"silimar in sharpness? I meant "similar" That's what I get for not proofreading before hitting submit.
"silimar in sharpness? I meant "similar" That's what I get for not proofreading before hitting submit.
After trading e-mails with Glenn on this issue, I got to thinking about how critical we are being here. Glen stated that at normal viewing distances you would not be able to discern any differences. He is right, of course. But I replied that if you get closer you can see the increased detail and reduced grain from the larger original.
After sending my reply to Glenn I reflected on something that happened yesterday. It's a bit of a story, but bear with me because it has a happy ending.
I'm working a six-month contract job at a high-tech firm in Silicon Valley. They auctioned off some old computers and unused office equipment this week. I went downstairs yesterday morning to see if there was anything interesting, and saw a black and white photo leaning against the wall. Hmmm. Nice shot from a distance. Turned out to be a 16x20 archivally matted and framed fiber print of Ansel Adams' "Aspens, Northern New Mexico." High bid at the time was $10.50. This was s silent auction where people wrote their name and bid on a sheet of paper taped to the item.
Not wanting to drive up the price, I decided to wait until a few minutes before the auction ended, fully expecting to pay $200 for this masterpiece (not an Ansel print by the way, it was printed by an assistant. I checked the Adams Gallery web site just to be sure it was what I thought. It was.
Ten minutes before the auction ended I went back downstairs. The bid was up to $14. This is robbery. I bid $18 hoping to discourage anyone from incrementing by one dollar. Just as the auction was ending the lady who had bid $14 walked up and asked me if the $18 bid was mine. I told her it was.
She said she was decorating her new house, she really liked the frame, and she wanted the photo. I told her I'm decorating a new house too, I shoot photos in my spare time, and I was prepared to outbid her. I asked her how high she was willing to bid. "$50," she replied.
Stunned, I told her I would gladly pay $200 for the photo. She looked at the photo, looked at me, and said "It isn't that good a photo. I could do as well up in the Santa Cruz mountains myself." At that I was determined to buy the photo at any price. I had visions of her throwing away the photo and putting a cheesy shopping mall picture in the frame. She broke my train of thought when she said. "No. I'd never pay that much for that photograph. I paid $200 for a 3D photo last weekend. This one isn't worth more than $50."
I hung the Adams photo on my living room wall last night.
The point of my ramble here is that what we're arguing, while interesting, is academic. We're quibbling on fine points along the lines of which Zeiss lens is sharper, a Rollei or a Hassleblad. Ooops. Wrong forum. Make that which LF lens is sharper, a Rodenstock or a Schneider.
The vast majority of people neither care nor could see the difference between a 6x9 printed to 24x30 or a 4x5 printed to that size using traditional methods, much less drum scanned and printed on a LightJet 5000. But I do find all this discussion fascinating.
So who's going to let me borrow a Rollei 2.8GX?
Darron,
actually, it is a very interesting dialogue and I don't view it as a Canon/Nikon debate at all. it does have to do with identifying a print quality that meets your criterium, and working backwards, to establish the minimum format you could use and be pleased with the results. I just bought this Linhof Technikardan, mostly out of the 'zen' of working with a precision machine, and slowing my pace. I am sure there will be many a time, that I will agonize over that fact that 6x6 would yield the same results and be a lot easier to tote around. and then again, my neighbour, Christopher Burkett, would argue that even 8x10 is too small at times!
Hi Darron;
"Noise floor" is a commonly accepted term for what you were trying to describe in the image capture world.
High-end drum scanners use photo multiplier tubes as their capture elements. PMTs have inherently superior noise performance compared to CCDs for any given level of sensitivity (scientific CCDs do have very good noise performance, but their sensitivity is low, so you really have to blast them with light to attain full saturation with a reasonable line period).
That having been said, there are CCD-based scanners on the market which offer extremely good noise performance. I've had a few 4x5's scanned on a CCD-based Scitex flatbed recently, and the results were good enough that I frankly can't see any reason to drum-scan them. Of course, the Scitex in question is a $50K scanner, so it's really not any cheaper than a drum.
All of this discussion has been wonderful, and of course, if simply carrying an 8x10 guaranteed that my images would all be like Christopher Burkett's, I'd already have one.
But Patrick's post about his experience with a "low-cost" scanner, and the balloon analogies, raise a question that I am wondering if someone can answer, and that concerns the independence of sampling area on adjacent pixels.
I work with Landsat images which are processesed to yield 28m pixels (ground resolution). But we always know that what we take to represent a 28x28 meter patch of ground is really a number that represents light entering the instrument from a circular gaussian distribution that overlaps adjacent pixels.
Given the optics and mechanics of the scanner and light scatter in the film base, how "independent" are pixels scanned at say 5000 dpi on a Tango (or for that matter at 11,000 dpi which I guess the Tango quotes as an upper limit), or what kind of footprint does the PMT actually see?
Afterall, what we are all discussing can be viewed in a different context. Given a print size and input resolution to the LightJet (res80 or res120), the input file will have the same number of pixels regardless of the original size of film scanned. What makes the three (RGB) 8-bit integers in those pixels "better" if the scan comes from a larger original? Clearly noise, film grain and pixel spread are some of the factors.
This discussion really leads back to the fact that most printing paper has a resolution that is lower than film. Once you have chosen a print size, then whatever the reproduction method you eventually reach a point where the extra detail on a larger negative simply cannot be expressed on the paper. Exactly where that point occurs is a subjective issue, but it has to occur somewhere.
Basic information theory would imply that finding the point of no extra print quality should be a simple mathematical excercise, but it turns out that the 'point' is approached gradually, so there is always room for personal choice in exactly how redundant you like your image data to be. In analogue imaging grain complicates the issue further.
This is true even in the digital world. Glenn mentioned remote sensing data, in which the pixel density is higher than the optical point spread function (roughly speaking, the resolution). The same is true in scanners, and in fact the optical path in many scanners is deliberately designed to blur the projected image so that adjacent sampling cells overlap slighly. Doing so automatically prevents aliasing (it's exactly the same as band-limiting an audio signal before digitisation) and is one reason why scanners which allow adjustment of the illumination level do so by changing the brightness of the light source and not with an iris diaphragm which would upset the amount of blur.
Drum scanners don't use imaging optics as such, but instead scan a spot of light across the film and measure how much passes through at each pixel position. Their resolution limit is set by how well you can focus the spot (which is ultimately determined by collimation of the light source and diffraction in the focussing optic) and such things as control of scattered and stray light. There are specialised techniques (SNOM: surface near-field optical microscopy) which allow you to go beyond the diffraction limit and measure optical absorbtion on the atomic scale, but nobody does that with film - at least, not when scanning to make a print.
Provided a scanner is 'good enough', it's true spatial resolution is not actually very relevant because beyond about 4-5000 dpi normal pictorial films are all grain, and increasing the scan resolution merely reproduces the grain structure at higher and higher magnification. Photographers value the result because prints with corner-to-corner sharp grain are regarded as good, but the amount of image information isn't changed much.
Taking an offshoot from Darron's latest post, I think a more interesting comparison to 4in x 5in vs. 6cm x 9 cm drum scans/lightjet prints, is 4" x 5" traditional prints (such as cibachrome) vs. 6 cm x 9cm drum scans/lightjet prints. Has anyone done any empirical comparisons along these lines? I'd be very interested in the results. My guess is that the 6 cm x 9 cm drum scans and lightjet print will be superior to the 4" x 5" traditional, cibachrome print. (Yes, that's right, the smaller original should, IMHO, yield a superior final image because the printing process is so much better.) Howard http://www.naturelandscape.com
Let me add my voice to Howard's for feedback on the 6x9 (or 6x7 or whatever MF you want to choose) digital vs. 4x5 "analog" print comparison. I've been curious about this one for a little while, and have spoken to a few people about it. My next step is an actual test (I know, the only way for me to really know). But in the meantime...I too would appreciate any information from folks who actually have made that comparison already.
By the way, in my talks with Bill Nordstrom and the folks at Calypso and Evercolor, they agree with Howard's thesis: they all say that a digital print of a certain size from 6x7 film will beat a conventional print of the same size from 4x5 film. Does anybody with experience agree or disagree?
for the record, I took my first image with the Technikardan on Polaroid 55pn, cut up the negative, and scanned it in as a panoramic 35mm negative on my PhotoSmart scanner at 2400dpi. the tonality is much smoother than the corresponding 35mm negative scan, and grain appears to be an order of magnitude less, as it required less unsharp masking and less sizing. smoothness of tone transition is the overwhelming observation, and I am very pleased with the results printed on my Epson 1200 in 1440dpi mode.
with the new Quadtone inks, I'll be ecstatic!
This may have no relevance to the original question of scanning 67 vs45, but here goes. I had my 1st roll of 35mm hi-res scanned "Kodak picture CD".I manipulated them to the extent the software on the disk allowed-sharpness/framing/contrast/brightness. I printed the results on my very low tech HP 820 ink jet printer on el-cheapo plain old paper-8 1/2X11. The results were "pointilist paintings". They really looked fantastic.I mounted 2 of them and they catch my eye every time I walk by. Has anyone out there had this result? Re: digital-this months Shutterbug had many articles on same. George Nedleman
Posting Permissions
Reply With Quote
Bookmarks