Drum Scan vs DSLR vs Epson Scan (From Nick Carver's Youtube Channel)

[sperdynamite]

https://youtu.be/Q9d8BukUgzI

I think you'll see some pretty surprising conclusions here, but they track with my experience.

Personally I would quibble with his DSLR scan technique. 1. He appears to be doing no masking at all. 2. He should be using a copy stand and macro rails with the film in a proper film holder elevated off the light source. 3. The 6D has the worst in class sensor of all semi-modern DSLRs. 3a. The 6D cannot pixel shift so he's fighting bayer interpolation and he has to do an extremely large amount of captures and stitching to achieve his resolution target. And 4. Though he seems to have an understanding of Negative Lab Pro, he doesn't seem to have a strong grasp of how to use it.

All that being said, I'm just not surprised to hear that a DSLR is so close to the drum scan in IQ. You really do have to try it to see for yourself. I'm also not surprised to see that he prefers the Epson scan. Epsons do well with larger formats, and he's both wet scanning it, and taking advantage of iSRD. Dust is a huge problem with ultra high res scanning, and one I'm still trying to deal with. (Particularly with 8x10!)

Enjoy the video!

[Peter De Smidt]

These are what I use to combat dust: https://www.amazon.com/Record-Cleaning-Cleaner-Reusable-Antistatic/dp/B07KVDK5Y7/ref=sr_1_3?keywords=awpeye+record+cleaner&qid=1580226479&sr=8-3

I roll both sides of the scanner platen, the light table, and the film. It really helps.

[Tin Can]

Good video, thanks!

[Peter De Smidt]

His dslr scanning method...was sub-optimal...in many ways. His camera was a poor choice, the lens was worse, the negative should've been carefully masked, there's no way his alignment was accurate enough for the best results, and his light source quality was questionable.

[Tin Can]

Correct on all points

The video was well done...

[sperdynamite]

These are what I use to combat dust: https://www.amazon.com/Record-Cleaning-Cleaner-Reusable-Antistatic/dp/B07KVDK5Y7/ref=sr_1_3?keywords=awpeye+record+cleaner&qid=1580226479&sr=8-3

I roll both sides of the scanner platen, the light table, and the film. It really helps.

Great tip thank you!

[brad martin]

Ha! Just watched it tonight.

Recommended.

[Peter De Smidt]

If you want to stitch with a digital camera, other than, say, a simple 3-frame stitch, make a dedicated movement system like Scanduino. Just like an Epson, you mount the film and press the button. Regarding stitching, with a precision movement system you can simply use a template in Ptgui. He says, "I tried everything." Clearly, he didn't. It's odd to me that he would go to all the work involved with the video, but he failed to do due diligence.

[Sasquatchian]

He got one thing right. Aztek makes shitty scans even though they've got great hardware. Just about everything else about this video reminds me of why people who don't really know what they're doing shouldn't make instructional or informational videos. This was painful on so many levels.

And without knowing the specifics of the drum scans he farmed out to Strickland, it's very hard to know if even those were any good. There are drum scanners and then there are drum scanners. And then there are the operators. And keeping the E6 scan from Aztek as a good example when their neg scan sucked so bad had me shaking my head. What inkjet print was he printing on? If it's Epson, then 360 or 720 dpi, if you've got it, will make very slightly sharper prints if the paper actually supports it. And printing slices is NOT the same as looking at an entire print. You lose the context. Printing full sized prints simply is not that expensive when they're fairly modest like his are. And, having done dozens and dozens if not hundreds of flat stitches using PhotoMerge in Ps, I've never ever had the issues he had. Well, once with 25 100mp Phase files where there was a lot of parallax going on, but never on a flat stitch. I simply don't buy that he tried every single option, or if he did, he clearly didn't understand how the tool works. I actually felt embarrassed for him from one photographer to another.

[sperdynamite]

He got one thing right. Aztek makes shitty scans even though they've got great hardware. Just about everything else about this video reminds me of why people who don't really know what they're doing shouldn't make instructional or informational videos. This was painful on so many levels.

And without knowing the specifics of the drum scans he farmed out to Strickland, it's very hard to know if even those were any good. There are drum scanners and then there are drum scanners. And then there are the operators. And keeping the E6 scan from Aztek as a good example when their neg scan sucked so bad had me shaking my head. What inkjet print was he printing on? If it's Epson, then 360 or 720 dpi, if you've got it, will make very slightly sharper prints if the paper actually supports it. And printing slices is NOT the same as looking at an entire print. You lose the context. Printing full sized prints simply is not that expensive when they're fairly modest like his are. And, having done dozens and dozens if not hundreds of flat stitches using PhotoMerge in Ps, I've never ever had the issues he had. Well, once with 25 100mp Phase files where there was a lot of parallax going on, but never on a flat stitch. I simply don't buy that he tried every single option, or if he did, he clearly didn't understand how the tool works. I actually felt embarrassed for him from one photographer to another.

Micheal Strickland does good work, he uses a Tango. I actually reached out to Nick Carver to bring up some of the issues I had with his camera-scanning technique. He was basically non receptive to the idea that he did anything wrong or anything could be improved. So...okayyyy. I do like some of his videos, even when I'm not crazy about his work overall. It's very 'commercial'.

[Pere Casals]

I'm also not surprised to see that he prefers the Epson scan. Epsons do well with larger formats, and he's both wet scanning it, and taking advantage of iSRD. Dust is a huge problem with ultra high res scanning, and one I'm still trying to deal with. (Particularly with 8x10!)



Not a surprise, very, very skilled people having a drum and an Epson prefers the Epson 95% of the times for MF and up. Probably 5% of the times the drum is needed. For LF it's really difficult to substantially beat the EPSON.

The Epson has been attacked by people having commercial interests because the Epson was a very hard competition, and many people prone to hype had been lured to buy expensive gear when just an Epson scan can match the result with a few mouse clicks.


The Epson requires some edition skills to digitally optimize the image in the way Pro machines do automaticly, so people lacking basics on Ps edition skills were easily misslead, some buying expesive/big/problematic pre-press gear to have absolutely no advantage for LF.



Every day a fool is born, it's just a matter of finding him to make a good business. (Citing Juan March, https://en.wikipedia.org/wiki/Juan_March_Ordinas)

[Alan Klein]

Not a surprise, very, very skilled people having a drum and an Epson prefers the Epson 95% of the times for MF and up. Probably 5% of the times the drum is needed. For LF it's really difficult to substantially beat the EPSON.

The Epson has been attacked by people having commercial interests because the Epson was a very hard competition, and many people prone to hype had been lured to buy expensive gear when just an Epson scan can match the result with a few mouse clicks.


The Epson requires some edition skills to digitally optimize the image in the way Pro machines do automaticly, so people lacking basics on Ps edition skills were easily misslead, some buying expesive/big/problematic pre-press gear to have absolutely no advantage for LF.



Every day a fool is born, it's just a matter of finding him to make a good business. (Citing Juan March, https://en.wikipedia.org/wiki/Juan_March_Ordinas)

What edits in PS for the Epson are you referring too?

[Peter De Smidt]

"Not a surprise, very, very skilled people having a drum and an Epson prefers the Epson 95% of the times for MF and up." Kinda hard to support that, isn't it? And if you look at this site, one where many people have drum scanners, including those who don't scan for money, you'll not find people agreeing with Pere. I've been here a long time. I pay attention to scanner threads, and I don't own a drum scanner or do scans for money. Made-up statistics are not a good sign of reliability. Epistemology, by the way, is my area of professional expertise.

[Jim Andrada]

I don't even have a drum scanner - just a Lowly IQsmart 2. And I don't prefer the Epson 750/850. I have one, but haven't seen a need to use it since I got the Creo/Kodak. Which is what I actually prefer because It's A Better Scanner.

None of the folks that bought prints I made with the Epson were unhappy with them. But I slowly started to feel that I could do better. I don't so much feel that way anymore. Of course I can always do better, but the emphasis is on the "I" not the tool I'm using.

[Pere Casals]

What edits in PS for the Epson are you referring too?

Hello Alan,

First, let me say that everytime a serious side by side has been made the Epson has shown impressive excellence in LF.

See this, the Epson equals a "modern" Creo and and a 11000 Drum, while surpassing an older Creo:

https://www.largeformatphotography.info/forum/showthread.php?150020-Scanner-Comparison-2019-Epson-Flatbed-Eversmart-Flatbed-Drum-Scanners&p=1479178&viewfull=1#post1479178

It only required 2.5pix 50% sharpening radius and a little touch in the curves shaping a bit in S to totally match the drum sample.


The challenge you will face with the V850 is Sharpening and "resolution management". Sharpening is always complex. Also you have to save a Hires image (4800 or 6400) in TIFF, 16 bits/channel and taking all histogram to get a dull image then sharpen, reduce image size with "bicubic, ideal for reductions" choice in the image size dialog, bend curves to fit the monitor/paper dynamic range, edit, save a Hires file copy, resize to the distribution size, make a final pixel level sharpening (also bicubic for reductions), then convert to 8bit per channel, and save that file in TIFF and in jpeg.

That workflow sequence is important... you may vary something but really not much. Pro scanners may do internally well digital sharpening and resolution managent, but with the Epson you have to edit like Professionals do make a perfect job.


If you search "Epson" and "Crap" in this forum you will find many results, those are absolute lies. If you know how to edit an image Professionally then you'll get absolutely Professional results with the Epson.



Do something, if you want... order a Pro drum scan of challenging negative, then try to match it with the Epson, if you have any missmatch then PM with dropbox and I'll guide you to nail a perfect match.


99% of complains about the EPSON for LF are in fact shorcommings in the edition skills or pure lies from people that had dirty commercial interests, just this: if for LF you can't match a drum job then just PM and I'll show you how to do it, I won't fail.

____

Disclaimer, it is true that for LF sometimes (5%) a drum ca do a better job. For example for a mural print from 4x5 where grain depiction is to be important a 8000dpi drum scan will show a way better grain structure, but not much difference will be there compared with a 4000dpi drum scan.

For very underexopsed Velvia use Multi-exposure in silverfast. For crazy underexposed Velvia a drum may be required.

[Louis Pacilla]

Hello Alan,

First, let me say that everytime a serious side by side has been made the Epson has shown impressive excellence in LF.

See this, the Epson equals a "modern" Creo and and a 11000 Drum, while surpassing an older Creo:

https://www.largeformatphotography.info/forum/showthread.php?150020-Scanner-Comparison-2019-Epson-Flatbed-Eversmart-Flatbed-Drum-Scanners&p=1479178&viewfull=1#post1479178

It only required 2.5pix 50% sharpening radius and a little touch in the curves shaping a bit in S to totally match the drum sample.


The challenge you will face with the V850 is Sharpening and "resolution management". Sharpening is always complex. Also you have to save a Hires image (4800 or 6400) in TIFF, 16 bits/channel and taking all histogram to get a dull image then sharpen, reduce image size with "bicubic, ideal for reductions" choice in the image size dialog, bend curves to fit the monitor/paper dynamic range, edit, save a Hires file copy, resize to the distribution size, make a final pixel level sharpening (also bicubic for reductions), then convert to 8bit per channel, and save that file in TIFF and in jpeg.

That workflow sequence is important... you may vary something but really not much. Pro scanners may do internally well digital sharpening and resolution managent, but with the Epson you have to edit like Professionals do make a perfect job.


If you search "Epson" and "Crap" in this forum you will find many results, those are absolute lies. If you know how to edit an image Professionally then you'll get absolutely Professional results with the Epson.



Do something, if you want... order a Pro drum scan of challenging negative, then try to match it with the Epson, if you have any missmatch then PM with dropbox and I'll guide you to nail a perfect match.


99% of complains about the EPSON for LF are in fact shorcommings in the edition skills or pure lies from people that had dirty commercial interests, just this: if for LF you can't match a drum job then just PM and I'll show you how to do it, I won't fail.

____

Disclaimer, it is true that for LF sometimes (5%) a drum ca do a better job. For example for a mural print from 4x5 where grain depiction is to be important a 8000dpi drum scan will show a way better grain structure, but not much difference will be there compared with a 4000dpi drum scan.

For very underexopsed Velvia use Multi-exposure in silverfast. For crazy underexposed Velvia a drum may be required.

Man oh man here we go AGAIN! I didn't miss your absence.

[Corran]

Man oh man here we go AGAIN! I didn't miss your absence.

+1

As has been stated a zillion times by many members, any sharpening done on an Epson scan can be done on a better scan too, and therefore is a poor argument. And there is a difference between resolution and oversharpened noise.

[Kiwi7475]

Why is this topic so controversial and brings so much animosity since it was first brought up years ago?

[Louis Pacilla]

Why is this topic so controversial and brings so much animosity since it was first brought up years ago?

It's people (mostly one) trying against ALL evidence to convince all others that the Epson flatbed scanner 700-750-800-850 match or OUTPERFORM a quality drum scan. Simple BS and most of the BS comes from this one member. Mentioned above.

Now whether or not the Epson flat bed is not good enough for most of us is a personal thing. Trying to convince others the Epson V700/800 is as good or better then a quality drum scan is some serious hubris.

[sperdynamite]

Man oh man here we go AGAIN! I didn't miss your absence.


This is how you do it. :cool:

200124

[Louis Pacilla]

This is how you do it. :cool:

200124

He has been on my list for a good while but I must admit I can't help myself at times & I check the post out. Probably should know better but I keep holding out hope.:rolleyes:

[Sal Santamaura]

He has been on my list for a good while but I must admit I can't help myself at times & I check the post out...I don't peek, but then there are people who quote his entire posts and make it impossible not to see them.

By the way, Pere is now also polluting PHOTRIO as member "138S."


...Probably should know better but I keep holding out hope.:rolleyes:To once again quote my favorite observed bumper sticker, "You'll feel much better when you give up hope." :)

[sperdynamite]

I don't peek, but then there are people who quote his entire posts and make it impossible not to see them.

By the way, Pere is now also polluting PHOTRIO as member "138S."

To once again quote my favorite observed bumper sticker, "You'll feel much better when you give up hope." :)

I engaged with 138S once and then quickly realized what was happening. Never again. Some people just want to watch the world burn... :D

[Pere Casals]

Hi guys, you know well that I can deal with all you with both hands tied :)

But don't worry, I won't participate much in this forum from know because I'm busy with CAST and ATL retrofit, just saying that from time to time I may bring evidence about the truth.

Just remember that when Pali made that side by side scans were made by someone knowing what's an scanner, and in that situation results are equal, of course some are not able to get perfect LF results from an Epson, me I'm able, anyone needing to know how to do it can count with all help I can give. Is for this that I may post in the future, helping those wanting to learn how to use the Epson optimally.

Regards, see you.

[Jim Andrada]

Hi Pere.

I don't think anyone says the Epson isn't a nice scanner - it is and for the price it's hard to beat. If it had autofocus it would be a better scanner. And we can all do good work with the Epson. That doesn't mean that if one has the space and the $$$ that there isn't a better tool for the job. I've used the Epson for years. I've had the IQsmart for a few years now and I think it's a better tool. Drum scanning might be even better (but I found that the workflow wasn't for me.)

[sperdynamite]

Hi Pere.

I don't think anyone says the Epson isn't a nice scanner - it is and for the price it's hard to beat. If it had autofocus it would be a better scanner. And we can all do good work with the Epson. That doesn't mean that if one has the space and the $$$ that there isn't a better tool for the job. I've used the Epson for years. I've had the IQsmart for a few years now and I think it's a better tool. Drum scanning might be even better (but I found that the workflow wasn't for me.)

If the Epson had ICE on the scanning bed for 8x10, and if it had AF, I would definitely be using one, even over my S1R workflow. How they could go from the V700/50 to the 800 series and not implement at least AF is beyond me.

[Alan Klein]

Well, having moved up to 4x5, I don;t really want to stitch with my V600. So I have to make a decision, probably go with the v800 or v850.

[Pere Casals]

If it had autofocus it would be a better scanner.

Hello Jim

V800/850 holders are better than having Autofocus, focus is ensured with an ANR glass: perfect focus without cost/complexity of an AF system.

Also new holders have variable height and are V700/750 compatible.

Problem with ANR is that we have an additional chance to get dust, so an HEPA purifier can be recommended even more.



I've had the IQsmart for a few years now and I think it's a better tool. Drum scanning might be even better (but I found that the workflow wasn't for me.)


Drums and the Creo are way better scanners than the Epson, but this is only well noticed with 35mm film, For MF and up this is way less noticed.

By 5x7" (with custom holders) the Epson atonishingly good, at 8x10 the Epson is not as efficient than with 5x7 because it uses the LoRes lens, but still it takes 300Mpix effective from medium because format is large.


Principal drawback of the Epson is that it requires good skills from operator, but if it's the case then impressive results can be obtained for LF, less impressive for 35mm. But this is a LF forum, IIRC.

[Sasquatchian]

Dude - a real drum scanner requires even better operator skills that a flatbed and far too many purporting expertise in the area simply have little or none. The skill it takes to operate an Epson is peanuts compared to what it takes to fully understand the ins and outs of a drum scanner, both hardware and software. Critical focus, actual resolved detail, real dynamic range and most importantly to those of us printing, smoothness of gradations especially in areas like clear skies are all areas a good drum scan far outshines an Epson, and even far more expensive flatbed scanner for that matter. But again, if you have a shitty operator who thinks he's good or you have hardware that is not optimally maintained or the optical system is misaligned or dirty, the the playing field is no level. Really, I can't tell you how many times I've had people come to me and say that their scans didn't seem right. I always tell them that I don't know for sure if my drum scanner will be any better but I'm more than happy to give it a shot and see what happens. In every single case that stupid little desktop HR8000 has kicked ass over ever competing scanner I've come across, including Creo's, Hell's, Imacon's and even the beasts known as Crossfield. And let me tell you that 8000 ppi on a Crossfield drum doesn't begin to compare to the same on a properly aligned Howtek. While it's true that at smaller sizes you won't see as much as a difference, you may indeed if you know what you're looking for. As Robert Earl Keen wrote so many year ago -- it's the little things...that piss me off...

[Pere Casals]

Dude - a real drum scanner requires even better operator skills that a flatbed and far too many purporting expertise in the area simply have little or none. The skill it takes to operate an Epson is peanuts compared to what it takes to fully understand...


Sorry, I was not clear. I totally agree with you.

No doubt that operating a drum properly requires a true professional that probably also has amazing edition skills and a refined aesthetic criterion.

What I wanted to say is that a Pro machine usually delivers a welll digitally optimized image, while the Prosumer Epson allways requires some edition skills to optimize images manually, a discipline that not all home users do master well.

I you review the Pali test you'll find that the epson matches the scanmate 11k drum for Portra MF, this is because job was well done.

[Sasquatchian]

When I was first exposed to drum scanning back in the mid 90's, one place where we had scans done in West L.A. had a Hell 3010 scanner. Yeah, on that, it was far more complicated. You never ever saw an image on the screen. The monitor was only to set all the scanning parameters but you had to use this little scope with crosshairs in it to pinpoint and set the ink percentages for white black and gray - and it was all in CMYK as that scanner could only deliver CMYK. So you'd scope the white point and set that to 5C, 3M and 3Y, set the mid tone to 50C, 40M, 40Y and set the black accordingly. You had to KNOW this shit upside down and inside out. It was more expensive to fix it after than to get it right in the scanner. I used to sit with the scanner operator at midnight over on Pico Blvd and tell them the numbers I wanted in my scans. There was no color management then. The scanner was calibrated directly to their proofing system and what you got was what you got from them.

Then as scanning for magazines and ads started to wane and fine art scanning started to take over, it was a much different experience. Color Managed graphic interfaces were introduced and a single click could get your scan pretty much 98 percent there, and a lot of the mystery of driving a drum scanner was gone. But while drum scanning software still has some complicated and esoteric features, all the packages out there today are really quite fast and easy and usually support IT8, Wolf Faust and Hutchcolor targets, making fast accurate scans simple. It's really pretty much the same in concept in drum scanner or flatbeds and they both do the basic stuff automatically - black point, white point and gray - if you want it to.

For color transparencies scanning it couldn't be easier than scanning a Hutchcolor target and saving the scanning setting used to scan the target and calling that up when you scan. For black and white I always scan flat to make sure no clipping at all happens. That makes for more work in Photoshop later but it's always worth it. Color negs are the one area where it really does take personal sensibility as there's no visual reference and having used at least half a dozen different scanning applications for color negs, it has always come down to one software package that has been just better at it than any other and that has been the long abandoned Trident 4.0 for Howtek. That's probably where the biggest learning curve is and what takes the most operator skill to get right.

To me, you sound like someone who has never actually operated a drum scanner or become familiar with the software and work like hell to convince yourself that your Epson is just as good. I started off with flatbeds then to fluid mounting on flatbeds then to drum scanning, but it was having the digital correction and toning skills from the very beginning that made the transition to drum scanning fairly easy. At least it seems easy in retrospect twenty-two years later.

[interneg]

Color negs are the one area where it really does take personal sensibility as there's no visual reference and having used at least half a dozen different scanning applications for color negs, it has always come down to one software package that has been just better at it than any other and that has been the long abandoned Trident 4.0 for Howtek.

What I've found works most effectively is not allowing the scanner to do any inversion whatsoever, getting the image into Photoshop, sampling & dividing out the mask, then inverting, then setting black & white points etc. Most scanner software and plugins seem to have real problems understanding how to deal with the mask - and I've often found that the impact of the onboard scanner software carrying out the inversion can be sufficient to reduce the performance of many high end scanners quite noticeably.

[Sasquatchian]

What I've found works most effectively is not allowing the scanner to do any inversion whatsoever, getting the image into Photoshop, sampling & dividing out the mask, then inverting, then setting black & white points etc. Most scanner software and plugins seem to have real problems understanding how to deal with the mask - and I've often found that the impact of the onboard scanner software carrying out the inversion can be sufficient to reduce the performance of many high end scanners quite noticeably.

While that is certainly true of many software packages, and I haven't used them all, but whatever secret sauce John Panazzo put into Trident is simply the best inversion I've ever seen. And once you really learn what it can do and how to affect color changes by simply dragging the white and black point probes to a different place in the image, you realize it's on a different level. There's a real learning curve to it and it took several months of scanning trying all the different options but once you figure it out it's fast and easy. The problem is that it only works for Howtek scanners. Well, there was a version for Colorgetters as well at one time but I don't think that version was ever fully updated. And just for experimenting and the sake of argument, I've shot 5DSR images of color negs on a Just Normlicht light box and converted them both in Photoshop directly with mixed results and also done the full inversion in Capture One, which was surprisingly good, but again, only if you're accomplished in that program. Probably the worst of the expensive packages in my personal experience was using Linocolor. Very complicated and very lackluster color neg scanning. Probably the reason most Hell scanner operators just flat out refused to scan color negs.

[Peter De Smidt]

https://www.dpreview.com/news/7648446596/fujifilm-says-new-400mp-pixel-shift-mode-is-coming-to-its-gfx-100-camera-system

[Tin Can]

And the beat goes on

[Pere Casals]

No available lens for the GFX will project 400 differentiated megapixels on a 43.8 mm x 32.9 tinny sensor like the GFX has.

Let's say we use a lens with same quality than a top notch Rodenstock Digaron HR Macro, $6400 at B&H. This MF glass yields "up to" 100 lp/mm (probably less in practice), which limits effective resolving power to 43.8*32.9*200*200 = 57MPix effective, no amount of shifts will enhance image beyong native optical quality.

Add $10,000 for the GFX... for an around $18,000 setup if counting the stand and illumination which is at the same level than the rest of the setup, and still we have to remove scratches and dust manually for an imperfect job, when the cheapo V800 will remove all dust automaticly through infrared detection (specially in color film) saving a lot of manpower and it will deliver 150MPix effective from a 4x5, beyond 200 from a 5x7 and beyond 300 from a 8x10.


If we are to stitch... then resolving power has little limitations, but in that case if we stitch a bit more crops we do the same with any cheap dslr, say an ancient D810.


My mobile is said to make 100Pix images from pixel shift, the USAF 1951 target downrates yield to 10MPix effective in my personal tests.


With present day pixel shift we potentially can exploit most of performance the lens has, if we also are able (!) to ensure perfect focus at optimal aperture, perfect alignment and total shake absence, but the lens itself is a limiting factor anyway.


(The Digaron Macro has a ring adjusting position of an internal floating group for optimal work at the magnificaton we work... we need a lens working specifically well at our scale, or something like an adjustable Macro Digaron, if not that 18k setup is a pitfall)

[sperdynamite]

https://www.dpreview.com/news/7648446596/fujifilm-says-new-400mp-pixel-shift-mode-is-coming-to-its-gfx-100-camera-system

I saw that! Very cool stuff. However medium format will still have disadvantages vs 35mm full frame. The native macros tend to go to 1:2, and DoF is much less. You could solve this with an extension tube and IIRC the Fuji system will still allow full AF, which is great. But then you're at 120mm on a 44x33 sensor, which means it'll be hard to have a lens that can do 1:1 on all formats from 35 to 8x10 on a single copy stand. I'm certain that sheet film will benefit from the resolution and tonal gains, but 120 and 35, you're already at quite a large enlargement from a single pixel shifted A7RV...(IV?) or S1R scan. Then there is the cost of course.

Who knows what the S2R or S1RII or whatever they choose to call it will be able to achieve? Of course if you happen to HAVE a GFX100, you should definitely at least try it!

[Peter De Smidt]

... I'm certain that sheet film will benefit from the resolution and tonal gains, but 120 and 35, you're already at quite a large enlargement from a single pixel shifted A7RV...(IV?) or S1R scan. ...

Agreed!

[Kiwi7475]

No available lens for the GFX will project 400 differentiated megapixels on a 43.8 mm x 32.9 tinny sensor like the GFX has.

Let's say we use a lens with same quality than a top notch Rodenstock Digaron HR Macro, $6400 at B&H. This MF glass yields "up to" 100 lp/mm (probably less in practice), which limits effective resolving power to 43.8*32.9*200*200 = 57MPix effective, no amount of shifts will enhance image beyong native optical quality.

Add $10,000 for the GFX... for an around $18,000 setup if counting the stand and illumination which is at the same level than the rest of the setup, and still we have to remove scratches and dust manually for an imperfect job, when the cheapo V800 will remove all dust automaticly through infrared detection (specially in color film) saving a lot of manpower and it will deliver 150MPix effective from a 4x5, beyond 200 from a 5x7 and beyond 300 from a 8x10.


If we are to stitch... then resolving power has little limitations, but in that case if we stitch a bit more crops we do the same with any cheap dslr, say an ancient D810.


My mobile is said to make 100Pix images from pixel shift, the USAF 1951 target downrates yield to 10MPix effective in my personal tests.


With present day pixel shift we potentially can exploit most of performance the lens has, if we also are able (!) to ensure perfect focus at optimal aperture, perfect alignment and total shake absence, but the lens itself is a limiting factor anyway.


(The Digaron Macro has a ring adjusting position of an internal floating group for optimal work at the magnificaton we work... we need a lens working specifically well at our scale, or something like an adjustable Macro Digaron, if not that 18k setup is a pitfall)

I don’t think it changes your conclusion but the pixel shifting of 4 means the lens would “only” have to be capable of resolving 100 Mpixels, not 400.

[Pere Casals]

I don’t think it changes your conclusion but the pixel shifting of 4 means the lens would “only” have to be capable of resolving 100 Mpixels, not 400.
Of course...

But more than 4 shifts can be made...

The way the camera makes the shifts may be undisclosed, for example shifts can be sub pixel matching sensitive area size inside the pixel.

[Tin Can]

I think Pixel shift is a expansion of shake reduction mechanisms and software

I am waiting for NIKON as I have zero need for a zillion pixels

[Leszek Vogt]

I think Peter is correct (#8)....Nick lacks diligence. If the camera results were inadequate, one can do various things to overcome that. If Canon color was "off" (?) then why not try Pentax or Nikon....with crappy Tamron 90/2.8 macro that delivers....or even my silly Nikkor 55/3.5. Strange, my idiotically old and cheap Elements can stitch superbly...so I don't quite get it why PS failed him?
Hmmm, if one wants good results....patience is essential.

No idea why he went through this exersize....his images don't offer any emotional impact that I was looking for. Subjective, indeed.

Les