DSLR scanning vs. Epson V800 scanning

[Smorton]

I have been using an Epson V800 for a short while now.

A debate I am having in my head is whether you can acheive comparable results using a camera ranging from a crop sensor to a full frame digital camera (eg. Nikon V850 with good macro) without stitching images together. I could never afford a medium format camera so I am not including that in my consideration.

My interest in the Epson V800 was renewed after I did a few fluid mount scans.

Thanks for any replies.

SM

[Kiwi7475]

What are you scanning? 4x5 ? 8x10?

If you believe that the epson resolves at least 2000 pixels per inch, a 4x5 will yield 80 MP and a 8x10 about 320 MP. Hard to get that with a single shot digital camera.

[Tobias Key]

For me it is about the quality of the optics and dynamic range. If a lower MP image is sharper because of better optics it will win in a head to head comparison. I have also found that my DSLR is much more able to deal with dense negatives. The scans are a lot less noisy and it's much easier to pull out detail. You could also argue that 20+ mp is good enough for most uses and if you do need more you have the option of stitching if you want to. Plus a scanner is only useful for one thing but if you put that money into a good macro lens you have more versatility for your money.

[interneg]

You have to trade nominal pixel resolution against MTF/ Sharpness performance. Whatever the actual pixel resolution limit of the Epson is (it's effectively 2x 1200ppi sensors, 1/2 pixel offset from each other & interpolated together), its MTF falls under the critical 50% response needed for good sharpness around 660ppi. An average DSLR sensor might be over 1000ppi at the equivalent MTF50 - so while you might only be getting 20-40mp (realistically all that is necessary for most people who aren't needing to do mural scale work) from a camera sensor used as a single shot scan, the actual sharpness & quality of image captured by the pixels on the camera sensor are likely to be considerably higher than what are consumer grade scanners (despite the 'pro' claim). You are also less dependent on software obsolescence & can adjust the precision of the setup to a greater degree than in a scanner largely made of injection moulded plastic & stamped metal parts.

[Kiwi7475]

You have to trade nominal pixel resolution against MTF/ Sharpness performance. Whatever the actual pixel resolution limit of the Epson is (it's effectively 2x 1200ppi sensors, 1/2 pixel offset from each other & interpolated together), its MTF falls under the critical 50% response needed for good sharpness around 660ppi. An average DSLR sensor might be over 1000ppi at the equivalent MTF50 - so while you might only be getting 20-40mp (realistically all that is necessary for most people who aren't needing to do mural scale work) from a camera sensor used as a single shot scan, the actual sharpness & quality of image captured by the pixels on the camera sensor are likely to be considerably higher than what are consumer grade scanners (despite the 'pro' claim). You are also less dependent on software obsolescence & can adjust the precision of the setup to a greater degree than in a scanner largely made of injection moulded plastic & stamped metal parts.

Is there a reference for your quoted MTF50 of 660 dpi? I’d like to read it.

[interneg]

Is there a reference for your quoted MTF50 of 660 dpi? I’d like to read it.

It can be found here. It also provides a (teardown based) considerable level of detail about the construction of the Epson V7/8xx sensor systems.

[Kiwi7475]

It can be found here. It also provides a (teardown based) considerable level of detail about the construction of the Epson V7/8xx sensor systems.

Thank you!

[interneg]

Thank you!

It certainly made clear a lot of the optical phenomena I'd encountered over the years from Epson scans - much more so than the (frankly rather useless) resolution tests that proliferate across the internet. For my own amusement about a year or so ago, I did actually see if I could recreate the 'look' of an Epson scan by combining 2 slightly offset 1200ppi scans from a high end CCD scanner & upsampling them to 2400ppi - and the resultant file had the rather distinctive characteristics of an Epson scan result - the resolution looked slightly higher than 1200ppi (maybe heading towards 1500-1600ppi in terms of useful resolution - even if nominally 2000+), but a native resolution scan done at 1600ppi on the aforementioned CCD scanner blew away the blended 2x 1200ppi in terms of the important noise:sharpness ratio. The relationships of sharpness, noise, latitude (bit depth too) and colour handling really matter a lot more to a good scan result than any high contrast resolution test.

[grat]

For 35mm, DSLR digitizing will probably be very close in useful resolution to the Epson. For any other format, you'll have to use stitching to beat the 2400-2800 PPI resolution most people are able to extract with the Epson.

interneg apparently was apparently bullied by an Epson in his youth-- the results of the somewhat questionable paper he references are flatly contradicted by real-world experience, and in spite of his claim of "(teardown based) considerable level of detail", provides one measurement, and a whole lotta assuming. Further, the paper was written about 5 years before the introduction of the v800, which is known to have better (even if only slightly) scanning resolution than the v700.

[Peter De Smidt]

With a camera scanner, you can also optimize the scan in a number of ways that would be a challenge with a consumer scanner. You can change the exposure time, for instance, easily and quickly.