results. Fine by me. If these results are
all that a given user is capable of obtaining from 35mm film in those
shooting conditions, that user indeed could be better off even with
today's digicam. Some of us photographers do aim higher though.

Well, I'm not showing these things to upset you or be a nay sayer. D# didn't clean the 35 well at all. But I've printed thousands of positives and negatives in my former life as a film photographer and frankly, not many people can get that little chip clean as a whistle. Retouching an actual print is far less friendly than doing it digital.

Perhaps you could find an example of a full frame 35mm scan on the web or even post one so we could all directly compare an example in the full frame. Clark's small blow ups don't tell me much.

-iNova

 

[FJM]

Perhaps you could find an example of a full frame 35mm scan on the web or
even post one so we could all directly compare an example in the full
frame. Clark's small blow ups don't tell me much.

That's not an unreasonable request. I will try to find something over a weekend. Failing that I could dig out some scans a friend had done for me and / or scan some prints on my flatbed.

FJM

 

[Peter iNova]

Perhaps you could find an example of a full frame 35mm scan on the web or
even post one so we could all directly compare an example in the full
frame. Clark's small blow ups don't tell me much.

That's not an unreasonable request. I will try to find something over a
weekend. Failing that I could dig out some scans a friend had done for me
and / or scan some prints on my flatbed.

FJM

See if you can find a direct scan of a negative or transparency. I've seen the flatbed results that are good with prints and puie with transparencies. What I'm hoping for is a definitive example of practical, affordable 35mm > digital results by the shortest production path.

-iNova