relativity

[david mertens]

Undoubdtably, the finest photographs ever captured are those expresing the human emotionion, or exposing the human condition, i.e. grief of a father, worry of the mother, inocence of the child, horrors of war, beauty of raw nature, beligerance of youth, sexuality of the young, etc,, that being said and understood, the question I have is this, before all chemically-based photography is discarded to history....What is the highest resolution film photograph ever taken, with details, such as film type, format size, and speed, camera, lense f/stop, lighting, shutter speed, that you have ever witnessed? How does this compare to a "common" 10 meg digital photograph?

Please excuse my spelling.

Dave wants to know.

[Ralph Barker]

FWIW, a 4,000 DPI drum scan of an 8x10 negative generates a 1,280 megapixel image. For improved resolution, a format larger than 8x10 is probably better. ;-)

[Bruce Watson]

Undoubtedly? I doubt very much indeed you'll get consensus for your hypothesis. But if that's what floats *your* boat, go for it.

Otherwise, I agree with Ralph. If you want more detail in your film, get bigger film. I'd hate to think about just how much detail you could get out of a 20x24 sheet of FP4+.....

Oh, why not:

(5000 ppi)(20 in)(5000 ppi)(24 in)(2 bytes/pixel) = 24 GB, 16 bit grayscale.

Well beyond the capability of modern scanners, desktop computers, and programs like PhotoShop.

[chris jordan]

Dave, I think some of my work might be pushing the envelope in terms of high resolution images from film. I shoot 8x10" originals and scan them at 2000 dpi (contrary to popular belief, using higher resolution doesn't provide any additional image information, just more grain), and that process produces files almost a gigabyte in size that have image details as small as one pixel. A few of my images are multiple 8x10's laced together; I have a couple of panoramic images that are composites of four 8x10's in a row-- the equivalent of an 8x40" film original. They are a nightmare to lace together in Photoshop, but the resulting prints have more detail than our eyes could see if we were there in person.

However, I think there is a limit to how useful that much image information can be; for my own work it matters because my images are all about small details, but for other kinds of work, too much information can actually reduce the emotional effect of the image. I've seen some very high resolution landscape photos recently, whose only interest is their extreme sharpness. My own reaction to a lot of that kind of work is, "amazingly sharp, but so what?"

It's kind of like 24-bit digital audio recording-- nowadays just about any musician can make an awesomely high-quality digital recording, but all that really does is level the playing field; there still isn't any substitute for good musicianship. Those scratchy old John Lee Hooker albums still beat the pants off of plenty of perfectly noise-free digital recordings of uninspired notes.

~cj

www.chrisjordan.com

[david mertens]

Cris,

You are getting closer to the answer I seek. Hogarth and Ralph missed the point. I could 4000 dpi drum scan a sheet of black construction paper and get the same number of pixels in a file as a finely grained 8 x 10. ..

Hogarth has nominated one component, FP4+ film. I am not familiar with this film, but I'll look it up. Now I am looking for a format/lens combination known to produce the "sharpest" negatives, i.e. information per inch, regardlesss of the total information contained.

Dave

[Ralph Barker]

Missed the p0int? I don't think so, although it might seem that way on the surface, David. I think the real point is that even within a 4,000 DPI scan, a large negative will have a level of detail that a hi-res scan or current original-digital capture technology is simply unable to capture.

A while back, I did a commercial shot of several mid-sized computer tape backup systems for a magazine. The layout included three tape "libraries", each about 24-30" in height, and their respective tape cartridges (different formats), in front of them. I shot it on 4x5, using Fuji Provia 100F, with the height of the libraries occupying about 80% of the landscape-orientation frame. With a 20x loupe on the transparency, I could clearly read the 4-point type at the bottom of the 4mm DAT cartridges in front of one of the tape libraries in the image. That's a level of detail that a current 10, or even a 14-megapixel digital camera would not be able to capture.

[chris jordan]

David, my own rig consists of an 8x10" monorail camera, Fujichrome Astia 100F fim (the finest-grained color film made), and top-quality optics (a Nikkor 450M and a Rodenstock 240 Sironar "S"). Assuming I get my exposure right and all that, then my film originals are wet-mounted and scanned on a Heidelberg Tango (which is the best scanner made), and printed on an Epson 9600, which is the highest-resolution digital printer made. Inbetween is much tweaking in Photoshop, which requires a great deal of skill to optimize the quality of an image.

Images made with today's digital cameras are stunningly smooth and sharp and accurate-- far more so than even the best film. But so far, the capture devices are not big enough to create a file that has as much information as an 8x10 film original-- that is why a scan from an 8x10" original has so much more information than a digital file from a 10 megapixel camera. If you compared the files inch-for-inch, the digital file would be better; it's just that the digital file is 1x2 inches compared to the 8x10" film. If they ever make an 8x10 digital capture device, it will be incredible. That's what I'm hoping for from Santa someday: a "digital film holder." That would be an 8x10 digital capture device that is the same size and shape as an 8x10 film holder, that slides into the back of the camera (with no additional cords, laptops, batteries, etc.), and captures a 2 GB file, which would be stored in its 50 GB memory. (You listening, up there in the north pole?)

Unfortunately, by the time they make such a thing, I will be too old and frail to carry the camera around anymore...

~cj

[Bruce Watson]

Somebody is missing the point, I'll agree with that.

The information in film is contained in the grain clumps, and the spacing of these clumps. Even chromes start out this way, before you process and come out the other side with something analogous, which is dye clouds, and the spacing between dye clouds.

The grain clumps are spaced stochastically. Scanners scan deterministically. Think of scanning as laying down grid on top of the film. The scanner "sees" through the holes in the grid. Because the grain clumps are arranged stochastically (think randomly), the grain clumps and the grid holes don't normally line up. Which leads to interesting effects which I'll leave as a thought exercise for the reader.

In order to get all the information off the film, the scanning grid has to be about equal in size to the average film grain clump. For large grained films like Tri-X, this is around 4000 ppi. For a really fine grained film like Velvia, it's up over 6000 ppi.

If you push the scanner resolution up above the grain size, you start to learn more about the film base between grain clumps. This is more data, but I don't think it's more information.

If you want more exact and complete information, talk to the folks at the last American drum scanner manufacturer, Aztek. They've apparently spent decades studying this very issue, and can tell you a lot more about it than I can. I had many an email exchange with Phil Lippincott who ran Aztek for about 25 years before he died this fall. This engineer believes that he knew whereof he spoke on most things scanning.

[Bruce Watson]

Heidelberg Tango (which is the best scanner made)

I'm not a believer in the modern myth of the Tango. It's certainly not the sharpest scanner out there, by a long ways, and we are talking about sharpness here.

Drum scanner sharpness is largely determined by aperture size. The Tango is limited to 11 microns if memory serves. A number of desktop scanners have 6 micron apertures, including some of the Howteks, the Optronics ColorGetters, the ICGs, etc. A few have 3 micron apertures, such as the Aztek Premier and the newest ICG, again if memory serves.

When you test these scanners with USAF 1951 resolution targets, the Aztek Premier is clearly the best of the best. It is one of the few (only?) scanners that can out perform film, if just barely.

For the record, I don't own or use either a Tango or a Premier. But I do own and use an Optronics ColorGetter 3 Pro, from which I get better scans than I've had done by others on Tangos. This is one of those things that depends a lot on the operator, so YMMV.

[paulr]

"In order to get all the information off the film, the scanning grid has to be about equal in size to the average film grain clump. For large grained films like Tri-X, this is around 4000 ppi. For a really fine grained film like Velvia, it's up over 6000 ppi."

This is presupposing that there's any useful information at the resolution of the minimum grain clump size. Using your numbers, detail at the grain clump level on Tri-X would require 78 lp/mm to resolve. Velvia would require 118 lp/mm. At f22, diffraction limits the modulation of any lens to a near-invisible 5% or so at 78 lp/mm. This is not even taking into account other elements of the real world: mtf response of the film itself, lens aberations, focus errors, movement, etc.

Resolving useful photographic detail at 118 lp/mm strikes me as fantasy.

What this suggests to me is that it's a dead end looking for detail or improved image quality at the very high frequencies of the grain level. The gold is at the lower frequencies where your eyes pick up the cues that indicate sharpness. Not to mention at the frequencies that are visible. Notice that the german lens manufacturers give MTF information at 5, 10, and 20 lp/mm. Not at 40, 60, 80, or 100. They know what the important frequencies are.

The only argument I can think of for capturing resolution at the grain level is if you are making very large prints, and you want to resolve the grain for its own sake.

[paulr]

Another thing ... I suspect that when scanning resolution approaches the implied resolution of a film's grain clumps, you can run into problems. To avoid aliasing and other weirdness, you probably want the scan resolution to be significantly more or significantly less than whatever the implied grid is of the film grain (as Hogarth said, grain behaves stochastically, so the implied frequency will vary depending on the film and on the tone being reproduced).

My suspicion is that this is the reason flatbed scanners seem to have about half the optical resolution of their physical digital resolution--not because the manufacturers are being cheap or stupid, but because they know we wouldn't like the results if we pointed perfectly focussed ccds at film grain at those resolutions.

But to indulge the original question, this site has some good data on different films:
http://www.normankoren.com/Tutorials/MTF1A.html (http://www.normankoren.com/Tutorials/MTF1A.html)

[Bruce Watson]

PaulR,

What you say is true. What the film can record, and what we can get to it, are not the same things. Most any film will outperform the lenses in large format work.

I think now I better understand what the original poster is asking - which I now think is "where is the sweet spot of maximum system resolution?" I'd like to know the answer to that one also. Perhaps it's in 8x10, as Chris says.

[chris jordan]

I have long been a fan of lower-resolution scanning. If you scan all the way to the film grain, then your scan has grain in it, which is all different colors. For example, if you photographed a grey cement wall and scanned it all the way to the grain, your scan would contain millions of little brightly-hued grains, sort of like the colored dots in a magazine page that from a distance look like grey. When you start doing your Photoshop work on this file, you will run into lots of problems trying to manage all the colored pixels.

However, if you scanned only to the edge of the image detail, but not all the way to the grain, then the pixels in your scan would all be grey as they are supposed to be. You would still have all of the detail that your lens focussed on the film, but without the extra noise of the film grain.

A super sharp scanner might be able to make the grains look sharper on a scan, but it can't make the image detail any sharper, because the fuzziness of the image detail is caused by the lens. So, despite the existence of sharper scanners than the Tango, I still prefer the Tango. It produces a beautifull smooth and photographic-looking result, while still easily capturing all of the image detail that my camera lens can resolve. Some carefully administered unsharp masking can then zing the image details into slightly sharper resolve without bringing out the underlying grain structure.

~cj

[paulr]

the beauty of unsharp masking, if you do it well, is that it allows you to choose which resolution gets its mtf boosted. Unsharp masking actually reduces the total amount of detail in an image, but it does it at frequencies higher than the one that gets sharpened. What gets thrown out is the frequencies that contain noise, including grain (which is essentially analog noise).

[Ellis Vener]

After reading your question I formed one of my own: "Why does an apple not taste like an orange?"

[david mertens]

Same question....

A beautiful, young naked figure is half-sitting on a stool our rented studio. Unfortunately, we are all bio-molecular scientists, and are interested in the very small parts that make up the whole, rather than appreciate the beauty of the collective molecules.

Paul, Chris, Horgarth, and Ralph are invited to bring their best film gear to compete for the "most got it together at the edge of the envelope biomolecular photographer of the year award".

Ellis is not invited.

From skin to film, there are four controllable components. Lighting, lens, camera, film. Every biomolec's lens must stay at least 10Ft from the surface of the subject, at least while the shutters are open.

Regardless of the format, all results of the session will be judged at "8"x"10", wether a reduction, enlargement, or contact print is required.

The prints will be judged on the following biomolecular-scientific scale, using the highest power loupe that still produces any discernable detail:

"I can see molucules!" 10 points

"I can see skin cells!" 9 points

"I can see (something between skin cells and pore-couldn't think of anything) 8 points

"I can see every pore!" 7 points

"I can see an arm! 0 points

You of course want to win this competetion, so you bring the best stuff you can beg, borrow, or steal.

What would you bring to the competition for each of the components to be sure to "win":

Lighting?
Lens?
Aperature?
Camera (format)?
Film?

If Ellis shows up anyway with the finest digital camera available, could he outdo you? What gear do you fear he will crash with?

Oh yaa, winner gets the girl!

[QT Luong]

There has been some press recently on custom-made cameras using 9x18 film with builders of those devices claiming to make to make among the highest-res images ever.


http://gigapxl.org/technology.htm (http://gigapxl.org/technology.htm)


http://patft.uspto.gov/netacgi/nph-Parser?Sect1=PTO1&Sect2=HITOFF&d=PALL&p=1&u=/netahtml/srchnum.htm&r=1&f=G&l=50&s1=6,795,648.WKU.&OS=PN/6,795,648&RS=PN/6,795,648 (http://patft.uspto.gov/netacgi/nph-Parser?Sect1=PTO1&Sect2=HITOFF&d=PALL&p=1&u=/netahtml/srchnum.htm&r=1&f=G&l=50&s1=6,795,648.WKU.&OS=PN/6,795,648&RS=PN/6,795,648)

[Struan Gray]

Sigh. You people just don't know when you're beat.





http://www.sljus.lu.se/People/Struan/pics/si111.jpg



Silicon atoms at roughly 1000000:1</br>
(Bellows factor is left as an exercise for the reader.)

[paulr]

"Paul, Chris, Horgarth, and Ralph are invited to bring their best film gear to compete for the "most got it together at the edge of the envelope biomolecular photographer of the year award".

In this situation I'll trade in my view camera for some nice beer goggles.

[Bruce Watson]

The Tango is limited to 11 microns if memory serves.

Nope. My memory served, but the information it contained was wrong. The Tango scanners have minimum apertures of 10 microns. This from Karl Hudson, Hudson Grafik Services, Inc (http://www.hudsongrafik.com). Karl maintains and services Tangos; he should know.

[Drew Wiley]

Do a little research into survelliance films. There were contests before the Second
World War to see how many texts of the Bible could be placed on a single microdot.
I've handled color photos of a surfaced Russian sub taken from a mile away where the names of the officers on board were known by reading the name tags on their lapels! Of course, some incredibly expensive lenses were also involved. Nothing the public calls
digital is even remotely in the ball park - but since we don't know much about classifed
films, we probably don't much about classified digital capture either.

[Robert Hughes]

Whatever became of those exotic films & lenses? Did any of them make it into commercial production? Or were they all thrown off the back of an aircraft carrier, along with the Arc of the Covenant, and the Area 51 UFO's?

[Drew Wiley]

Robert - the folks who made those lenses are still in business, though my personal contact who I used to chat with regularly has retired. They were directly involved in the correction lenses for the Hubble space telescope too. They custom make lenses for the DEA, NSA, etc, typcially at a cost of over a hundred grand apiece, so I don't think you'll see them for sale retail anywhere soon! What I can't figure out is how they get some of these lenses to see through atmospheric conditions, though it might be analgous to some of the magnetic image alignment used in extremely high-end optical
microscopes. They were demonstrating a lens system for the Coast Guard and took a
photograph across Boston Harbor when it was filled with fog and the visibility was only
a couple hundred feet. Two miles away there was a car being broken into and stolen,
in a parking lot clear across the harbor. The resulting film photograph was so sharp that it was handed over to law enforcement and the two thieves were convicted! And that was some time back. Same thing with the sub shots I saw. They were shot through heavy atmosphere. In this high tech neigborhood there are all kinds of James
Bond technologies in production. Most people know about Silicon Valley and the
gazoolions of dollars it produces, but defense tech is nearly as big around here, and
pharmaceutical/biotech nearly as big. Someone at this moment is probably even working on a digitally-controlled version of the Frankenstein monster, but with an
owner's manual too long for anyone to read.

[Drew Wiley]

Should have added, Robert -since you think I'm crazy anyway - I did see a UFO once,
when in a desert mtn range. Lifted up my head from the darkcloth, and there was this
huge gray thing with a nose like a Klingon warship, stubby wings with four absolutely
enormous engines, totally quiet, no vapor trail, just a couple hundred feet above me.
No, it didn't say Klingon Empire but US Navy. Probably a billion dollars worth of structural graphite there, and some kind of incredible mufflers on the engines. One of
the stealth prototypes before mass-production of the current Delta-wing designs. My
nephew was along and snapped a shot with his Pentax. Took us six months before we
found out the actual model number.

[Robert Hughes]

Someone at this moment is probably even working on a digitally-controlled version of the Frankenstein monster, but with an
owner's manual too long for anyone to read.

"If the iCreature begins to choke your new bride, press <ctrl><alt><delete> on your remote for the reboot menu. Otherwise you may log in through our customer support site."

Oh, shoot! No batteries! Help?