Scanner review Plustek Optic Film 7600i, a different conclusion

[ThomasH_normally]

Considering my film collection, I am always happy to see that someone really still cares and analyzes the scanning equipment.
However, in this case it strikes me how different are the conclusions of Mark Segal's test compared to the biggest scanner test collection on the web by Patrick Wagner:

FilmScanner-Info-Pages

In this test its author claims that the effective resolution of the scanner is mere 3250dpi, even less than the predecessor model OpticFilm 7500i. Of course even M.Segal's test images show that this scanner has by no means higher resolution than the Nikon LS5000, thus Wagner's conclusion seem to have merit. Wagner writes: "It is annoying if one purchases a scanner that in practice only provides 43% of the resolution that is actually promised by the producer."

Segal also mentions Epson V750, here I would like to direct your attention to Filmscanner.info as well. Clearly the V700/750 are admirable performer for flatbed scanner, but many people often do not see the clear difference to V600 based only on specifications from Epson. Neither did I, until I saw the tests: the effective resolution of V600 is mere 1560ppi!! Conclusion: "This is less than a quarter of the resolution of 6400ppi declared by the producer." I was quite shocked by seeing this difference, Epson's image got a dent in my book.

If you have lots of uncut film and look for an alternative to Nikon LS5000 and SA30, lookup some of the Reflecta scanners on Filmscanner! Similar in construction and resolution to Plustek, some models simply pass the entire roll of film horizontally without any adapter. Much better than the convoluted SA30 adapter from Nikon. But... LS5000 can also use the SF200/SF210 slide feeder, the Reflecta does not provide automated slide scan.

However considering the present ebay-only price for LS5000+sa30+sf210, it might be more cost effective to take two scanners, the Reflecta for film rolls, and something else for the mounted slides. Many scanners come bundled with Silverfast, just like the Plustek.

Thomas

[Ray]

Thomas,
Thanks for the link to that scanner review site.

I guess there's a limit to the amount of time one can spend scrutinising and comparing the resolution output of scanners. To do that properly one ideally needs a sharp photo of a resolution chart. Mark did comment and demonstrate in his PDF that the Nikon 5000 ED produces marginally more detailed results than the Plustek, in real-world images.

In general. I would say Mark's conclusions are pretty similar to ScanDig's in the sense that the scratch removal really does work and is a great improvement on the earlier model.

However, Mark was probably not in a position to actually quantify any resolution differences or compare the resolution with the earlier Plustek model, the 7500i which apparently had a slightly greater resolution potential.

What surprised me in the ScanDig review was the claimed maximum resolution figure of 3250 dpi. In itself, that's quite respectable. Not too far behind Nikon's resolution of 3900 dpi (averaging vertical and horizontal).

What is perhaps disturbing is that Plustek claim 7200 dpi for their scanner, which is downright misleading. Nikon on the other hand claim only 4,000 dpi.

The other issue is that, in order to achieve that maximum resolution of 3250 dpi from the Plustek scanner, one has to scan at full optical resolution generating files of 400mb in 16 bit mode. That's a terribly slow process.

On the issue of Dmax, the good news is that the multi-exposure option allowing for the merger of an overexposed and underexposed scan, really does improve the quality of the shadows, using this latest model. In the previous Plustek model, multi-exposure apparently didn't have much effect.

[Bart_van_der_Wolf]

Segal also mentions Epson V750, here I would like to direct your attention to Filmscanner.info as well. Clearly the V700/750 are admirable performer for flatbed scanner, but many people often do not see the clear difference to V600 based only on specifications from Epson. Neither did I, until I saw the tests: the effective resolution of V600 is mere 1560ppi!! Conclusion: "This is less than a quarter of the resolution of 6400ppi declared by the producer." I was quite shocked by seeing this difference, Epson's image got a dent in my book.

Hi Thomas,

A gentle word of caution, the resolution test is somewhat flawed. The target that was used is not appropriate for quantifying a CCD scan in such absolute terms. It was designed for analog systems, film, not for discrete sampling sensors. The USAF target will give a visual clue, but it is too dependent on accidentally lining up with the sensels to derive accurate quantification.

To give you an example, I looked at Wagner's site. The only scanner I could find that I also use was the Epson V700. I ran a quick test with a proper method (there is an ISO standard for it), using a slanted edge target to derive an MTF curve. That showed me that features in the vertical scan direction (so horizontal resolution, the actual resolution between sensels, not carriage), the scanner performed better than Wagner's 2300 dpi (small niggle, it should be expressed in PPI). My results showed 49.1 cycles/mm or ~2494 PPI (close but higher) at 10% MTF response (resolution at 10% is considered to correspond well with human visual limiting resolution). The V700 continues to discriminate detail right upto some 90.874 cy/mm or ~4616 PPI where the response is reduced to only 1.562%. While sharpening may recover some detail (although adding film MTF will reduce that resolution) at such low responses, it wouldn't look pretty at that resolution, noise and all. Part of the difference in testing outcome could be explained by the lack of a focusing facility for the V700, but the testing methodology also has a significant impact on what the outcome is.

Another thing that needs to be considered, besides an absolute resolution figure (at some arbitrary cut-off point, when does sharp become unsharp), is the way the scanner controls things such as flare (which hugely reduces dynamic range), something flatbed scanners are usually not very good at. Why the interior of such transparency enabled scanners e.g. is not black but bare metal, still puzzles me, although cost reduction comes to mind. Also the sensor technology, such as Epson's double rows of staggered (half sensel offset) sensels, delivers a different MTF shape with regards to aliasing reduction (which will reduce grain aliasing), so there's more than resolution alone to consider when comparing scanners.

Cheers,
Bart