Mark Segal's Article on Epson V850 Scanner

[BradSmith]

Mark,
You recommend purchasing one of Silverfast's IT-8 targets to go along with the scanner and upgraded Silverfast software.     
I went to their web site and for film, they only seem to have transparency targets so I assume that is what you meant.  Correct?
They offer different targets by film brand and in sizes and two different types: Calibration Target and "Individually Measured Target".  I assume you meant the Calibration Target.  Correct?   
I also assume one should purchase the target in the most common film size they will be scanning.  Correct?

Thanks for the highly detailed article.
Brad Smith

[Ray]

Thanks for the very detailed review, Mark. When I first glanced through your article rather quickly, I thought it was a bit brief and not particularly helpful. However, on the second reading I saw the link to the very long pdf which contains numerous test examples. There's so much there to digest.

One point you make in the pdf, which puzzles me a bit, is the following comment.

(“Right--‐sizing” scan resolution is controversial, some claiming that maximizing resolution and
then downsampling produces cleaner scans. My research indicates no comparative advantage to this approach.)

Would I be correct in assuming that in all your comparisons between different scanners, you have not always used the maximum optical resolution of 6400 ppi that the V850 is capable of?

When I bought a V700 Epson scanner about a year ago, and scanned (or rescanned) a few hundred 35mm slides and negatives, my distinct impression was that a 6400 ppi scan, at least in the few tests I carried out, produced slightly sharper and more detailed results, compared with a 2400 or 3200 ppi scan, provided the 6400 ppi scan was fully sharpened in Photoshop before downsampling to 2400 or 3200 dpi.

If the image was downsampled before sharpening, then I saw hardly any benefit in the full-rez scan. My explanation is that the larger file size at 6400 ppi provides the advantage that more sharpening and less destructive sharpening can be applied, and that this advantage becomes apparent after downsampling to the same file size as the lower-rez scan.

[Bart_van_der_Wolf]

Thanks for the very detailed review, Mark. When I first glanced through your article rather quickly, I thought it was a bit brief and not particularly helpful. However, on the second reading I saw the link to the very long pdf which contains numerous test examples. There's so much there to digest.

One point you make in the pdf, which puzzles me a bit, is the following comment.

Would I be correct in assuming that in all your comparisons between different scanners, you have not always used the maximum optical resolution of 6400 ppi that the V850 is capable of?

When I bought a V700 Epson scanner about a year ago, and scanned (or rescanned) a few hundred 35mm slides and negatives, my distinct impression was that a 6400 ppi scan, at least in the few tests I carried out, produced slightly sharper and more detailed results, compared with a 2400 or 3200 ppi scan,

Hi Ray, that's correct.

provided the 6400 ppi scan was fully sharpened in Photoshop before downsampling to 2400 or 3200 dpi.

This is not absolutely necessary to benefit from the (marginally) higher resolution that a 6400 PPI scan delivers. However, now that we have mostly avoided grain-aliasing effects, and are close to over-sampling the resolution that the V7xx/V8xx scanners can deliver, we do have an opportunity to sharpen in detail (and even do some selective noise reduction), and optionally down-sample which will reduce random noise that originated from the scanner itself.

Besides the burden that large scan files bring, they also allow to do some detailed retouching/sharpening, as you have noticed. If the subsequent down-sampling is done carefully, we do not lose too much resolution, if any, and have a better quality and more manageable file size to deal with.

Cheers,
Bart

[Ernst Dinkla]

Hi Ray, that's correct.

This is not absolutely necessary to benefit from the (marginally) higher resolution that a 6400 PPI scan delivers. However, now that we have mostly avoided grain-aliasing effects, and are close to over-sampling the resolution that the V7xx/V8xx scanners can deliver, we do have an opportunity to sharpen in detail (and even do some selective noise reduction), and optionally down-sample which will reduce random noise that originated from the scanner itself.

Besides the burden that large scan files bring, they also allow to do some detailed retouching/sharpening, as you have noticed. If the subsequent down-sampling is done carefully, we do not lose too much resolution, if any, and have a better quality and more manageable file size to deal with.

Cheers,
Bart

I did not read the PDF thoroughly yet but it looks like one of the better articles written on LuLa for some time. Thank you Mark.

The Epson V700 up to the V850 oversample without doubt when their highest Samples Per Inch setting is used. The sensor design itself is aimed at oversampling. I recall a technical article on the sensor and a similar one used by Umax, the well/pixel size is larger than the pitch between them, achieved by more linear CCDs shifted 1/2 or 1/3 of the pitch to one another, the same for the stepping distance. It shows in the usable dynamic range that Mark did reveal in some of the tests with the V850.  Next to a true reducing of grain aliasing, in my case V700 scans compared to Nikon 8000 scans, both wet mounted. I use Neat Image to get rid of say half the "grain" left after further development of the scan. The last step could be downsampling but I am still not sure which path to take then. I tried different downsampling routines and output resolutions to see whether that would hit an optimal to further decrease the "grain¨ without losing further detail. My gut feelings is that Bilinear is best and estimating the grain size to pixel size a clue on the ratio of downsampling. On the other hand, printing the image to the stochastic dot formations of the inkjets may prove a better way of downsampling for IQ if file size does not matter.

Met vriendelijke groet, Ernst

http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
December 2014 update, 700+ inkjet media white spectral plots

[ErikKaffehr]

Hi,

Oversampling is always a good thing…

I have shot a test target on ADOX 120 CMS. What I could see was that it resolved around 150-180 lp/mm with my Zeiss Planar 80/2.8 lens around f/8 near the optical axis. Sampling that with 4800 PPI on my Minolta Dimage Scan Multi Pro  still resolved at Nyquist, albeit with low contrast and some aliasing was seen.

The 150-180 lp/mm figure was seen in microphotography using a Photar lens at f/4 (or so) at something like 6X magnification using a 3.8 micron pixel sensor.

From what I have seen, the ideal scanning resolution for that kind of film may be around 6000 PPI or higher.

Best regards
Erik

[greggan]

I have also read and found the article very interesting.
There are many interesting findings and comparisons that have been done by Mark.
However there is not a single word about the film-holders quality and design that many other users have been upset about.
Since the holders have a very important task (to keep film flat) it is strange not to review them as a part of the scanner review.
About the Epson V850 Pro film holder problem!
Note: Epson have been sending the old V750 holders for free to some buyers while the awaiting a MK2 film holder design that have been
promised to some customers, this is not yet confirmed to me by my Epson contact but..... ;-)

Best regards/Stefan