Optimal resolution for digitizing slides

Discussion in 'Film and Processing' started by mark_pierlot, Jul 20, 2015.

  1. I'm going to be having my father's slides digitized, and am unsure what the optimal resolution should be. I had assumed that the highest available resolution would be best, but the manager of the place where I'm having it done claims that ultra-high resolution scans would exceed the resolution of slide film and expose and exaggerate imperfections, thereby yielding inferior results compared to scans done at more modest resolutions. (The cost per slide will be the same regardless of the output resolution.)
    Almost all of my dad's slides are Kodachrome. They were shot between the late 1960's and late 70's with his Minolta SRT-101 and Rokkor lenses, and are well exposed and in remarkably good condition, having been stored in cool, dry, dark conditions.
    I will be sharing most of the photos online and on computer via DVD or flash drive, but would like the option of being able to make fairly large prints of some of them. But, even more importantly, since my main motivation for the project is to archive part of my family history, I don't want to later regret that I chose a sub-optimal resolution.
    So, in a nutshell, which output resolution(s) would be optimal for my purposes?
  2. awr


    I can't see where "too much" resolution is a problem. It would only more sharply define the existing film grain structure (and Kodachrome doesn't have much)
    I scanned some B&W 35mm at 4000 lines, here's the result.. http://goo.gl/1oyWrL
  3. Minimum of 2000 dpi would get you 2000x3000 or about 6mp equivalent. That would give you very nice prints at 12"x18"
    and even acceptable prints at 16x20

    But higher resolution is better.
  4. In my experience, 4000dpi is about the ideal resolution. Any higher and you're just magnifying the grain.
  5. At the maximum optical resolution for the scanner being used, and recorded in the TIFF format using the Pro Photo RGB
    color space. At 4000 pip, the files will be approximately 110mb per slide.
  6. Thanks for the quick responses, guys. It seems that the consensus is 4000 dpi. What would be the approximate megapixel equivalent to that resolution? Would higher be even better, or overkill?
    Anthony, those are stunningly good concert portraits of some of my idols! (I'm a jazz lover.) I greatly admire your ability, and even more greatly envy the fact that you were able to witness (and photograph) those legendary musical artists live.
  7. awr


    Thank you, Mark. You can see why I took pains to get the scans right. Those were done on a Howtek, under liquid. I didn't own the machine, nor do I want to buy one now.. but I'm in the market for a scanner, as I have thousands more frames to digitize, 35mm and 6x7cm.
  8. Check out this site. They use a Heidleberg Tango scanner the best.
  9. You didn't say what the lab's scanner is and what are the practical limitations of their scanner. Asking them to provide 4000 when their scanner's effective range is only 2400 will not provide more resolution and may increase the noise around grain. Find out what they provide and if it's not high enough resolution, then you may need a better scanner from another lab. Note that better scanners cost more money to scan. So you'll have to take that into consideration. Also, you can select the photos you may want to blow up and scan those at higher and more costly resolutions. Leave the rest for smaller prints and archiving all at a lower cost.
    Kodachrome is wonderful film. Here some scans I did from 45 year old slides. It was my own flatbed type scanner - Epson V600. I used 2400dpi which was reduced for the web. I didn't print these so I can't comment on how big I could make them. Good luck.
  10. Note that Kodachrome scans will need a fair bit of post processing to look good. And unfortunately it's no matter of finding out the right setting for a certain type of film or for a certain batch, you'll need to adjust every scan individually.
  11. What would be the approximate megapixel equivalent to that resolution? Would higher be even better, or overkill?​
    Technically the megapixel equivalent of a 35mm slide at 4000dpi will be around 21.5 megapixels. However, that does not mean the resulting scans will match the image quality of a 21 megapixel DSLR. They will be nowhere near as detailed but at least you can be confident that you have captured just about all the detail possible. Scanning at a higher resolution will just be overkill. You'll just be enlarging the grain and filling up disk space.
    Make sure the scans are being done on a proper dedicated film scanner and not a flatbed scanner. Flatbed scans are only good for 2800dpi at best.
  12. OTOH, a 21.5MP DSLR actually shows much less than advertised.​
    But more than a frame of 35mm film, which was my point.
  13. Discussing the theoretical maximum resolution of a frame of 35mm film compared to a digital camera is of little use to us
    now, when we are considering scanner technology which hasn't progressed in more than ten years. Alas, scanner
    technology is not likely to advance at all from that historic point.

    We are left with what we have. And of that, what is available for a reasonable cost and quality (based on the skill and
    care of the service provider)

    Now, it may even be best to simply photograph a slide show projected on a high quality screen with a state of the art
    digital camera.

    Back in the olden times, it was considered that the maximum size print one should attempt with a 35mm frame was 8" by
    10". Given that standard, a 2000x3000 scan provides much more resolution than is necessary.

    One should consider how many of those thousands of frames would be printed at larger than 8x12 and take that into
    consideration for their scanning strategy.

    Of course with careful cataloging and preservation of the originals, an occasional candidate could easily be retrieved and
    sent out for an advanced scan.
  14. Oh, by the way, Digital ICE supposedly doesn't work on Kodachrome.

    The film is more like black and white in that regard. So they will probably have to turn ICE off and you will have to clean
    up the dust and dirt in photoshop.
  15. Find out on what exact scanner they scan. A serious lab will tell you that.
    Also tell them, that part of the films are Kodachrome. This practically excludes ICE.
  16. The last Nikon Coolscan scanners (like the V) had ICE that would sometimes work with Kodachrome. When you need it and it works, it's great.
  17. Look at the prices for scancafe.com, who do 3000 dpi scans from slides for $0.22 each. That's plenty of resolution, and they have a fine reputation.
  18. Is this unqualified statement based on your own personal experience/results or are you referencing third party info? Can you share your data supporting this?​
    Not falling into that trap Les :)
  19. It seems like doing some test scans at different resolutions would be a good idea.
  20. Not falling into that trap Les :)
    Too late. I would argue that, by making a broad brush statement, without being prepared to back it up, you are already IN. That's OK, by the way, people will draw their own conclusions. Cheers.
  21. Whoa! I ask an innocuous question about slide scanning, and end up fueling the flames of the film vs. digital debate. Just to set the record straight, I appreciate the respective virtues of both film and digital photography, and would never argue for one over the other.

    When I asked the question about the megapixel-equivalent resolution of Kodachrome, I just wanted to get a rough idea what "digital resolution" would best fit the film's resolution. (Since I use my DSLR's a lot more than my film cameras, I tend to think in terms of digital resolution.) I know that it's comparing apples to oranges, since their are many other variables at play, but I was looking for a crude approximation.

    And it seems that Google is my friend (thanks, JDM!), since a cursory search revealed that the approximate digital resolution equivalent of 35mm Kodachrome slide film is 20MP. That "fact" suits me fine, since my full frame DSLR happens to be 21MP. Now that that "academic" issue has been settled for me, I can get on to the practical task of getting my Dad's slides scanned - at 4000 dpi.

    Upon further inspection, I have discovered that not all of his slides are Kodachrome, anyway. Many are Ektachrome, and some have no film-identifying markings at all, since they were presumably not developed at official Kodak labs. The good news is that most of them have survived well. Even the Ektachrome slides have retained their colour and not faded to an orange hue (as cheaper slide film tend to do over time). Some slides have lost some of their emulsion around the edges, or look a bit splotchy in the background, but in all cases the subjects of the photographs are clear and identifiable.

    I want to thank everyone who has contributed to this tread, especially those who provided examples of their own slide scans and those who recommended a 4000 dpi minimum scan resolution. And I would appreciate any further commentary.


  22. ICE works on Ektachrome but not Kodachrome. Good luck with your project.
    Here are my Ektachromes. Like my Kodachromes, these were scanned at 2400dpi and 16 bit color. I used the auto correction built into the scanner (Epson V600) which on retrospect caused many shots to clip. But I was into my learning curve at the time. I don't know if outside labs have similar problems with their auto adjusts.
  23. awr


    This is an interesting discussion for me, as I am in the market for a quality scanner. I would prefer multi-format, as I have 35mm and 6x7cm frames to scan, but I am looking for some wisdom and experience before I take the plunge. The Nikon 9000 looks good to me; any relevant advice would be appreciated.
  24. I gotta chime in here because I went down this road.
    I scanned a TMX negative that was developed in TMAX at 7200dpi on a Plustek. I got a HUGE file - that I since deleted because it was pointless. [EDIT] Got rid of the scanner too and now I'm 100% analog for film.
    Wanting to see how big I could make the image, I zoomed in /enlarged to the maximum size and I saw sliver grains.
    There's a point where scanning at a high dpi becomes counter productive because there is only so much information a negative holds about the image and subject. Capturing the individual silver grains is just irrelevant information for a digital file - unless you are saving the file so that your great great great great grandchildren can replicate the negative at the silver level in their replicators on their star ship.
    I would even argue that if you want a huge print to scan at a lower dpi so that the silver grains are blurred. It would look better at a distance and up close - and add a bit of silver print character.
    This whole scanning at obscene DPIs is just another form of pixel peeping.
    4,000 dpi is more than enough for any negative.

    And to add, I took a 5MP image on a Kodak point a shoot and printed it at 11x14 at Wolf and that thing is sharp and clear. Meaning, that thing captured a small fraction of the information you can get from a 35mm negative and it looks AWESOME!
    Don't get me started on these new 50+ MP SLRs on the market ...OY!
  25. Do you have supporting data for this or is this statement just based on a casual observation from your own personal results or third party reference?​
    Here he goes again... my advice to R David would be not to respond. That is unless he wants to see lots of pics of coloured pencils and 100% crop scans of human hair.
  26. is this statement just based on a casual observation from your own personal results​
    Yep. And I'm not ashamed of it either.
    At - it looks like - 3,200dpi you're just scanning irrelevant information - unless you want to know the silver grain structure of the negative. But for better resolution, I suggest a scanner that can scan the molecular state of the emulsion - what? 1,000,000,000 dpi? Or better yet, the quantum state of the silver in the emulsion. But then, Schrödinger's cat may die and I don't want that.
  27. I think it depends what you`re trying get out from your negatives or slides. If you want to have an excellent scan from carefully exposed and developed TMAX 100 negative and really want to get out everything there might be, you`ll need a 8000 ppi drum scan.

    Having said that...

    I also believe that in 99% of cases 4000 ppi is definitely high enough. But much more important than extra high dpi is the dynamics of the scanner - especially when you`re scanning Kodachromes.

    Nowadays I like Nikon D5200 body, reversed EL-Nikkor 50 mm f/2.8 N + Heliopan 216 set-up ring (52/40,5mm) with Nikon BR-2A Lens Reversing Ring. Small, sturdy & fast combo with > 4000 ppi resolution. No moving parts like bellows between the body and the lens.

    Esa Kivivuori

  28. The attached photos were taken using a Leica M3 and Ektar 100 film with a Summicron 50/2 lens, and a Leica M9-P, using a Zeiss Biogon 35/2.8 lens. The purpose was to compare the relative quality of the image using the same camera and same subject, within seconds of each other. The film was scanned on a Nikon LS-4000 at 4000 ppi. The camera has an 18 MP CCD, without an anti-aliasing filter.
    The film scan clearly resolves the dye clouds. As you can also see, the digital image has more resolution, less noise, and a better depiction of the textures in the subject. Scanning at an higher resolution would serve no useful purpose. I was not diligent about using the same lens, but if anything, the 50 mm (film) lens has greater magnification and should show more detail. The lens is not the limiting factor in either image.
    The first two are film and 1:1 detail. The second two are digital and 1:1 detail.
  29. Kodachrome is a relatively contrasty film with very dense blacks. Don't confuse the dynamic range of the slide with it's ability to capture the dynamic range of the subject. There is almost no detail in the dense areas of Kodachrome, and none at all at the other extreme (white or clear). It is the intermediate areas that count, consequently most scanners are more than capable of dealing with Kodachrome. Furthermore, Kodachrome is treated with an UV/IR absorbing lacquer on the emulsion side to better preserve the image against abrasion and time. It is this lacquer which renders ICE ineffective.
  30. Ed: What sharpening did you apply to the film scans? Why are the film pictures so green? (foam is green)
  31. No sharpening was applied to any of these images. The color balance is pretty much as it came from the scanner. It can be adjusted, the need for which is par for the course with negative film. If I adjust the foam to a more neutral hue, the water takes on a magenta hue. Pick your poison.
    In any case, color balance was not my primary reason for this comparison. Any attempt to make the film scan match the digital image is probably beyond hope. In general, I find I can match one primary color fairly well, and sometimes two, but something always falls out of bounds. Every process distorts something. You have to pick something that distorts in a pleasing way and move on.
    In terms of sharpness and contrast, using a rangefinder (or mirrorless) camera is about as good as it gets. No DSLR lens comes close to the performance of a Leica, Zeiss, or even Fuji lens. This serves to highlight differences in the medium and process of producing an image rather than the optics.
  32. No DSLR lens comes close to the performance of a Leica, Zeiss, or even Fuji lens.​
    As Les would say... "Do you have supporting data for this or is this statement just based on a casual observation from your own personal results or third party reference?"

  33. Crayon pictures aren't casual? MTF data would show something, but there is no standardization for MTF data either. Nikon, Leica and Zeiss publish curves based on measurement taken from production lenses, whereas Canon (admittedly) generates it's curves using ray-tracing software form the theoretical design. One method includes manufacturing tolerances, the other doesn't. DXO data is derived from the sensor with which the lens is used, and not all sensors are the same either.
    To make a long story short, I most definitely use my personal experience to reach the conclusion that lenses designed without the need to clear a swinging mirror are, in general, superior to SLR lenses. The differences narrow in the center of the field, as the lenses are stopped down (e.g., f5.6), and for focal lengths beyond 135 mm (there's nothing to compare). However I can use practically any lens on a Sony A7 with the right mount. I don't have "all" lenses at my disposal, but I can compare the ones I do have, namely Leica, Zeiss and Nikon, on a common platform.
    Here I compare a Nikon 50/1.4 at f/1.4 and f/2.0 with a Zeiss Loxia 50/2 at f/2.0 at the right edge of the field, starting with an overview of the source. It was a flat day, so there's no "pop" to compare, which is perhaps the strongest hand of Zeiss. The overview is using the Loxia, showing a hint of Zeiss "pop". There is no sharpening or other adjustments beyond Lightroom defaults.
    Nikon 50/1.4 @ f/1.4
    Nikon 50/1.4 @ f/2.0
    Zeiss Loxia 50/2 @ f/2
  34. For Ektachrome 100:
    Kodak claims 50 cycle/mm, or about 2500dpi at 30% response.
    That should be most of the size of the dye clouds.
  35. If you can scan at 4000 ppi or higher, that is preferable in my experience. Scanning at lower frequencies may capture all of the image details, but there is more to a film image than simply the image details. The texture and granularity of film looks more pleasing when scanned at higher resolutions. A lot of what people have come to expect from film over the last 15 years, in my opinion, is due to nothing more than grain aliasing artifacts aggressively sharpened in post-processing. At lower frequencies, grain aliasing becomes harsher. If you scan at high resolutions, it becomes less visible. The result is that you end up scanning at a resolution that is higher than the spatial resolution of the film, but the grain becomes more pleasing - even finer, in my experience. Tri-X is rated to 50-100 lp/mm. I scan routinely at 116 lp/mm (about 5900 ppi), and find the grain looks better and the fine details look *fine* as they should, compared to when I scan at 80 lp/mm (4000 ppi). At 4000 ppi, the finest details are visible, but they look thicker and blockier than they do when scanned at higher resolutions. I imagine 8000 ppi scans are even better in these regards, but I don't have the means yet to sample at those frequencies.
    So, the bottom line is, oversampling a negative produces better results, even if you aren't necessarily getting more information about the scene. You *are* getting more information about the medium, and this seems to reduce artifacts.
  36. Here is a crop of the above image:
  37. Rory,
    How was this scanned?

Share This Page