New to digitally copying film and just two months into shooting film, I’ve been reading Kodak Technical Specifications for Portra 160, 800, Ektar and E100, but what really helped the most in developing final digital resolution goals was Grok 4 going step by step through the interpretation of publish lp/mm to an estimate of MPix digital representation of that film. Grok puts 645 Ektar and Portra around 90 MPix, E100 a bit higher and Portra 800 around 60 MPix. Now, I don’t for a minute believe that such ideals can be attained with most subjects, lighting and so on, but it does provide a reasonable goal for downsized pixelshift negative copy files.
From what I’ve seen on various boards, most don’t try to obtain a high degree of fidelity in the analog to digital copies and it seems that file size is a major obstacle, especially with medium format copies. I’m tethering with Lightroom Classic and storing the final files on a NAS. While it is true that file operations bandwidth isn’t the same as when I use SSD, it is relatively fast and more intensive intermediate work, such as combining 16 shot pixelshift images, negative to positive conversion and downsizing can be done on SSD and these files deleted after the final files are stored on the NAS.
You undoubtedly have more experience with regard to the above and I would be happy to learn where I’m misguided and how I might better my workflow.
You have no idea how difficult it was in 2002 to scan 6x9 to 120MP 16-bit files at 4000DPI with 16x sampling. That was pain and suffering!
I recall compressing the files losslessly after that and leaving them for a couple of years until I had a better computer with more RAM. However, I did the right the right thing by getting the best quality I could at the time. You don't want to go back or regret it later unless you really don't care about the quality.
Today working with individual large files is nothing, though of course huge volumes (100s of thousands) may be an issue for bulk storage.
I'm not clear what that Grok scanner is, or are you digitizing with a camera and macro lens?
If using an actual film scanner and for image processing like Abode, of course it is ideal to work from internal SSDs. I have a WIP SSD space of 20TB and then another 20TB internal for other purposes. I keep the local files also copied to at least two NAS.
EB-1 wrote:
You have no idea how difficult it was in 2002 to scan 6x9 to 120MP 16-bit files at 4000DPI with 16x sampling. That was pain and suffering!
I recall compressing the files losslessly after that and leaving them for a couple of years until I had a better computer with more RAM. However, I did the right the right thing by getting the best quality I could at the time. You don't want to go back or regret it later unless you really don't care about the quality.
Today working with individual large files is nothing, though of course huge volumes (100s of thousands) may be an issue for bulk storage.
I'm not clear what that Grok scanner is, or are you digitizing with a camera and macro lens?
If using an actual film scanner and for image processing like Abode, of course it is ideal to work from internal SSDs. I have a WIP SSD space of 20TB and then another 20TB internal for other purposes. I keep the local files also copied to at least two NAS.
Ah, sorry for the confusion. I use an A7CR tethered to Lightroom and using a Tamron 90mm Macro lens on a Negative Supply light/mask setup and a Kaiser copy stand. Grok 4 is an AI agent that I used to try to understand how the resolution, granularity and MTF numbers/graphs might translate into a digital resolution that I might use to maximize information retrieved from the film.
My V850 does not give any more resolution when setting the scan resolution above 3200 dpi, and even at that setting the actual resolution is not 3200 dpi. That is fine, as I made a decision to not chase resolution when shooting with film, so I am doing all of my scans at 3200 dpi. For 6x6 to 6x12 frames, that gets me a ~250-500 MB file.
In terms of storage. In short, I cull. I only scan frames that look like they could be keepers. I do not bother scanning every frame. Scans (in tiff format) stay on a massive external drive, only the edited keepers (in png format) reach my local drive.
theHUN wrote:
My V850 does not give any more resolution when setting the scan resolution above 3200 dpi, and even at that setting the actual resolution is not 3200 dpi. That is fine, as I made a decision to not chase resolution when shooting with film, so I am doing all of my scans at 3200 dpi. For 6x6 to 6x12 frames, that gets me a ~250-500 MB file.
In terms of storage. In short, I cull. I only scan frames that look like they could be keepers. I do not bother scanning every frame. Scans (in tiff format) stay on a massive external drive, only the edited keepers (in png format) reach my local drive....Show more →
Same. My V700 is more of a contact sheet creator - I can decide later if I need more than 25mp or so for a 6x6 scan. Maybe I can do better with my camera, more maybe it makes more sense to get it drum scanned.
I think you focus too much on some theoretical resolution maximum that’s based on contrast ratios that don’t exist in real images.
For a real increase in „fidelity“ esp from negative film it seems more effective to focus on the light source and consider true narrow-band RGB lighting.
With pixel shift and fixed white balance the camera should be able to capture a separate true color image for each channel which you could then extract in photoshop and merge again for a true RGB file.
Might as well buy a Fuji frontier or Noritsu at that point though..
corposant wrote:
Same. My V700 is more of a contact sheet creator - I can decide later if I need more than 25mp or so for a 6x6 scan. Maybe I can do better with my camera, more maybe it makes more sense to get it drum scanned.
The thing with a camera is that you lose digital ICE. When I compared my V850 scan to an IQsmart3, the difference was not worth it to me. I would much rather do analog prints, but I lack the room for it, while I am also terrified of even more dust management.
I just stick to digital when I know that resolution matters (which is rare).
theHUN wrote:
The thing with a camera is that you lose digital ICE. When I compared my V850 scan to an IQsmart3, the difference was not worth it to me. I would much rather do analog prints, but I lack the room for it, while I am also terrified of even more dust management.
I just stick to digital when I know that resolution matters (which is rare).
That is true - I will fluid-mount a cam-scan for things that I feel like are really worth it. Other than that, I'll take the dICE of my flatbed.
bwcolor wrote:
Ah, sorry for the confusion. I use an A7CR tethered to Lightroom and using a Tamron 90mm Macro lens on a Negative Supply light/mask setup and a Kaiser copy stand. Grok 4 is an AI agent that I used to try to understand how the resolution, granularity and MTF numbers/graphs might translate into a digital resolution that I might use to maximize information retrieved from the film.
I'm not seeing how you would enough data to train an AI model, but that's not my area of expertise.
You can't have too much info. I'd scan up to 5000 DPI if I had a good scanner. 60MP is not so much for smallish medium format, maybe 50MP with cropping. The problem with copying is getting the film flat, removing blemishes on the film, and the corners may lose IQ. I'd use a scanner if you can find a good one and it matter to you.
EB-1 wrote:
I'm not seeing how you would enough data to train an AI model, but that's not my area of expertise.
You can't have too much info. I'd scan up to 5000 DPI if I had a good scanner. 60MP is not so much for smallish medium format, maybe 50MP with cropping. The problem with copying is getting the film flat, removing blemishes on the film, and the corners may lose IQ. I'd use a scanner if you can find a good one and it matter to you.
EBH
Yes.. good points, but the use of AI light allows me to understand the factors involved. I do care about resolution and that is why I shoot 60-100MP digital cameras, but that doesn’t mean that I don’t appreciate film. I have a Coolscan 9000ED, 5000 and a couple of others, but I re-entered film from the back door, looking to scan images going back to the mid 60s that predate the introduction of the Coolscans. For that, I wanted a workflow that was faster than using a scanner and higher quality than my Epson 700. Some above don’t care about film resolution, and maybe haven’t thought much about color fidelity. For me, I’m intentionally leaving my digital behind in order to get more experience with film, so my first issue was what film. I’ve narrowed that down to Ektar, Portra and E100. B&W is only digital, with my M11 Monochrom. I’ve had little issue with dust. I’ve been sending my film to Citizen’s Photo in Portland and it comes back uncut and sleeved. I remove the sleeve as it enters the film holder and that seems to be sufficient to keep dust away. My copy room has a large, higher end air filter and keeps room dust to a minimum. My interest in film resolution is simply a basis for how much to downsize the combined pixelshift image. Those above that point out the folly of believing theoretical resolution figures produced under perfect lighting with high contrast are correct, but I’m not taking those numbers as absolute, but as relative numbers that allow me to compare different film stocks.
I’m going to link to a 60MP (after downsizing) M3/90mm Super-Elmer/Ektar image which contains much more resolution and fidelity than I expected. I’m guessing that I could have downsized this to between 24 and 30MP without loss of detail.
Do resolution and color "fidelity" matter for film images captured in the 60s? And fidelity to what - to the colors of photographed scene, color reproduction as recorded on film back when the images were captured or to how the film represents them today?
And how do you measure both resolution and especially fidelity? Without a reliable measuring system usage of such terminology is kind of pointless.
What is the intended purpose for the digitized high-resolution film images? Print them all 20x30 inches or larger, frame and hang for a close visual inspection?
SergeyT wrote:
Do resolution and color "fidelity" matter for film images captured in the 60s? And fidelity to what - to the colors of photographed scene, color reproduction as recorded on film back when the images were captured or to how the film represents them today?
And how do you measure both resolution and especially fidelity? Without a reliable measuring system usage of such terminology is kind of pointless.
What is the intended purpose for the digitized high-resolution film images? Print them all 20x30 inches or larger, frame and hang for a close visual inspection?
Good points. Film by design is a film stock lighting dependent rendition of reality. What I have intended was eliminating the Bayer representation, much like an RGB scanner. My copying of old images is based upon first creating contact sheets and individually copying culled images. I do print mostly larger print, 24”x36” and 24”x48” and 12”x12” albums. I’ve yet to get past the contact sheets, so future use of images is in flux. Right now, I’m working on a best effort, from which I can copy with less rigor depending upon final end use.
If they survived 60+ years of storage on the original medium (film) it is highly likely that they will survive for much longer and may even outlive the digital files (think about hardware failure, discontinuation of interfaces and other technology standards, user mistakes).
Every image may require a unique treatment and with that in mind , beside of making contact sheets, I would think what is the best for digitizing and processing of each image at the time when I am ready to edit and print it.
On the other hand, if some of the objectives is to kill time and contribute to the global warming then digitizing everything at the highest resolving power is the way to go
I print film & digital, put them on the wall and rotate them into storage. I also produce 12”x12” albums. I expect all of my negatives to be trashed and digital files abandoned after I pass. Other than myself, nobody has any interest in these. Regarding scanning all at maximum resolution, I think that you mistake exploring this option with intending to use it frequently. I do care about the ability to produce high resolution images when the need arises and I’ve been exploring a workflow to accomplish this. I have a heavy duty copy stand and medium format camera for 645 and 6x7 and a smaller setup for 35mm. Unfortunately, finding a Hasselblad 1:1 macro isn’t that easy, so I haven’t ventured beyond making contact sheet copies with medium format. It turns out that the 90V has a sufficiently flat field of view to copy a page of negatives/slides in plastic three hole binder pages.
If you have a Coolscan 9000, why not use that? 4000ppi is beyond the resolution of most films whether it’s 35mm or 120 since the scan scales with the size of the original.
A smarter person than me explained the issue this way:
“ As long as a 10.000DPI scan at 50% zoom looks as sharp as a 5000DPI scan at 100%, then the resolution of the slide is still beyond 5000DPI.
Fuji once specced their 35mm Provia to be 140 lines/mm. With a bit of maths this equals a pixel pitch of 3.6um for a color sensor, what equals pretty much the pixel pitch of a Fuji GFX100, which is 3.76 um.
So if i have done the maths right, Provia should max out at around 4900DPI, while the Fuji GFX sensor should max out at around 4600DPI (presumed the lens doesn't max out even earlier).
Given the fact that Provia is one of the highest resolving commercial film stocks ever, i would never feel a need to scan any other film stock at higher resultions than 3200dpi. That's why i'm recommending relatively simple scanners all the time. Myself being happy with the ancient Minolta Elite MK1 capable of only 2850DPI.
However, there still is a difference when scanning with drum scanners, one of them is the absolute evennes of light distribution across the frame. The other is what i call "bare metal sharpness". These scanners suffer none of the issues ordinary devices have to deal with. Like uneven sharpness from the lens, aberrations etc...”
This above quote is in the context of a different question but it’s related. While you “can” scan film at 5000 or even 10000ppi, it’s largely pointless and just makes the scan looks worse.
I would take your ideal print size and make a scan that equals that at 360/720dpi for Epson printers or 300/600ppi for Canon. The Coolscan should be able to do that for you and ICE is not to be underestimated.
I do camera scanning but it’s largely a back up solution for real scanners, in case there is a problem with hardware.
speedgraphic wrote:
If you have a Coolscan 9000, why not use that? 4000ppi is beyond the resolution of most films whether it’s 35mm or 120 since the scan scales with the size of the original.
A smarter person than me explained the issue this way:
“ As long as a 10.000DPI scan at 50% zoom looks as sharp as a 5000DPI scan at 100%, then the resolution of the slide is still beyond 5000DPI.
Fuji once specced their 35mm Provia to be 140 lines/mm. With a bit of maths this equals a pixel pitch of 3.6um for a color sensor, what equals pretty much the pixel pitch of a Fuji GFX100, which is 3.76 um.
So if i have done the maths right, Provia should max out at around 4900DPI, while the Fuji GFX sensor should max out at around 4600DPI (presumed the lens doesn't max out even earlier).
Given the fact that Provia is one of the highest resolving commercial film stocks ever, i would never feel a need to scan any other film stock at higher resultions than 3200dpi. That's why i'm recommending relatively simple scanners all the time. Myself being happy with the ancient Minolta Elite MK1 capable of only 2850DPI.
However, there still is a difference when scanning with drum scanners, one of them is the absolute evennes of light distribution across the frame. The other is what i call "bare metal sharpness". These scanners suffer none of the issues ordinary devices have to deal with. Like uneven sharpness from the lens, aberrations etc...”
This above quote is in the context of a different question but it’s related. While you “can” scan film at 5000 or even 10000ppi, it’s largely pointless and just makes the scan looks worse.
I would take your ideal print size and make a scan that equals that at 360/720dpi for Epson printers or 300/600ppi for Canon. The Coolscan should be able to do that for you and ICE is not to be underestimated.
I do camera scanning but it’s largely a back up solution for real scanners, in case there is a problem with hardware....Show more →
Thank you for the informative reply, My interest in copying film came about when faced with decades of negatives and slides to be scanned. The Coolscan was just too slow. It may indeed be better to quickly scan this large backlog via copying and save the high resolution work for the 9000ED.
That's a workflow I recommend. I mean it wasn't too long ago that if you were a pro you dropped off your rolls at the lab, got a contact sheet. Made your selects, ordered prints of 1 or 2 if at all. As we all know, they're not all keepers!
My path forward is clear to me, but might take a bit of time to enact. First, copying 35mm color film such as Portra, Ektar and E100 benefit from the aspect ratio of the A7CR, but there is no advantage to sixteen shot pixelshift, so I will use four shot.
645 and 6x7 formats are better served by using my X2D and once again four shot pixelshift is sufficient in all but the most demanding end use in which case the 16 shot pixelshift might be used with stocks such as E100 when using high resolution lenses such as the Mamiya 7’s 80mm. The 16 shot pixelshift mode is still in Beta. Also, I don’t yet have a macro lens for this system. I’ll eventually pick up a 120mm Macro ii.
The X2D/90mm XCD v combination seems to be a fast, easy approach to making contact sheets from my negatives and slides that reside in three ring negative binders. Stopped down, I’m getting good edge to edge sharpness and quite high resolution.
I’m leaving open the possibility of using the Nikon Coolscan 9000ED for critical work, but I’ll need to find room for the scanner and setup a modern interface to my MacBook Pro. I’ll be using my Kaiser copy stand with the X2D and the smaller Negative Supply stand with the A7CR. The A7CR copies directly to Lightroom using electronic shutter and the X2D copies directly to Phocus Desktop using electronic shutter. The Negative Supply copy system works great, except for the borderless medium format masks are a real pain and not worth the effort.
I have just made my first 110cm (43") print from the GFX 50sii (50mp) and it is easily good for further enlargement. I have also made my first scan of 645 velvia 50 using the GFX and a Pentax 645 macro lens. Judging on a 28inch 4k monitor I would say that scan is good for a 13x19 print, and possibly a little larger. My camera scans seem a little better and are certainly more flexible than what my lab does with its Noritsu, so I don't think modern camera scanning is giving up anything to 20 year old professional lab scanners - a drum scan is likely much better though.
Incidentally the conventional wisdom is 35mm is good for enlargements up to 8x12, while 645 is easily good to 13x19 assuming prints are going to be subject to close inspection. This suggests short of drum scanning there is not much more to be had from my GFX scans of 645.
So based on this, I would expect that using a medium format camera to scan 6x9 should produce a file good enough for a 40" print, ie double that of 645. This might be done by stitching, or pixel shift but 100mp might be good enough for a single pass scan.
