Film's true DPI?

I've got a mediocre 8x10 print from an enlarger on my wall right in front of me, hanging next to an 8x10 inkjet print of the same shot from a 2400 dpi scan. It's not the discrete data points that make the difference, it's the overall look of the image. And the mediocre wet print is far superior to the scan+inkjet. This includes easily visible differences in level of detail at normal viewing distances for an 8x10. It's not an issue of pixelization or resolution, it's an issue of not having the image parcelled into discrete data points which inherently loses details and destroys what we might call the "integrity of the image."

I know people will argue that the issue is my electronic equipment or my skills, but that's just making excuses. My photographic and wet printing skills and equipment are modest, yet the wet print is just so much better for far less cost and effort.

I made them both. If anyone wants to argue one needs $20K in equipment to produce an inkjet print that comes anywhere close to a wet print made with a used $100 enlarger with a used $15 Nikkor lens, save your breath and just say wet printing technology far exceeds digital scanning and printing technology. Because that's what you'd really be saying.

Verdict: Scanning and digital printing technology has got a long ways to go to beat a cheap used enlarger.

FWIW, I scan all my negatives. I wet print the onces I like. I use the scans as a contact sheet and to share the images electronically. I prefer the look of scanned film to digital photos, but won't suggest the scan is anything but a crapped out approximation of the "real" shot. A scan is not much different than taking polaroids of my wet prints. And nobody would argue the polaroids would be the actual and real test of the original camera, film, and enlarger.

For example, the scanner can only read each pixel as white or black or a shade somewhere in between. It cannot read each pixel as a cross or parallel lines or speckles or curves or gradients. So even though one might try to break things down to line-pairs vs. pixel count, it's not line-pairs that make up an image, it's the direction, intersection, curving and thickness of edges and the gradient and density of regions that make up a B&W film frame. You can overlay a grid and attempt to reproduce the image by taking an average of each cell, but that ignores the actual detail of each cell. Obviously, the simple act of averaging the density across a cell reduces the smoothness of the gradient, to say nothing about what happens to details when edges cut across cells. You can decrease the size of each cell to get a more accurate reproduction, but at some point, the entire effort surpasses the cost and complexity of simply printing the image with known technology. At what point does one admit that the entire effort is re-inventing the wheel with squares?
 
Last edited:
Hmm... figures from a company selling drum scans by the Mb and a company trying to stay in business selling film.
Many films can resolve 100 lp/mm; some can exceed 200 lp/mm. The same is true of lenses, at least centrally. The limiting factor is film flatness and film location: 125 lp/mm centrally, and more often than not, is the best I have ever seen (from the 75 Summicron).

Ilford's figures were for a slow transparency film, which they do not manufacture. The 18-20 Mp figure is widely accepted; I have yet to meet a camera or lens manufacturer who goes much outside the range 14-24 Mp, usually with the qualification 'insofar as it means anything'. Most agree that this last qualification is fundamental.

And the 30+ Mp figure was from a software designer, selling image processing software, with no axe to grind, working from a combination of first principles and experience. I do not know his working.

Cheers,

Roger
 
Last edited:
Dante, do you really think my image earlier in this thread is equivalent to 2-3 megapixels? I can provide you full resolution scan, print it and a 2mp digital image to 8x10" and see the difference.

Your math is not entirely correct too; 1/lpmm = 1/lens_resolution + 1/film_resolution. Which for your example of 100 and 100 lp/mm for film and lens gives us 50 lp/mm for the system. Also, take into account that it takes a 2 lines of pixels to approximate 1 linepair, and more if the linepair is not spatially aligned with pixel rows/columns (and usually they aren't).
 
. . . in the end a good image is a good image and nothing else matters. . . . Photography is a weird art because there is this tendency to quantify everything and make it scientific. The science is used as a crutch.
Dear Patrick,

I couldn't agree more, and I'd add that it depresses me to see people looking for a precision that isn't there even in an optimized system, and CERTAINLY isn't there in the real world of hand-held pics, different developers, non-optimum exposure (optimum for sharpness, that is -- which is unlikely to give optimum tonality)...

Cheers,

R.
 
Hi Varjag:

The formula might not be right, but the math is. You have to add the scanner (output stage) too as a third factor (5000dpi = 99lp/mm), which drops the result from 50 to 33. The scanner itself is a camera (you would have to add the enlarging lens if you used one instead of a scanner). My gut feeling about this is that (at least in this expression of the resolution limit) this extra intergenerational loss is a pretty big handicap and that in a lot of circumstances, optical printing would be a better way to get the data out into a big enlargement. The megapixels might be there, but you may be better served to use as the last stage an apo enlarging lens that has two or three times the resolution of a scanner.

I don't doubt that your image has more than 2-3 million pixels of data as printed - and putting more data into the printer, as I have seen, generally results in something that is more "sniffable." But where I am not on the train is on the question of how much of that data actually carried through from the lens and film and not from interpolation at the various stages of processing. We all love to deride digital people for all of the "interpolation" and "upsampling," but my suspicion is that hybridized workflows are doing the very same thing.

I like making big scans too (witness my collection of 3,000 6x4.5 scans at 4,000 dpi :eek:), and this is not going to change my scanning habits (what's a few more terabytes of disk space more or less... :)).

Cheers
Dante



Dante, do you really think my image earlier in this thread is equivalent to 2-3 megapixels? I can provide you full resolution scan, print it and a 2mp digital image to 8x10" and see the difference.

Your math is not entirely correct too; 1/lpmm = 1/lens_resolution + 1/film_resolution. Which for your example of 100 and 100 lp/mm for film and lens gives us 50 lp/mm for the system. Also, take into account that it takes a 2 lines of pixels to approximate 1 linepair, and more if the linepair is not spatially aligned with pixel rows/columns (and usually they aren't).
 
Hi Roger:

Is it 100lp or 200lp/mm or is it 100 or 200 lines per mm? Prior to yesterday, I thought I remembered TMX doing 200 lp/mm, but when I looked it up (as with many others), the figure was 200 lines/mm (using the ISO test), but the MTF chart showed a groundout around 133 cycles/mm (the Kodak MTF charts are measured in cycles, which means black stripe white stripe, or line-pair). And the best numbers always seem to be drawn from subjects with 1000:1 contrast between light and dark.

Points well taken re lens resolution, film flatness, falloff, etc. 100 might be optimistic. Did the 75 Summicron really do better than the 50 Summicron? Wow.

The neat thing about projection printing is that the demands on getting image to paper are relatively low, which means that the failings of everything in the chain can be pretty big but still yield a viewable picture. Grain sniffable? Maybe not. But that may be a pathology.

Cheers
Dante

Many films can resolve 100 lp/mm; some can exceed 200 lp/mm. The same is true of lenses, at least centrally. The limiting factor is film flatness and film location: 125 lp/mm centrally, and more often than not, is the best I have ever seen (from the 75 Summicron).

Ilford's figures were for a slow transparency film, which they do not manufacture. The 18-20 Mp figure is widely accepted; I have yet to meet a camera or lens manufacturer who goes much outside the range 14-24 Mp, usually with the qualification 'insofar as it means anything'. Most agree that this last qualification is fundamental.

And the 30+ Mp figure was from a software designer, selling image processing software, with no axe to grind, working from a combination of first principles and experience. I do not know his working.

Cheers,

Roger
 
I recollect arguments among friends who bought stereo equipment. They debated, compared, contrasted and generally lived stats on equipment reviews for amps, receivers, styli and speakers.

One day I was in a stereo shop and asked a guy who worked there for a number of years what thoughts he had regarding the latest NAD amps. His answer was simplicity itself. He said “this number stuff is all for the “techno freaks” because the vinyl you’re playing may be slightly warped and cause the stylus to lift and not track correctly. The stylus itself may be good or poor and that will affect your ability to hear the best sound from the LP assuming it’s an LP that is physically ‘good’. Next the amp may clip the range of signals sent through the stylus or even pick up vibration or feedback hum in the tone arm. Then we get to the speakers. Some are going to clip the range of sound due to physical limitations so you don’t hear everything that is on this perfect LP. Finally there is the link between the amp and speakers, the wiring. Bad wire will degrade the transmitted electrics from the amp to the speakers. Finally there is the room where the music is played. You can imagine the variables there too.

His conclusion; stereos will get better every year. Buy something that plays the music you like and gives the sound you like. Violin concertos are going to sound better over one system than rock or opera.

To paraphrase, buy a camera to take shots that suit you and use the equipment designed for that type of image, because “image is what this visual medium is about”. Science will advance and the images will follow or not depending on how people decide to use the technology.

BTW I find this techno fest entirely interesting, subjectively I’m not sure it’s going to give me a better image right now. I think I’ll go pull out that Diana I got on Craigslist and load it with some 120 Ekatchrome for cross processing.
 
Last edited:
BTW I find this techno fest entirely interesting, subjectively I’m not sure it’s going to give me a better image right now. I think I’ll go pull out that Diana I got on Craigslist and load it with some 120 Ekatchrome for cross processing.

That exercise will definitely be more artistic than technical. Do post your results in a different thread. :)
 
In my humble view, the phenomena of Photography is much much wider than "the final image". Few people have been born with the unfighted for privilege of a "photographic eye"', and the best have became The Masters.

A much wider number of pros and aficionados are just folks with an eye talented to this or that degree. But up to this point we have not started to count the broad and different aspects of the photographic phenomena. And this broad spectrum opens craft creative opportunities, almost unending in their numbers.

Where are we going to cathegorize a Master Printer ? Or a great camera designer ? Or a camera historian ? Or a master repairman like Henry ? Or those of us with deep and sincere love for the hobby ?

What is this kind of narrowness about the "Final Image" ?

Sorry folks but some among us have lost perspective that Photography is the most performed Art by the hundreds of millions owning whatever camera. Do you think you are outside of this crowd ?

It is within this context that the technical issues raised by Dante and discussed "once more" do have a warm place in the photographic phenomena. Photography is not just Art (kindly quote me "just Art", but also Science (and its many branches), Weaponry, Politics, History, Psychology, etc, etc.

And in all of these unmeasurable areas there is an unending room for creativity, for discovery, for originality, and for humanity.

Cheers,
Ruben
 
Last edited by a moderator:
That exercise will definitely be more artistic than technical. Do post your results in a different thread. :)

heh Andrew, there are a couple of threads like that right here at RFF including one by Todd Hanz. Don't worry I won't put that image into this thread. Not even a link. I promise. ;- )
 
1) I ve closely compared Minolta's 5400II disaster-machine @ 5400ppi with itself at 4000ppi, and with Nikon V at 4000ppi. They seemed identical, except that the extra 1400ppi created bigger files that served no purpose (like big files from flatbeds).

2) 6MP DSLR are nowhere near film @ 4000ppi
3) 10MP DSLR approach film at 4000ppi
4) 5D seems identical to film except for less latitude
5) Pentax K20D (14.6mp on unique APS-C) whups 5D files, (partially because Pentax does less noise reduction) whups film detail resolution However, like 5D and like slide film, it runs out of latitude with more than a stop of over-exposure (so I set it for three automatic 1/2 stop brackets whenever possible).
 
Last edited:
Hi Roger:

Is it 100lp or 200lp/mm or is it 100 or 200 lines per mm?
Dante
Dear Dante,

I was sufficiently uncertain about this that I called an old (ex-Ilford) friend. He recalled 125+ lp/mm (not l/mm) with the low-contrast target; as he put it, 'God knows what it was at 1000:1, almost certainly over 200 lp/mm'. This is with Delta 100, a significantly sharper film than TMX. A long conversation explains my posting this at 01:30!

From memory, I am reasonably confident that Zeiss topped 90 lp/mm on the film with the Biogon on an Alpa. My test targets both offer 100+ lp/mm with the 75 Summicron -- I suspect this is the optimum f.l. for depth of field AND depth of focus, hence its outperforming the 50mm -- though of course the chart manufacturers may have been indulging in sloppy terminology (I have been too lazy to calculate the angles subtended, the more so after dinner with a good deal of brandy on board). I'll try a more rigorous (third) chart sometime soon.

Likewise I am reasonably sure sure I recall Zeiss promising 200+ lp/mm on the film, with focus bracketing, with some ZI lenses, though of course the film flatness/location issue renders this effectively nugatory.

The important point here is that I am talking about maximum possible resolution under ideal conditions, NOT what is likely to obtain in the real world, where I cheerfully concede that 6 Mp will normally do it. But the more you can afford to throw away, the better: hence the argument for top-quality wet printing, where you normally throw away less.

Cheers,

R.
 
Last edited:
40oz, 2400ppi is pretty low for a film scan, and if it's a flatbed, it is more like an actual 1200ppi. regarding sampling theory etc., at some point you get enough samples that the difference between tradition print rendering and a sampled rendering is imperceptible using a human eye. Regarding your prints, you will have to convince me you are an expert in scanning, post processing and printing to make that part meaningful to me... With my digitized prints, only the best wet darkroom prints are better (my own and my photo buddy's)...

[edit... spelling error]
 
40oz, 2400ppi is pretty low for a film scan, and if it's a flatbed, it is more like an actual 1200ppi. regarding sampling theory etc., at some point you get enough samples that the difference between tradition print rendering and a sampled rendering is imperceptible using a human eye. Regarding your prints, you will have to convince me you are an expert in scanning, post processing and printing to make that part meaningful to me... With my digitized prints, only the best wet darkroom prints are better (my own and my photo buddy's)...

[edit... spelling error]

How about a couple of A/B comparisons using your own wet prints and ink jets. I'm interested but finding an actual example of a quality wet vs quality inkjet is hard to do.

regards, Jan
 
I used to scan my 35mm film (mostly colour neg) on a Minolta Dual IV at 3200 dpi, and thought I could see lots of grain. Now I have a Nikon Coolscan 5000 which scans at 4000dpi, and the apparent grain is *much* less. I read somewhere once that at lower resolutions the scanning pattern can interfere with the grain pattern, thus exaggerating the latter. Not sure if this is true, but it certainly fits with my experience.
 
How about a couple of A/B comparisons using your own wet prints and ink jets. I'm interested but finding an actual example of a quality wet vs quality inkjet is hard to do.

regards, Jan

Dear Jan,

Isn't that the problem? That people who are REALLY GOOD at one or the other are seldom better than competent at the other one?

Also -- from many photo-fairs such as photokina, Focus on Imaging, PMA -- I know that almost any process can be made to look good with the right subject. In B+W (not always, in colour) I remain convinced that wet-process silver delivers far better images of a far wider range of subjects.

Cheers,

R.
 
>>How about a couple of A/B comparisons using your own wet prints and ink jets. I'm interested but finding an actual example of a quality wet vs quality inkjet is hard to do.

Jan, I haven't been in the real darkroom for years (after having a horrible reaction after years of over-exposure to the chemicals)... but if I found some time, I'm not opposed to putting this test together (really, I want someone else to do it!), but how would you propose to normalize everything???? We all already know that under a loupe that film will kill an otherwise beautiful inkjet print for resolution (just like 100% digital views always look better than 100% film scan views)... when in fact, it is how the *print* looks to an actual person, when viewed in person. You will have to process the scans as both will scan differently due to the different medias used, how do you normalize that? My main contention/justification for saying that *my* good inkjets look better than *my* average traditional prints comes from the fact that I get so much more control over the tonal rendition.

Just something to think about... side by side, I am happy with my HP b&w prints (on the Premium Plus Satin paper).

Everyone that knows anything always assumes my b&w prints are *real* until I tell them otherwise. For a gallery show, I signed the archival statement (based on Wilhelm data) and listed the work as "b&w photographs," the gallery owner read it and replied, "I'm glad you're not doing that digital crap any longer." I let him hang the work and kept it list as b&w photographs, which is what they looked like and what they are... ask for details and I'll give them... otherwise, "archival b&w prints" is technically accurate.
 
40oz, 2400ppi is pretty low for a film scan, and if it's a flatbed, it is more like an actual 1200ppi. regarding sampling theory etc., at some point you get enough samples that the difference between tradition print rendering and a sampled rendering is imperceptible using a human eye. Regarding your prints, you will have to convince me you are an expert in scanning, post processing and printing to make that part meaningful to me... With my digitized prints, only the best wet darkroom prints are better (my own and my photo buddy's)...

[edit... spelling error]


My scanner is a dedicated film scanner.

And you have provided the evidence that supports my claim - "you will have to convince me you are an expert in scanning, post processing and printing." Essentially, you believe one must be an expert in three fields to be able to trump an inexperienced wet printer working in his kitchen. Need I say more?

I'm far from an expert in wet printing. But you are saying I have to be an expert in three separate disciplines to give my experience any validity for you?

How about this - I've scanned a couple thousand frames at least. If I'm not an expert at scanning my own images, I've no excuse. I've spent many, many hours in photoshop, and would not consider myself an expert. That said, I've spent countless hours researching techniques and tools. I'm hardly a neophyte. And I've wasted far too much inkjet paper trying for something better than I can make.

Now, I could certainly benefit from a newer, better film scanner. I could certainly make better use of more expensive papers and inks in a newer and more expensive inkjet (or whatever) printer.

But my used $100 Beseler 23C, a used Nikkor lens, and pretty standard Ilford Multigrade IV paper manage to totally shame what comes out of my electronic equipment. And I print in my kitchen with sheets over the living room windows. I rinse in my kitchen sink.

What more evidence is needed to express the superiority of one technology over another? On the one hand, there is never a point at which you've spent too much money or time on electronics and supplies, whileon the other, really rather modest and perhaps barely adequate means turns out a superior product. It's not really a question *if* wet printing is superior at this point for me.

I understand if a person feels they don't have the time or resources to access a darkroom or make one, but take all the time and money put into scanning, editing, and printing on a computer, and you'd have access to whatever darkroom you'd want. It's simply a matter of trying to make it work. Instead of spending thousands on scanners, printers, software, and computers, spend that money on darkroom rental. Instead of spending huge numbers of hours on scanning and editing, spend that time wet printing. Not everyone has that flexibility, perhaps, but it is hardly as difficult as I think most believe.
 
Roger, thanks. Your comments have now crystallized a thought in my head. Certainly the "look" on the wall is what it is about. Then it all becomes personal taste.

MH2000, interesting response to the matter of interpretation. I'd probably do the same. Still it hits me that it might be as simple a matter of choice as "do you want your prints glossy or matte finish" at the local lab. I'll know which I want and tell them so. I think the matter of inkjet vs silver will eventually become one of taste, disregarding implict costs of each method.
 
>>What more evidence is needed to express the superiority of one technology over another

I'm not biting, I'm not expressing the superiority of either... but some people are getting great results from both. My tendency is toward favoring traditional prints, but with the control that you get from digital processing of film images, it often makes up for the difference.

For B&w, only the HP gray dye set, the Epson gray pigment set and the 3rd party quadtone sets can be compared to traditional prints... are you using one of these systems?

Comparing prices... well, since everyone on this forum already has a computer and will probably add a printer anyway (IE. it won't cost any extra to just choose a dirt cheap HP that can take a gray cartridge), all they need add is a scanner... and a used Minolta SD-II goes for like $50 (and that is a really capable b&w scanner!)... cost can be considered comparable... cheap.

Cost per print, seems about the same for fine quality enlargements from either... once you include everything you put into a darkroom.

Regarding a 2400ppi film scanner, it is either very old or very cheap... no offense, but judging results from that and extrapolating to general conclusions isn't really convincing.

I've been working with b&w photography for 30 years... plenty of darkroom experience... if I hadn't developed a sensitivity to photo chemicals I probably would never have gone to scanning... but photography is too important to me to abandon... so I had to move on.

well... I guess you got me to bite a little bit! ;)

I really appreciate both processes and the results... really, enjoy what is giving you the best results.
 
Back
Top Bottom