|
Articles and Reviews
|
Home >> Photography >> Digital Nikon D100 - Page 4 - Digital vs. Film The Contenda' It has taken quite some time for me to produce this update to the D100 review. Some of it was taken up with figuring out what kind of shot would be suitable for comparison purposes between film and digital, but most of it was biding my time for an opportunity to scan the film via a Creo Eversmart Supreme flatbed scanner. I could have taken the easy and quicker route by having one of the professional labs to scan the film for me, but, no offence, I was not willing to spend a few hundred dollars for a handful of high resolution scans. I much prefer spending the money on other consumables such as paper and ink for my Epson printer, priorities you know. Once the timing was set-up, I had the test scenes taken with 35mm and medium format film scanned. I already had the same shots taken with the D100 in the computer and burned to disk for posterity. All taken at the same time of course. The eagle-eye readers will have already noted that the scanner used is a flatbed and I know what you’re thinking – Edwin, you cheap bugger, couldn’t you have at least tried to access a real film scanner and not compromise the results for the comparisons to digital?
So, yes, the Eversmart Supreme is a flatbed but it ain’t your cheapo consumer Epson or HP flatbed and it is certainly leagues ahead of even the better Agfa, Umax, and Microtek units costing in the single thousands of dollars. One side note though, Microtek’s top of the line XY 6000 scanner was once re-badged as a Creo Eversmart Jazz (discontinued), a lower-cost and lower-resolution version of the Supreme. It is quite conceivable that Creo has outsourced the production of the monstrous Supreme to Microtek. Any downsides to the Eversmart – only a couple I can think of:
I did some basic checking of the pro labs in Vancouver, and of the four that I found information about their scanners, three use high-end flatbed units and the fourth uses an Imacon Precision film scanner (G. King). The three-flatbed users are:
As you can see, the use of high-end flatbed scanners is common in the professional world and one should not think that using one is worst than a dedicated film scanner. I have also read some commentary by those who have extensive experience with scanners, both film and flatbed and the conclusion is that the Eversmarts, Lanovias, and XY-15s blow away the dedicated 35mm film scanners that many pros and serious amateurs purchase for their own use (Nikon, Minolta, et al). With that background out of the way, let’s get into the meat of the comparisons. The Technical Parameters The Oxygen software that runs the Eversmart scanner seems fairly straightforward but one thing I noticed right away is not being able to select an option for a full optical resolution scan of the film. There is instead a scaling option to allow the user to select the intended print output size and dpi resolution. This means that instead of doing a preview scan and then being able to crop to the intended medium size and selecting 5600 ppi scanning resolution, as would be the case for most other scanners, I had to think about how big of a print I desired first and then input the requisite parameters in the options fields. You do not want to go crazy and select a humungous print output size because you risk having the scanner and software interpolate the data if the size is more than what the true optical resolution of the scanner is. There is a crazy 14,000 dpi figure for one axis of the Eversmart scanner but that just means interpolation for the other axis. I decided to scale the medium format print by 900% (100% scaling being the original 2.2 inches of the film itself) to produce a 20x20 inch print at 300 dpi. For 35mm, I chose a scaling factor of 1500% for a print size of about 14x20 inches. The Eversmart is in no way taxed at these scaling factors and my rough estimate of the scanning resolution would be about 2700 ppi for medium format and just over 4000 ppi for 35mm film. I chose 20 inches as the maximum printing length to provide me with come cropping leeway for actual printing of 12x18 inch prints via my Epson 1270 printer. Doing direct comparisons between the scans and the D100 images would already see a lot of down-sampling of the scans, so there was no need to go wild with the resolution available for huge file sizes. The raw scans of the medium format film were already over 100 MB while the 35mm scans were in the 70 MB plus range. 35mm film Unfortunately, due to a lack of thinking on my part, I found the scans of the 35mm slides to be not good enough to use for comparison purposes. I left the film in their slide mounts and that tiny bit of space between the film and scan bed left me with scans lacking in sharpness. No amount of sharpening or technique I tried would amount to anything. The medium format films were placed directly on the scan bed and there are no qualms as to the quality of the scans. I hope to have another scanning session to remedy my oversight and will update this page with those test scans. Without the 35mm scans to compare to, I can only do direct comparisons to the medium format scans. Is this fair? After all, the D100 images originate from a small APS-sized CCD whereas the medium format film is 2.2x2.2 inches (closer to 645 once cropping has taken place). This represents a huge discrepancy for the original image capture. Technically, I would state that it is obviously not a fair comparison and one can easily guess at which format will win out. However, a number of people and commentators suggest that the various 6 MP cameras are suitable for large print sizes, even poster-sized prints and as such, this comparison helps to satiate my own innate curiosity. I also down-sampled most of the medium format scan images so that they would be similarly sized to the D100 image for the sample images shown below. Also note that none of the medium format scans have had any editing or sharpening done to them except for cropping and resaving as maximum-quality JPEGs. The D100 images on the other hand require some editing and most importantly, generous sharpening to bring out the details. Digital On its own, the D100 NEF files appear quite good after processing the images through Bibble or Photoshop and having a healthy amount of Un-sharp Masking applied (I’m finding 300-350% required for my tastes). All the primary details seem to be there, but I do find that the micro details are harder to detect – micro meaning the really fine details like the words on signs from a distance or the texture on a given surface such as a brick wall or piece of wood. However, until now, without a point of comparison, I could not suggest how much was missing and how much was not. The digital images are also quite different in colorcast to the scanned films. This is not likely a white balance issue so much as it is a profiling issue. Not a big deal as Photoshop editing can come close to matching the color of the scans or vice versa. The D100 files are quite a bit warmer, however, the extra warmth does serve my own creative interpretation for some prints I have made. Medium format Holy smokes does that Eversmart resolve details in the film. I am completely flabbergasted by the quality of the scanned image; within my limited experience these are the best digital/digitized image I have ever seen. Looking at the raw scan file, the micro details are all there with great clarity, so much so that the scan does not even need USM applied. Doing so would just accentuate the film grain that can be clearly seen at full magnification. There is indeed significant raw information in the film that the Eversmart pulled out exceedingly well. And all this from only about half the resolution the Eversmart provides. Flatbed-schmatbed, I think any film user would be more than happy with what the Eversmart can do, and so they should if they actually shelled out $100k for this big unit. However, Creo has not come up with any kind of dust removal hardware/software combination that Applied Science Fiction did with ICE (and used in various Minolta, Nikon and now Microtek scanners). As such, much editing is required to clean up the scan for actual print use. No matter how much dusting/blowing and cleaning you do, some dust always manages to end up on the film and dust will also find its way into the scanner itself. Image Evaluation and Comparisons on the Monitor It is quite plain to my eyes on my monitor that the D100 images are not up to the standards of medium format film scanned via a high-quality scanner. Despite the grain and dust specks, the medium format scans still resolve more detail than the D100 image and this comes as no surprise given the useful reviews Michael Reichmann has already done in comparing the similarly specified Canon D60 to scanned 35mm and medium format film on his Luminous Landscape web site. Print Evaluations and Comparisons Viewing an image on the monitor is not the same as viewing prints and the print comparison is the real key for most of us interested in photography. I printed the sample images on Epson Premium Luster letter-size 8.5x11 inch and Epson Premium Semigloss Super B 13x19 inch paper. Although I did edit the scan files for getting rid of the most obvious pieces of dust, I did not alter the colors, levels, curves or apply any USM, unlike the D100 images that had mostly been processed through Bibble with further refinement in Photoshop. The grain resolved by the scans were less noticable in the letter-size prints and the prints' clarity were concertina-wire sharp. The D100 images were smoother, had lower contrast and although all the major details were there, the lack of micro-detail resolvement gave a softer appearance. The larger prints carried over the status quo from the small print sizes. I cropped the industrial scene to provide the same view as the D100 image, which resulted in a 12x18 print output only at around 250 dpi. I interpolated the D100 image via Genuine Fractals for a 12x18 output at 360 dpi with some added sharpening to compensate for the larger print size. Interpolation, of course, adds no further details but Genuine Fractals claims no loss of details either and the D100 print seemed all right but upon comparing to the print from the scanned file, there was again a softness to the image and the micro details were again AWOL. Boosting the contrast and increasing the USM for the interpolated D100 shot of the condos on the waterfront seemed to help bring out some of the micro details in ths scene, however, the print from the scan file was still superior. I found it interesting to see noise in the clear blue skies on both prints. The print from the scan being film grain while the D100's was the typical digital noise artifacts that often plague such light toned areas of a scene. They looked pretty much identical. There were no surprises in the print evaluations, all they did was confirm on paper, what I was seeing on the monitor. And no, I did not view any of the prints with a loupe. I try to write common sensical reviews and articles and viewing prints through a loupe is not common sensical and provides no useful information for my everyday use and enjoyment of the prints I make. The quality of the prints was easily detected from normal and closer than normal viewing distances. Conclusion There have been various postings and musings about 6 MP D-SLRs being able to match medium format quality. Some of these comments refer to double-truck spreads in magazines, full two-page images, as often found in National Geographic or Sports Illustrated. The ability to do this type of A3 sized image seems to be a popular indicator of digital quality. For newspapers, the resolution requirements are much lower, as the 2.7 MP D1 and D1H still seem to reign supreme among digital photojournalists. Magazine image quality is higher than newspapers (175 dpi versus 300 dpi) and depending upon the magazine, even 6 MP cameras are not up to snuff for large print sizes. National Geographic has apparently stated that the Nikon D1x can only output a half-page sized image at their stringent quality levels. I think many people can live quite happily with lesser quality than the mighty National Geographic. I’ve been quite pleased with 10x15 inch prints at a measily 200 dpi run through my Epson 1270 printer and with some judicious USM applied for prints res’ed up to 12x18, the print quality is quite good on its own merits. However, in direct comparisons to medium format scans and prints, the D100 falls short of the lofty levels that some are suggesting 6 MP D-SLRs are capable of. The CCD just does not have the ability to resolve all the micro details that the larger film surface can. So no, 6 MP is not comparable to medium format and no, this is not based upon line pairs per mm or test charts or other nausea –inducing data. This is based upon taking actual images of real-life subjects and looking at and comparing the quality of the results, because how many of us would actually frame and hang on the wall, a test chart. Enough said. Time does not stand still though and in the very short time during which I did the scanning and then printing and compiling of the data, the next generation of digital products were announced at Photokina.
6 MP is now old news with this wave of new products that push the envelope of digital technology. Be that as it may, it does not mean that my D100 is now somehow irrelevant. Within its resolution limits, it will still produce great images that satisfy, however, the tests did re-ignite the flame for medium format quality. What I really want now is an affordable medium format digital film back for my Bronicas, but I’m under no illusions that anything of the sort will ever come about. Kodak produces the self-contained film back that I desire but at such extraordinary costs, that I would do better buying their new DCS Pro 14n instead if I wanted higher resolution. Given its consumer origins, similar to the D100, my interest in the unit is low and I expect Nikon to introduce the much-anticipated D2 at PMA in February 2003. Not that I would be able to afford either in any case… Sample Images In case you're wondering how a scan from my cheapo consumer scanner fares against the Eversmart, here's a pixel-magnification sample below.
Note the detail apparent for the medium format scan above left. You can see that the wall in the lower middle is made of bricks and the glass walls on the top left come out clear enough that one can count the individual glass blocks. There is either none or minimal details available from the D100 image at right. The full size 20x20 inch original scan is even better (of course). |
|
|||||||||||||||||||||||||||||||||||
Correspondence & About this website Copyright © 1998-2008 Edwin Leong |
|
 |
|
How
about some basics about the Creo Eversmart Supreme?
