In 2019 where are we at with drum scanners?

[fotopfw]

In Large Format, that's where we are here, it is obvious that a professional drum scan form a 8x10" film is very much better than what the flatbeds can produce.https://www.flickr.com/photos/139951...posted-public/

[Pere Casals]

In Large Format, that's where we are here, it is obvious that a professional drum scan form a 8x10" film is very much better than what the flatbeds can produce.https://www.flickr.com/photos/139951...posted-public/

This is true, a drum is better, but way less that the sample you point, first the EPSON scan lacks the digital image enhancing work that the drum software has applied, so apply the same before comparing, with only an slight edition, see traffic plate, the quality distance is way less (click in the image to enlarge):



You can try on your own with Ps from the sample you linked, or you just can apply any commercial image enhancing software, and a denoising tool... In the drum scan we don't see the film grain, this is not because the PMT sensor has not seen it, but because digital image enhacement has removed it, also the image enhacement has worked the color and the adaptive contrast very noticeably. This has to be said.

There is a remarkable difference in the detail in the rear pneumatic, but it's not explained what kind of film and if multi-exposure was used with the V750. Also in one case (the V750) we have an antenna over the trunk removed, not in the other, so it's an edited image, and we don't know if it was resampled or edited in 8bits... who edited the image and what he did ?







Then you can also take samples from the collaborative scanner comparison and edit each crop to its best: https://www.largeformatphotography.i...an-comparison/

My results are these (And I'm not a very good Ps user...):





You can also check that on your own.


Then drum scans for 8x10" are usually offered at 2000dpi because at 4000dpi (when offered) usually have prohibitive costs from $300 to $400, if drum scanning at 2000dpi then the EPSON matches resolving power.

Another thing is if we have extreme high densities in velvia, then the drum can recover it if necessary, the EPSON may use multi-exposure, but with noticeable less capability .

[Pali K]

I will do a controlled test between Epson, Eversmart Supreme, Scanmate 5000, and Scanmate 11000 soon and everything will be done in RAW files so we can eliminate the notion of what software did or did not do.

I also have a Eversmart PRO and Tango but neither of these have a true RAW output option but will include them in the test with a disclaimer.

Pere, what DPI settings should I scan Epson files at so I get the most from this setup? I plan to use the same DPI across all scanners so that we can eliminate the doubt about downsampling as well. I will need to stay under 5000 DPI so Scanmate 5000 can be in the mix.

Plan is to scan the following in RAW:

- Resolution Target
- Color Slides (8x10, 5x7, 4x5, 6x7 MF, 35MM)
- Color Negative (4x5, 6x7 MF, 35MM)
- BW Negatives (8x10, 4x5, 6x7 MF, 35MM)

I do not have 5x7 BW or Color Negative so if someone wants to send me theirs, I would be happy to include it. I will start a thread for it so we can collaborate and make sure it's as honest a test we can all make it.

Pali

[chassis]

Thanks for this discussion, it is helpful.

Pali, looking forward to your test results.

[Pere Casals]

Pere, what DPI settings should I scan Epson files at so I get the most from this setup?

Pali, it would be interesting you make that side by side, having the same person making all tests adds consistency, and not many have such a diversity of gear like you.

For the EPSON to perform optimally you have to scan at 6400 dpi and to apply a degree of sharpening. IMHO the comaprisson have to be made after optimizing all scans to its best. You should measure performance for the high res lens used with holders and with the low res one used with the area guide. With holders you also may try to gauge a bit up to see if it has an effect.

The Epson requires the medium is flat, because it lacks a focusing system, so if curled you should keep the negatives several days inside a book to ensure best possible flatness,

Also for your V700 you should install a Silverfast version having the Multi-Exposure feature, it makes a significative difference with slides. Something like this: https://www.ebay.es/itm/SilverFast-S...Q2xW:rk:1:pf:0 , I guess that in that case it is included, but it should be checked.

Also for the resolution target I'd suggest you follow this methodology: https://archivehistory.jeksite.org/c.../appendixc.htm , this is making several different passes and to get the reading for each, having max, min and average of 5 readings is good because (depending on the pass) the bars can have a different degree of aliasing with the pixels.

To me the "collaborative scanner comaprison" https://www.largeformatphotography.i...an-comparison/ is also a good reference, that test offers complete information. But sure your test would be extremly interesting. It also would be nice to find at what enlargement we have a difference, to me for BW 8x10 a drum scan makes some difference over an EPSON if the print is beyond 1.5m, but it would be very interesting to check it from your test.

Also IMHO images have to be edited to its best before comparing, to separate what's the machine performance from the digital image enhancing processing, to me the interesting point is not the auto image enhancing, so the way we have is editing the images to its best potential, because many machines hide a digital processing that cannot be deactivated.

Please tell me by PM were I should send you a negative.

[Pali K]

Pere,

I have SE which does not have multi exposure and the upgrade is $70 USD right now which is not worth it for me. Can we assume it's only needed for slides and at least start the test with negatives? I'll be using adjustable height holders to maximize Epson perform.

[Pere Casals]

Pere,

I have SE which does not have multi exposure and the upgrade is $70 USD right now which is not worth it for me. Can we assume it's only needed for slides and at least start the test with negatives? I'll be using adjustable height holders to maximize Epson perform.

Pali, yes, I find multi-exposure it's useful for high densities. A flatbed has more flare than a drum, but IMHO multi-exposure cannot solve that. ME solves the dynamic range shortcomings of CCDs compared to PMTs, I guess that some high end ccd scanners may make two exposures (each with a different level) before advancing the carriage, having ME embedded in the normal operation, while a EPSON needs to make two passes to obtain the same. If you want I can send you a dense velvian negative after having scanned it with ME, to compare with result from other devices.

[Serge S]

Hi Pere,

Is there a sweet spot that you know of where to place the 4x5 negative on an area of the platen to take advantage of the lenses?

(I usually place it directly on then glass platen in the middle area)

Thanks,

Serge

You wrote:

<<Yes, the EPSON has no optical zoom system. A V850 has not bad two lenses, the one that covers 5.9" is resolving around 3000 dpi in the full scan width, resulting in some effective 2800dpi in Hor axis, the Vert axis resolves less (2300) suposedly because of vibrations (etc) in the carriage driving and belts, ...but with 120 film the V850 only uses 1/3 of the scan width, being a suboptimal result, if EPSON would have mounted a narrower lens to take 4" a better performance for 4x5 and 120 would be there, but only one 120 strip would fit, or two 35mm strips. Also a better Y carriage driving would be required.>>

[Pere Casals]

Hi Pere,

Is there a sweet spot that you know of where to place the 4x5 negative on an area of the platen to take advantage of the lenses?

(I usually place it directly on then glass platen in the middle area)

Thanks,

Serge

You wrote:

<<Yes, the EPSON has no optical zoom system. A V850 has not bad two lenses, the one that covers 5.9" is resolving around 3000 dpi in the full scan width, resulting in some effective 2800dpi in Hor axis, the Vert axis resolves less (2300) suposedly because of vibrations (etc) in the carriage driving and belts, ...but with 120 film the V850 only uses 1/3 of the scan width, being a suboptimal result, if EPSON would have mounted a narrower lens to take 4" a better performance for 4x5 and 120 would be there, but only one 120 strip would fit, or two 35mm strips. Also a better Y carriage driving would be required.>>

If you scan on bed you have a worse situation. For best performance you have to ensure film flatness, just keep curled film inside a book for several days if necessary, and you have to use holders.

If you obtain a worse yield with holders this is because height is not right of because of curled film. With holders you use the highres lens covering 5.9" and focused at holder's height. When you scan on bed (with guide) you use the low ress lens that is focused on bed surface, the low res lens has a some 25% loss in effective dpi, but covers 8x10.

The new V850 ANR holders can ensure flatness and have adjustable height.

we can use a glass slide to see if there is sweet spot, but I guess performance may be quite uniform. Think that many lenses don't have the sweet spot in the center, nor in the far off-axis, but in the middle. Design and assembly may target best correction for the centre with worse corners, for the corners with worse center, or in the middle of those extremes, to have a good average, my guess is that this last one it should be the case.

[fotopfw]

Thanks Pere, for your insight and making it explicit.
I've chosen a realistic approach for me, I hate to work on Gigapixel TIFF image.
No processing at all from me (time is money) after a drum scan, great shadow detail included.