rscheffler wrote:
Also of interest, all/most of the images at that site, shorpy, etc. culled from the LOC collection, have been retouched and colour enhanced, therefore you're not really seeing how the film *really* looks now, after 70 years.
That makes sense - the first reaction that I had when looking at the samples was "how digital they look". The originals look much more like film. And I am not saying that to mean that looking digital is good or bad, its just that the samples didn't match what film shots typically look like.
Its great that today you can get quality which is at least as good as these 4x5's, if not better, in a much smaller 35mm format - technology has made a lot of progress.
curious80 wrote:
Its great that today you can get quality which is at least as good as these 4x5's, if not better, in a much smaller 35mm format - technology has made a lot of progress.
O_o Where? I'd like to know so I can get that setup.
Zaitz wrote:
O_o Where? I'd like to know so I can get that setup.
To be clear I just meant these particular 4x5 shots as displayed, not 4x5 in general. There is nothing in these shots at the sizes that they have been displayed which cannot be captured by any recent FF DSLR. It may be the case that the resolution on offer is higher but we don't have the full-size shots so we don't know. In terms of the rendition of the scene, it is certainly within the realm of today's DSLRs.
To be fair, the "originals" look to my eyes as original unprocessed scans. I doubt the original kodachromes had such flat look, rather the opposite. Scanning has always been the weak point in digitizing film, it rarely reproduces the richness and pop of the "originals" as seen through a light box or projected.
luminosity wrote:
Kodak is not out of business. That is misinformation that needs to stop being perpetuated. A great many current companies have been through bankruptcy, some of them more than once. The best case for all of us is that Kodak ends up like Ilford, which of course was bought out and allowed to continue as it was.
No chance of that happening. Kodak has stated in their press releases that film is NOT part of their core business anymore and will be sold off. Kodak has no interest in remaining a film producer. They are putting ALL their eggs into the commercial printing business.
Ilford has fanatical commitment to film. Kodak does not have ANY commitment.
edwardkaraa wrote:
To be fair, the "originals" look to my eyes as original unprocessed scans. I doubt the original kodachromes had such flat look, rather the opposite. Scanning has always been the weak point in digitizing film, it rarely reproduces the richness and pop of the "originals" as seen through a light box or projected.
I think the 'originals' are actually dupes of the originals... you'll notice an orange digital-looking three digit number along the border of many of these images, implying relatively modern duplication.
Some of these, such as the landscapes, do look like Kodachrome to me. It was never a super high 'pop' film like Velvia, though it was high in contrast. Personally, I believe all of these images would have been 'better' rendered with a quality digital camera, when it comes to colour quality, as opposed to sheer resolution. The Kodachrome appeal for some, might be that they've seen the look in filters applied by Instagram and Hipstamatic, lending familiarity to the truly vintage images. Not saying it's the case for those posting here, but for the general public...
It would be interesting to see the actual original scans and know how they were done. I remember hear that around twenty years ago that either the LoC or the Smithsonian, and since it's been a while - that one or the other at that time bought two ICG drum scanners, arguably one of the best drum scanners ever made but long since defunct as a company.
The "original" scans post here, if that's what they are, certainly remind me of how much of the vintage Kodachrome I've scanned from the 1950's looks. A lot depended on how the film was handled, exposed and processed, so there could be a lot of variability in the final developed film. The other problem is that Kodachrome does not react to scanning the same way other color films react. Actually, it's the inverse of that. Scanners see a blueish-ness in the blacks of most Kodachrome that we never see with our eyes and drum scanners can actually see that color in the blacks and will register that black with 15 to 20 points of blue in the black, which also gradates up into the midtones as well. I would always fix that at the scan level but it didn't matter if you did it there or in Ps later on. And, of course, if you wanted to record all the shadow detail on the film, you needed to use a drum scanner rather than anything with a CCD chip.
Overall I think the LoC scans feel much better than the overcooked ones referenced in the first post. Just because it was Kodachrome doesn't mean it had to have garish colors. Velvia was still nearly half a century off.
I've been doing research on how to digitize Kodachrome because I've really need to liberate my transparency library from the dark storage mausoleum. I went through the past 20 years of posts on this site searching for "Kodachrome" to find good and bad examples of the digitization process. Unfortunately because I'm a consumer I won't have the ability to drum scan my library. The only workable option I've found so far seems to be "slide copying" with a digital camera, a macro lens, and an extension tube/slide holder device (Nikon ES-1) or a bellows/slide copier setup (PB-4/PS-4 or PB-6/PS-6).
One of the biggest impediments I've had toward getting this project going has been the dismal failure of most of the Kodachrome digitizing processes I've encountered. Kodachrome is something that I think of as a warm-toned emulsion, not a cool-toned emulsion. It's also something that I think of as balanced, without selected colors pushed to an extreme. Yet just about every Kodachrome digitizing project that I've encountered has used poor scanning correction that has resulted in bastardization of the "Kodachrome look" to the point that what new generations recognize as the Kodachrome-look based on scanned images has nothing to do with the look of real Kodachrome. Yes, it appears that the scanning process is very difficult for Kodachrome, and it's a very rare occurrence when somebody actually gets it right. (I bumped this thread primarily because it has good comparisons between the more natural LoC images and the perverted digitally processed images that look like they were colored by a comic book inker.)
The difficulty obtaining accurate results has led to "standards" for digitizing Kodachrome that are quite poor. Much of the industry has just accepted the "blue look" of Kodachrome scans as they come out of poorly-corrected scanners as if they are true to life. A garish looking artifact of improper scanning is now being accepted as an accurate representation of the Kodachrome look, such that the real look of Kodachrome just won't be available to anyone who lacks their own dark-storage library. This problem is so extreme that the industry is now remembering Kodachrome as a cool-tone emulsion. Gone are the days when people remember Kodachrome for neutral coloration, perhaps with a little bit of excessive vividness in red. Now Kodachrome is being remembered more as a flattened dull Ektachrome look, or a blue-to-the-max look, than as itself.
The perception of what Kodachrome is supposed to look like has been so badly corrupted that companies are selling preset "profiles" for use in post-processing that intentionally "emulate Kodachrome" [sic] by shifting the black channel to blue! It appears evident that they are not trying to replicate the true look of Kodachrome -- they appear to be trying to replicate the true look of a half-assed scan of Kodachrome.
2 years ago Fred had started a thread for a promotion on this site by Cobalt Imaging, a company that provides "film emulation profiles". Everyone jumped on the Kodachrome emulation profile bandwagon and it seemed that I was alone in taking issue with the fake-blue Kodachrome profile. Maybe I'm just more sensitive to color variation than the average guy, maybe I've just got a good mental image of the real Kodachrome look, but those Cobalt profiles for 1960s Kodachrome looked like absolute garbage to me. They took the blue-channel errors, amplified them, and based an entire "film emulation" product on exaggeration of the improper rendition of Kodachrome in half-assed digital scans.
Cobalt: 1960s Kodachrome Emulation Pack <-- scroll down and look at the colors in the image of the bearded fellow in blue jeans, sitting against a brick wall.
Thanks for the link to Cobalt. It turns out that I previously examined their products after reading one of FM's posts that offered special cobalt pricing for forum members.
Being a lifetime shooter of PKR, PKL and PKM -- and someone who was reluctantly dragged into digital when Kodachrome processing got shut down -- I was particularly intrigued by the Cobalt "Kodachrome" emulations, but I was completely disappointed by how much they adulterated the blue channel balance on all of their Kodachrome emulations, changing the tone of the emulsion from warm to cool.
I used to pay premium prices to buy refrigerated Kodachrome emulsions that were aged for exact coloration at the time of sale. The Cobalt 60s Kodachrome profile looks nothing like my color-controlled professional Kodachrome slides, and the colors look more like they were painted by a comic book inker than a true Kodachrome rendering. The image of the bearded fellow sitting on the sidewalk, in front of a wall that completely changed color from gray to blue, and his blue jeans that were pumped up to a surrealistic 2^14 saturation killed the appeal of the Cobalt package for me. I hate to say it, but I felt like I was looking at something with the subtlety of a FroPack. The look of their simulated "Kodachrome emulsion" appeared cooler to me, more like they were channeling Ektachrome than Kodachrome....Show more →
I also think the Cobalt "Kodachrome Last Roll" profile is particularly bad, as it fails to emulate the look of real Kodachrome and instead replicates the look of a crappy digital scan. I have a great many rolls of PKM, PKR and PKL that were processed during the final month. When projected, not one of those rolls mis-renders blacks as blue -- that's entirely an artifact of poor quality scanning.
I'm still trying to address the color accuracy problem. Last night I pulled some random Carousels off of the storage racks and projected them onto a Da-Lite screen to refresh my memory... and yes, the process of digitizing Kodachrome just doesn't match projection of the real thing. At least I haven't found a workable solution yet. I'm hoping that someone can teach me how to get it right.
I don't have any experience with 1940s 4x5 Kodachrome, so this is more of a question than a comment.
When I look at that LoC collection on flickr I'm seeing an awful lot of images that used flash photography where someone has post-processed them and forced the artificial light source to render as 5400K -- which has caused objectionable color shifts in the images. 5400 K flash didn't exist back in the day, and the result was that Kodachrome when exposed using artificial flash did not render colors as seen in the LoC photostream. That is to say, I think the LoC photostream's colors on the interior shots made with artificial lighting just look wrong.
To my knowledge the flash sources that were used in the 1940s were not white lights -- at least not the color temperature of white that is depicted in the Kodachrome flash photos in the LoC photostream.
To me it appears that the "editors" who were proofing these digitizations have applied modern electronic white balance standards to all of the 4x5 Kodachrome images to render them white according to today's flash light sources that have a color temperature of about 5400-5500 deg K. Electronic flash had not yet been invented back in the day that those Kodachrome images were taken.
In that era the industry used chemical (magnesium or zirconium) flash produced by controlled explosions, not electronic (Xenon) flash.
(Does anyone remember burning your fingers when trying to change out your flash bulb so you could take the next shot?)
Once replaceable flash bulbs were invented by Leitz Wetzlar, flash was typically provided by an explosion of a fulminate that triggered the ignition of a magnesium (or later zirconium) foil in an oxygen environment. Because the chemical flash didn't have the proper color temperature for daylight film, the bullbs were coated with a blue coloring to aid in white balance. This technology persisted through the development of flashcubes for Kodak Instamatic cameras, until electronic/Xenon flash eventually took over.
Does anyone remember the color temperature numbers for magnesium / zirconium flash bulbs? I'm trying to determine what the real color temperature would have been for those indoor flash-illuminated LoC images, and how the color would have shifted when those images were scanned / corrected and forced into a white balance of 5500K.
The skin tones on those factgory worker photos have certainly suffered -- I've never seen Kodachrome render like that, and I'm thinking it has to be an artifact of forcing the white balance from flashcubes to comply with modern standards of 5500 K. To me someone has forced a square peg into a round hole. In other words, LoC didn't even get the Kodachrome colors right.
So, after looking through more of the LoC Kodachrome 4x5 photostream, I found two images that clearly illustrate the black and blue problem that Peter described.
Drill press operator, Allegheny Ludlum Steel[e] Corp., Brackenridge, Pa. (LOC)
by The Library of Congress, on Flickr
Drill press operator, Allegheny Ludlum Steel[e] Corp., Brackenridge, Pa. (LOC)
by The Library of Congress, on Flickr
While I agree that there is a difference and the blue of the first scan is not seen in the second, the exposure/brightness is also different, and that may well explain some of the difference. As none of us know what the originals look like, we presumably are guessing at what is correct. When scanning it is often not possible to match the original precisely, so choices have to be made and decisions made as to what part(s) of the image you want to make the most important. If you consider showing greater shadow detail is better, then one might consider the second one superior. Rebalancing the shadow color may well have adversely affected the flesh tone, so was not done.
Ironically, cheap CCD scanners (and for me that means pretty much any CCD scanner) don't have as much of a blue blacks issue as high end drum scanners, and that's because they simply cannot differentiate colors into the d-max of Kodachrome where the black can be as high as 3.8. The drum scanners have more than two stops better dynamic range with real results up to 3.8-3.9 density where the CCD scanners max out at about 3.0.
In every instance of scanning Kodachrome on my Howtek, and we're talking many hundreds of scans simply manually ticking the blue that registered in the blacks down to where the reds and greens were not only removed the blue-ish cast from the shadows but did it in a way that it gradually tapered off as you hit the highlights, and it actually improved the skin tones as well.
Also what I found is that using a Velvia scanner profiling target from HutchColor and generating the scanner profile in Gretag ProfileMaker Professional worked quite well overall for Kodachrome and then that last little bit was taken care of with the black point tweak. But to be honest, none of this is any kind of secret. Anyone who reads the actual RGB pixel values while scanning or in Photoshop would instantly see this and know what to do - or, well, I'd hope that they would. Really, it is not that hard if you have the very basics of color correction and setting end points down. And that the bulk of Kodachrome's problems are from blue shadows, you always want to check for casts in the whites as well and correct them if those whites are supposed to appear neutral. Obviously you would not want to neutralize out a white (or anything else) of an image shot at sunset, so it's up to you to look at your images in their full context.
Robin I think you're right. It's likely that I wrongly assumed that one image had attempted correction while the other had not.
But then I'm not sure that we're even looking at Kodachrome transparencies in the LoC stream -- I tried looking up those notch codes, to see what emulsion is indicated by two "V" notches followed by two wide "U" notches. The closest I've been able to come in matching that pattern is Kodak Professional Copy 4125, not Kodachrome. I've not been able to find any notch codes for 4x5 Kodachrome.
Peter Figen wrote:
Ironically, cheap CCD scanners (and for me that means pretty much any CCD scanner) don't have as much of a blue blacks issue as high end drum scanners, and that's because they simply cannot differentiate colors into the d-max of Kodachrome where the black can be as high as 3.8. The drum scanners have more than two stops better dynamic range with real results up to 3.8-3.9 density where the CCD scanners max out at about 3.0.
I need to stop for a minute and ask for some help to make sure that I'm following -- you're referring to the d-max of Kodachrome while referencing a dimensionless number, which makes it hard for me to understand exactly what scale you're using. I imagine that you're so used to using a particular scale with your scanner that you don't have to stop to think about what the dimensions of the number are, or how to convert them to a different scale. I am familiar with 3 ways in which optical dynamic range is measured, and I'm assuming you're using one of the most common scales. Please let me know which one so that I can assure that we're on the same page:
Dynamic Range units
a) f-stops, scaled as log2. Changes by a factor of 2, also referred to as Zones or EV. Commonly used by photographers.
b) decibels (dB), scaled as 20 log10. 1 density unit = 20dB; one f-stop = 6.02dB. Commonly used by electrical engineers. The measurement normally used on sensor data sheets.
c) density units, scaled as log10; 1 density unit = 3.322 f-stops. Commonly used by optical scientists.
I'm assuming that you're referring to the decadic (base-10) logarithm of the reciprocal of the transmittance through the film (as measured by a scanner), or (c). Please let me know if I'm correct on this. If so then I think this is where you're getting the result of 2 stops:
f stops = antilog (3.85 - 3.0) = 2.14
Assuming that I'm right on that, where are you getting the values of 3.0 and 3.8 to 3.9 as outputs when you do your scans? Is one of your software packages reporting those numbers?
