Back in 1992, I had the opportunity to work in a professional photography studio for a short time. One of the experiences that stood out for me was taking a loop and inspecting the details on an 8x10 transparency. It was an architectural piece a photograph of a six-story building taken with a tilt lens. I was able to look under a magnifier at one of the windows of the building and I could see the details in the jewel lens of the ceiling trough lights inside the office space of that window. That was quite an amazing amount of resolution captured on film considering the entire frame encompassed the entire six-story building.
This is a complete misunderstanding of how film works. Silver halide crystals have quantized size. Not only do they tend to clump together to effectively form larger clusters, but smaller crystals have a negligible effect on the image. Color film is even worse as it uses dye dots, which are significantly larger than silver halide clusters. Furthermore, the line pair ratings are for contrast of 1000:1, or in other words 10 stops of light. That's close to the maximum dynamic range a piece of film can capture. A test chart cannot recreate such contrast - paper will never get that bright/dark. So you have to work with much lower numbers. In practice there is no way you can get remotely close to 54mp from a piece of 35mm film. Color film nets a equivalent resolution of 5-10mp, bw is around 10-20, both depending on the exact film stock. With modern sensors digital has significantly denser resolution than even the sharpest film stock. If you take fuji's 100mp medium format as a benchmark for digital you'd have to go to 4x5in to beat in in bw, and to 8x10 to beat it in color. Frankly this video is just misinformation. It is a complex topic, but that doesn't take away from the fact that this is wrong.
I don't intend to be argumentative, but am curious about some of your assertions. A contrast ratio of 1000:1 is 10 stops, agreed. Also agreed that paper will have trouble with that dynamic range. But where is your data for 5-10 MP for 35mm color film and 10-20 MP for black and white? I know that Fujifilm calls its GFX series "medium format." But, the sensor size is about 44x33mm, certainly larger than 35mm but quite a bit smaller than 6x4.5cm. Furthermore, the Bayer interlace pattern means that about 3 pixels are required for each color pixel rendered in the final medium. This general principle (multiple digital pixels required on the sensor for each rendered) applies to digital sensors regardless of the RGB sensor pattern used. As you say, this is a complex topic, but one that I am fascinated by. There are lots of reasons to prefer digital over film (and vice versa), but I'm thinking that resolving power is not the main reason for the preference. Thoughts?
@@nvrumi In most documentation on film a secondary line value is included, at contrast 1.6:1. If I remember correctly it was enforced by the ISO since the 1000:1 value is basically never relevant. If you take those values (there are some in the documentation provided in this very video) you'll end up with resolution similar to what I described. Of course different film stocks are going to vary. I read a paper recently which compared some of them; if I can find a link I will attach it below. The bayer pattern does make this somewhat more complex. In this case it is more helpful to look at luminance and color resolution separately. The luminance resolution is preserved really well even through debayering, while color gets a bit more lost. The multi layer approach of color film conserves both types of resolution equally (or foveon sensors). Applied to numbers we could assume that a sensor with 24mp has its full resolution for luminance, but only about a quarter for color (or half in each dimension), whereas color film will have the same 5-10mp for both. In practice color resolution is less relevant, as we see that less well than luminance differences, but on a technical level yes, digital suffers from debayering. These numbers are only estimates though. I don't want to hate on film - it's a lovely medium. It is fascination to divulge on the technology, but in practice it's just about what look one prefers. In technical applications digital is significantly superior nowadays, but photography is and remains art. Personally I shoot digital for various reasons and I don't care for the look of 35mm film, but medium format film I adore (and larger) - maybe I'll get into that some time ;)
@@ChrisThe1 I was just reading Tim Vitale's white paper on this topic (and remembered that I'd read it before) and reviewing what Ken Rockwell (love him or hate him...) wrote as well. In the end, this discussion seems like it misses the target (no offense to you, but in general) that digital photography and film photography are just different beasts. The sort of "resolution comparison" is the wrong question because I don't think there is a final answer because of the inherent differences. Both mediums (media?) produce images that can be very pleasing, or informative, or alluring, or choose an adjective. Particularly if film is wet printed (which is very different than digital printing of scans) there is an even greater difference between the two results. I think people want to say one is "better" than the other. But both are art and so much of art is subjective. I shoot both digital and film, although I am shooting more film now (35mm and medium format -- the latter is a blast). I love the convenience and instant gratification of digital. But I prefer the appearance of film (mostly) and I know this is going to lead me back to the wet darkroom again. Maybe this is nothing more than I grew up with film. I don't know... But I do enjoy the discussion with reasonable people. Thanks...
@@nvrumi Totally agree, they are both just very different and incredible in their own way. Although I haven't shot film in quite a while I have very nice memories of the wet printing process. Thanks for the discussion :)
I shot 35mm film in the 90's into the early 2000's. I gave up on film when i got a 10mp Nikon D200. Scanning film with dust, scratches, curved negatives etc. at home was not worth the effort any longer. For me, the benefit digital gave me was the freedom to shoot more, experiment more, and not needing to worry about the price of film + processing and hours of time scanning. I get the idea of people wanting to shoot analog end to end, developing and printing at home, but I honestly don't get shooting film to have someone process it and then scan it to then edit digitally.
Each to their own (preferences.) I was an early digital adopter, my first camera was a 2.1mp Oly and USED memory cost over a $1 a MEGA pixel. Film? Since 1953 since my dad was a professional photographer. Learned how to develop and print as a kid and teen. I upgraded cameras as the technology got better and I could afford them. That ended with my 8mp Konica-Minolta A2. That was about seven years ago and that's when I said to myself. "I miss film. It was a lot more fun." That includes making my own B&W developers from scratch and playing with my own formulas. I've also developed color neg film since the 1980's. Difficulty and challenges is why I like film. OTOH, inexpensive phones can do things within their SW that inexpensive digital cameras can't, based on recent research. (I was hoping to find a fixed lens, hand held camera under $500 to do what my old phone can do.) I won't go into what having super tiny pixels but a lot o them (64+mp!!) allows to be done by pixel binning, combining, and comparing. Instant single shot HDR. Expanded, full dynamic range. Things like that, right in my pocket. Not a perfect metaphor, but digital is like flying, film is like a good road trip. One gets you there w/o drama (hopefully!), the other puts you into the sights and smells and interacting with the world.
@@frequentlycynical642 That's a horrible metaphor. I can just as easily switch each other for road or air travel and neither format makes that type of difference. I can use by plane or car as the less drama method depending on what I want to say, but more importantly, neither film nor digital gets you *more* into "the sights and smells and interacting with the world."
@@JH-pt6ih Jeezuz, man, cool off. If you do your own chemicals and processing, you are much more involved than you can ever be in digital. If you will reread my comment, I have decades of experience with both. Do you? And it said it wasn't the best metaphor, so don't pile on me with your anger. All digital cameras are auto-everything. Kind of like letting a pilot fly you. While there were a few autofocus film cameras, they were horrible and never sold well. The most auto film ever got was auto exposure. But mostly, you don't have to prove yourself right to me or anyone else. Nor do I. Just let go and go have some fun, OK?
@@JH-pt6ih Get BTFO'd homie. What he's saying is digital photos can be fired off a hundred times a minute, all perfectly calculated, all properly exposed. Film means you slow down, take your time, enjoy the process. It's a different mindset. Trump2024
You're really only going to get 54MP out of those color films if you're shooting a black-and-white test chart (i.e., that's where you get that 1000:1 contrast ratio that gives you the 125lp/mm number). With actual real-world color subjects, the resolution's going to be more like that 50lp/mm number, which works out to only about 8MP, which is well below what modern digitals are rated at. Of course, all of this is really a red herring, since resolution is one of the least important aspects to what makes a good photo. As long as it's higher than about 4MP, things like subject, composition, lighting, etc are going to be far more important to making a good photo than just number of pixels. A well-composed picture of an interesting subject taken with an iPhone 5 is going to be a better photo than a test chart shot on a 16x20 view camera, and a well-composed picture of an interesting subject taken with a Pentax 110 is going to be a better photo than a test chart shot with 150MP Phase One IQ4 full-645-frame digital. The only sort of photography where pure resolution matters if you're taking documentary photo scans of other artworks or historical artifacts etc, which is what films like Adox are designed for. For anything else, you wanna go with whatever setup works better for you artistically.
Actually, you won't even get that contrast in a test chart. 1000:1 is 10 stops, close to the maximum dynamic range of film. Paper's dynamic range isn't close to that. Matte, as is used in most test charts is anywhere from 4 to 6 stops.
By the same marker, this would only apply to monochrome digital cameras or foveon types. Most digital sensors need a grid of 4 pixels to accurately record colour, while film does this through layers stacked on top of each other.
@allankcrain That's also my finding. My old Fuji Velvia 50 slides (scanned with a LS5000) have about the same amount of details as a 6 to 8MP digital image.
I'm in the process of scanning my family's photo negatives. There are at least 400+ rolls waiting to be scanned. Most are just typical family stuff, taken with anything from cheap compact camera to pretty decent SLR - but no award winning pictures here obviously. After conducting some test I've settled down the resolution to about 8.6Mp (3,600*2,400) - also in glorious JPG format too. This should be more than enough in this context. Scanning at 56 Mp / in RAW format would not only be waaaay too overkill, but extremely impractical so I won't bother considering that. Maybe just for some particular / important images, but not all of those 14,000 images - for sure.
Yap, just wanted to post similar. I have scanned thousands of negatives on Nikon and Acer scanners, and my conclusion that at the best resolution is max 12mp of color negatives 100asa. B&W ILFORD 400asa which we use to shoot while I was photojournalist is far worse than that.
Purely theoretically this might be correct. But real life useable resolution is much - MUCH lower. I digitised my film archive with DSLR scanning. First of all I have to admit - my film equipment was pretty average. Not bad, but not great either. I had 12 and 24 mpix DSLRs at that point and a vintage slide duplicator. As a reference for the slide duplicator capabilities I used film defects - scratches. I focused to achive best sharpness on them. After trying different ways to focus this gave me the best results. Contents of the film were much less sharp than these scratches and I couldn't achieve anymore sharpness improvement. I tried both my DSLRs and found that even 12 mpix gave me more resolution than I can actually extract from film without diving into counting individual film grains. I admit that if all the analog process if focused for maximum resolution - you can achieve noticeably better result, but that's not how an average film works. So for real life usage digital surpassed film long ago.
"Film Grain, Resolution and Fundamental Film Particles" by Tim Vitale pdf compares various color films, states avg line res 64 lp/mm that makes about 14 Mp for 35mm film, but that's line res, not color res, dye clouds are quite large. In Kodak Gold 200 35mm, I didn't see details worth 14 Mp. But it's important to understand that not every camera lens resolves too much, many good lenses only resolve something like 20-30 Mp, some cheap zooms cannot resolve even 14 Mp and only few lenses can resolve more than 50 Mp. Also for high-quality printing we often need only 2-8 Mp (that covers sizes from 15x10cm to 20x30 cm), for larger prints it's logical to drop the ppi because with greater print size viewing distance grows so 8 Mp is really all we need, no point to chase megapixels in digital or in film. P.S. Because of the Bayer's matrix in digital sensors you need more digital pixels to capture finest details accurately. In other words, 14 Mp digital pixels directly from digital sensor won't have 64 lp/mm due to aliasing/debayering issues. So while 8Mp is all we need, it really matters if these final 8Mp were obtained from 10Mp sensor or from 24Mp sensor.
@@vasyapupken that's what I say, yes. Lenses do not have infinite resolution, especially cheap zooms. For example Canon EF 24-85mm f/3.5-4.5 USM and Canon EF 55-200mm f/4.5-5.6 II USM when mounted on 50MP Canon 5DSR resolve details worth 12 MP max and ONLY if we close aperture to f/8 and ignore the fact that corners are blurry, have fringes etc. But Canon EF 85mm f/1.4L IS USM on the same camera is perfectly sharp even at f/1.4 with measured sharpness: 45 MP.
This theoretical analysis is very different from real-world photography, scanning and printing. I’ve been a graphic designer for 40 years, and have been using scanned photographs in design software for over 30 years. A high-res 8x10 scan from a $250,000 drum scanner of roughly 100MB was not nearly as good an image as I get from my LUMIX S5, which is 24MP. The files open up at about 115MB in Photoshop. The detail in normal professional 35mm or even 120 film cannot produce real-world results of a modern 24MP full frame sensor. Even my Canon 7D and 5D III produced better files than I ever saw from the service bureau scans. When I shoot film today I have to really reset my expectations for resolution and detail or I’m very disappointed. Still, there is a beautiful roll-off from specular highlight to the regular highlights on film that you don’t usually get with a digital sensor.
I agree, and I think the problem comes from overinterpreting MTF curves (which admittedly is easily done). I was a commercial photographer in the 1970s, and there is absolutely NO WAY 35 mm film could yield anything approaching the quality of hundreds of megapixels. We NEVER used 35 mm film for commercial work, and even medium format struggled to make the mark. We had to use large format, at least 5x4" but often 10x8", for the kind of work that is now routinely done with 50-100 MP digital cameras.
@skfineshriber Yeah I'm trying to figure out this you tubers results on thinking film gives more detail. I remember learning thst 9 to 12 megapixels is a goid range for the digital equivalent to film. Because film depends alot on how good the lens is. Cheap lens = Maybe 9 Megapixels. Pro gear abd lenses = 12 maybe 15 megapixrls if your lucky
@@SchardtCinematic Exactly, and in real life no-one used ISO 20 film. 50 or 100 for studio work, both in b/w and colour - and 200 or 400 for handheld, on-site jobs. Yes, ISO 20 for copy work and the like, but never for anything that had depth or moved. The thing about real photography is that it's a practical exercise, not a theoretical one.
This essay has one major flaw: If we take that Fuji 200, it only has 100% resolving power up to about 20lp/mm. A 1.3MP digital camera can match that resolving power. (There is some fudging with demosaicing, but let's assume a monochrome camera). Past 20lp/mm, the accuracy starts to drop. Meanwhile a digital camera is 100% accurate well up to the point where the film drops to 50% response. A 12MP camera is "better than film" up to that point. A 45MP camera will have a 100% response at the point where that film has dropped to 20%. So it's disingenuous to say that Fuji 200 film has a resolution of 54MP. It has a resolution of probably more like 4MP. Sure, it can still show difference between those values in some sense at a finer and finer scale, but a modern DSLR will wipe the floor with it at clarity, right up until the sensor's own MTF curve takes an abrupt downward turn into noise, somewhere around 115lp/mm.
A 45MP camera will not have 100% accuracy because, even if it does not have a low-pass filter, the MTF will be smoothed out by the objective lens. The same happens with the 1.3MP and 12MP cameras.
But given a scan of sufficient resolution, you can boost the film-typical poor contrast at the finest detail level that you describe as "less accuracy" and recover more of the inherent resolution. The same thing that a photographer did in the old days when printing on "hard" contrasty paper to make the details crisper and less soft, only much better controllable now in the digital domain. From a low-res scan however that throws everything that isn't 100% contrast (in your example, 20lp/mm) away, no finer details than 20lp/mm can ever be recovered. Astronomers (as a typical example) demonstrate every day that lens flaws and focus problems can also be reverse-engineered digitally if the input is sufficiently finely resolved, to get images that fully exploit, sometimes even exceed, the original specifications of the equipment and materials used. Yes, raw full-res film scans will always look "soft" and unfocused because of the gradual dropoff of that curve. Mirroring its characteristic (aka sharpening) in the digital domain can however bring back close to 100% of detail at any point along that curve that is distinctly above 0%.
Sure, a 1.3MP camera can match that but it won't do anything more, while the film will. 4MP is still a super low number to assign to a film, even one as boring at Fuji 200. I shot some recently, using a 88 year old lens that is losing its coating, and had it scanned on a Noritsu at 30MP as TIFFs. I see
There's another aspect of this comparison that is of great importance, and that is the spatial spectrum of the noise (grain or electronic) in an image. Film images will always have much larger noise energy out at the standard limiting resolution where 1000:1 contrast resolution target is reproduced with 20% MTF. By the way, the correct term is "modulation transfer function." The comparison between digital and film sharpness (not resolution) was addressed in the early 1950s by Otto Schade at RCA. He developed the technique of comparing the squared area under the MTF curves of the different media to find a numerical estimate of the apparent sharpness. There has been some argument since whether it's the squared area, just the area, or something in between that best estimates the visual sharpness, but the main principal of examining the area under the MTF curve. has remained. Essentially, higher contrast at mid frequencies (as in electronic cameras)can produce about the same apparent sharpness as higher limiting resolution with a lower mid frequency contrast and more gradual rolloff (film).
Excellent video. I am a boomer, working in the movie industry for a few decades now, so I really had to deal with film. My experience is that scanning 35mm film higher than 4k makes no sense because with the usual film stocks (Fuji, Kodac) you already have like 4-ish pixels per grain. But keep in mind 35mm film runs vertical instead of horizontal in photography, that means our negatives are closer to MFT than to Full Frame. On the other hand even though our eyes can't see much more than 2k, scanning a 35mm in 2k is still decently sharp but you can't recreate the original corn structure due to a lack of resolution, so you can spot it as digital right away. Watch the balcony scenes of De Niro/Amy Brenneman in "Heat", that was scanned in 2k and you can tell right away that it is a greenscreen scene, not because of bad vfx but because of a lack in resolution.
I just retired after over 40 years in Hollywood in post (colorist), visual effects, film restoration and HDR R&D. I just looked at the scene you offered in Heat. The wide shot just isn't a great comp. Possibly too much spill on them from the chroma screen. If they'd feathered the matte a bit it may have taken away some of the edge harshness. The singles are better. It's not a resolution issue. Being released in 1995 those comps were done optically. They probably looked more natural in the release print. I once had an issue with a feature from the early 2000s that had 8 bit digitally composed opticals that had been made into film elements and cut into the O neg. The entire feature was scanned 4K for a home video release and as one would expect, the scans revealed the extreme disparate quality of the negative versus the opticals. They probably would have looked as bad in 2K. I had to jump through a lot of hoops to match them to the rest of the show. The photochemical process covers a lot of issues that scanners unforgivingly reveal. When restoring old features, we had a process for completely removing grain and restoring it to whatever degree the client asked for. Our process could boost resolution of the image by sampling images before and after to add more detail. This was possible since the grain position is different in each frame of film. The finished deliverables were a digital intermediate film element and a home video version. The point is, we could get away with all of this stuff since we were using motion picture images. When you scan a still, you're just sampling the grain of the film...which is itself, the image. Unless you have a large format negative, at higher scanning resolutions you're just getting a better look at the edges of the grain. Sorry to be so long winded.
As a professional photographer, I shot film for 40 years until starting digital photography in 2000. My experience was, in the early days of scanning film on a Kodak 35mm film scanner, (about the year 2000) that if I made a print from a 35mm color negative onto Type C paper, in the darkroom, lets say for an 11x14 print, I would get a beautiful, smooth toned, relatively grainless image. If I scanned that 35mm negative, and then made a print from that digital file, on a color printer, I found that the scanning would bring out the grain. Instead of a smooth grainless image, I got a printed image that was full of grain. I am guessing there was something in the wet process and the combination of the color negative film and the Kodak Type C paper, that made the grain so much less. When the Nikon D2x digital camera arrived, and I did some aerial photographs with it, I discovered that I got a a beautiful, smooth toned image, compared to shooting negative color film and scanning that! I hope this adds some information to this discussion.
I no longer do detailed posts on this kind of thing, but a lot is not covered here: RMS values for positive film, specialty films which required meticulous processing- notably Tech Pan film in HC-110, and Gigabit film in its dedicated chemistry. Use of pure azo dyes in now defunct Kodachrome 25. Tricolor (trichromatic) process making 3 identical exposures in black and white through a 29, 47, and 61 filter, then converting them to a single color print. A LOT of variables. Few people know about Linhof tested lenses. At one time, Linhof had agreements with Zeiss, Rodenstock, and Schneider to buy lenses, but then to only keep a % of them which tested best. They found 30% variations in the SAME make and model of lens, as sample variation. Personally, the best lens I have ever seen was a Voigtlander Apo-Lanthar in 105mm, although some process lenses are probably better. That Voigt was a different company than the lenses you see today, which are quite good, but made by Cosina.
Great video! It’s worth noting that in audio, a higher sample rate than 44.1kHz doesn’t actually result in any greater perceived audio quality. Only an increase in the maximum frequency which is able to be captured, but as those higher frequencies are outside the range of human hearing they go completely unnoticed. This can be useful if those higher frequencies are going to be shifted down into audible range but otherwise it won’t make any difference aside from larger file sizes and greater CPU usage.
@@TheScrubsfanforever I’ll expand: I do audio engineering part time on top of being a musician and I spend a lot of time listening to high res audio through studio gear, so it’s possible my ears are attuned to it
@@EntonationStudios ah nice. The Anecdotal evidence. Fun fact: I studied that shit 😄 You only need higher frequency’s if you a) don’t use a low pass filter, or b) if you somehow slow the audio down There is probably no human on earth that can hear frequency’s higher than 22Khz, maybe a few newborns
Nyquist says you must have GREATER THAN twice the frequency. If the rate = the frequency you get a constant value that is arbitrary. As you approach the Nyquist rate, the number of samples you need tends toward infinity.
Thank you for this very interesting video. My experience with the Fuji GFX line is that while the sensor is not as large as a 6x4.5 film image, it is substantially larger than a "full frame" digital sensor, and produces fantastic, highly detailed images at a very reasonable price.
This is the clearest and most concise explanation on the subject I've ever heard. And I've heard a LOT in my day. Even if film didn't meet or exceed digital in the technical sense, I'd still shoot it for the beauty and fun of the process.
@@jaimeduncan6167 You're right. It would be interesting to see a photograph of checkerboards to illustrate this point. Basically if you compare a checkerboard with pure white and black squares (TOC 1000) to one with light grey and dark grey squares (TOC 1.6), and you have amazing lighting, if you move the checkerboards away from you as you take a bunch of pictures, you will be able to clearly see the black/white checkerboard from much further away. The grey checkerboard will turn to mush as soon as its size at the film plane reaches 0.25mm whereas the black/white one will do that at 0.08mm. So despite using the same exact film, the same exact size checkerboards, the same exact image, you will be able to clearly see the black/white checkerboard from over 3x further away than the other one.
@@jaimeduncan6167 He mentions that these measurements are theoretical and ideal, and the reality is dependent on all the systems between the media itself (film or sensor) and the viewer (optics, scanning, enlarging, monitors etc). At any rate, it doesn't really matter, it's just a mental exercise. We choose film or digital for subjective reasons
Interesting video. But from a practical viewpoint, you will often get a sharper image from a digital camera. With film. It depends on what film, how it is developed and scanned. Most of the time we don't use drum scanners. Too expensive. And film is best at low ISO. Where digital will often give great results at 1600 and higher speeds. I remember my first 5 megapixel camera. I had doubts about the quality. I had a model come in, shot a few photos and printed an 11x14 inch print on an Epson printer. It looked great. As good as from my Nikon film camera using color negative film. My current Fuji 26 megapixel camera makes great 2x3 Foot prints. Much better than my old 35mm film cameras. Why shoot film? It has an artistic quality. I even like the grain I get. It seems more authentic. But technically digital passed 35mm film up years ago. And is competitive with medium format film and even approaches 4x5 film cameras, at times.
One thing to not forget is the scanning process when using a scanner like a Nikon 9000ED will yield colour information for all three colours in the RGB spectrum and at 64Bit including infrared information. That’s 16bit per colour and all colours present in one pixel instead of a interpolated Bayer pixel which offers 16bit max for one colour channel in one pixel, but most often 14bit unless a highend medium format sensor. It’s just like how Foveon sensors capture more details per pixel.
And still those sensors blow away the crappy chemical spec based celluloid carrier, film is based on. How could it be otherwise, sharpness in film is just luck, because of the perfect spreading of the chemical particals needed which is off course never the case. And it shows there`s always some crappyness about it, and grain lots of grain and looking at it is uneasy on the eye. And we`re not even talking dynamic range.
Microfilms were used a lot in the past by libraries or mechanical workshops to store a lot of manuals and information. The resolution was incredibly high. I wonder if it compares to the resolution of small image sensors on phones for example.
Back when digital cameras literally were first coming out, I read a pretty well researched article in one of the photography magazines. Unfortunately, i can't remember which one or exactly when. They said that 35mm 100 iso film would be the equivalent of 24 megapixels. And this was back when 24 megapixels was a pipe dream. Their calculations were based on the actual visual information that a film image can contain, counting only the level of resolution that actually contributes to what the human eye can discern as part of the original scene, and what optical equipment can convey. In other words, a 24 megapixel image at 8x10 would look indistinguishable from a 35mm 100 ISO 8x10 film print... For most people in most normal conditions. Those are the numbers that I have been going by. My Sony A7R2 is 42 MP, and the a6700 i want is 26 MP. So I'm happy.
I am a photographer who started in 1975, we used very fine grain low ISO/ASA B&W film and that was extremely fine grain. Shooting reversal/slide film gave very fine detail and we used every format upto 5x4 in the field (sometimes 10x8) . My last proper use of reversal film was to photograph camouflaged military equipment at varying distances and height from a helicopter to be used for aerial recon training.
When you send in film to develop in a shop, most of them only send you 12mp images, and even then I often can see softness in the pixels. Looking at that MTF curve, only at 20lpmm can it resolve 100% contrast, but 2 digital pixels can resolve 100% contrast all the time. Accounting for lines not alined with the pixel grid let's just say 4 pixels required for the equivalent of 20lpmm. This gives us a digital resolution of 2880x2160 or 6mp, a much more reasonable number. I mean just look at the curve at 120lpmm it has 20% contrast, that doesn't really count. I will say under the most ideal circumstances, the best 135mm film photo I've seen, can be reasonably scanned to 24mp. I've never seen a 50+mp scan that doesn't have serious sharpness issues at the pixel level.
This answers a lot for me: I only recently got into a serious digital camera a couple years ago. I played with it for a while, but just was not happy with the images, and soon set it aside. Back in the day, I used to shoot Panatomic-X which was ASA 32! And, I got amazing images that I could easily enlarge to 16x20 without any serious degradation in the image. When I would shoot it on 120, you could project it across a room and not see anything that looked ‘too-far!’ Anyway, I’ve learned that you have to settle for limits, and be happy with the current state-of-the-art; otherwise you’re pretty well stuck with accepting what you get out of your phone… 😢
Spent quite some time testing our and other manufacturer's digital dental intra-oral sensors and film for our competitor comparison when I worked in dental radiology ... we had a (very expensive) test pattern to shoot giving us the "resolution" proof
In audio Nyquist theorem assumes that there is no destructive effect from having to brickwall filter aliases above the chosen bandwidth - but in the real world there is. In the same real world, there is no way that a 35mm film has the resolving power of an MF digital image.
These numbers are very interesting, but I think also very wrong. My father had a specialized high quality photo lab ("photo finisher") for professional photographers and customers who needed enlargements for trade fair show wall-sized pictures. We actually did have one photographer who shot on 8x10" - and we were the only lab in hundreds of kilometers that even had an enlarger who could take these huge slides. I say slides because this was the reason the photog shot this big: to show his work to the customer without any other party being able to mess up. So even in the 1990s there really was no good reason to shoot film this large. Bread and butter was 4x5" and 5x7" for high end and most was shot in medium format. None of those still shoot film professionally. Even 10 years ago most went digital in 24x36 and when the first medium format cameras/backs came out large format was almost instantly dead. So digital already surpassed film in quality and I would say by a large margin. I guess the misconception revolves around the contrast: Of course you can resolve B&W test charts fine, but real world pictures are a differnt story. They don't have contrast 1000:1 usually.
This was supremely understandable, while also being sufficiently technical. Prior to watching this, I'd not considered how the 'resolving power' of a film format could be affected by micro-vibrations caused by moving camera gear (shutter, etc).
One of the best content I have got on the internet on this topic And the most informative, arguments against it in the comment box Thanks a lot This settled a lot of debate
If you get down to the molecular level, film is digital, a silver crystal can be activated or not activated. while a digital sensor is analogue, it measures the amount of light activating a photodiode. but since pixels are much larger than film grain, and the photodiode is immediatly converted to a digital signal it makes sense to call it digital. Just to confuse the conversation. lol
And why anyone with experience scanning will modulate the scan resolution dependent on the stock/processing, scan delta 3200 pushed at 5400DPI and it's literally black and white, there are no greys.
You're confusing film grain with individual silver halide crystals in the film emulsion. What you see as 'grain' is the aggregate of thousands of crystals. So, no, grain is not digital.
@@emotown1 A silver halide molecule has an electron exited to a higher energy shell when hit by a photon. This silver halide is either in an excited state or not. it becomes soluble when developed, or not. On or Off. (yes this is simplified, but is essentially true. I went to college for photography back in the film days)
Anywhere from 9 to 12 Megapixels is considered 35mm Film Quality. Like you said you really can't define film as Megapixels due to there being so many diffrent grades of film. Plus it also depends on the quality of the lens you used.
It’s a relief to know that even a camera phone these days will give a highly satisfying image and pro-sumer cameras are good enough for any professional, and pro-level cameras are more about personal taste and work flow than image quality.
There's something that I have never seen addressed in these videos, and is the fact that, AFAIK, each grain in a film can only be either black or white (or, if it's a color film, either the color or black), which means that any light/color gradation must be done by dithering. So yes, you can have very small lines for very high-contrast elements, but as soon as you have normal-contrast elements, the image dithers and you loss a lot of resolution (which, BTW, explains the two values of lines per millimeter). Instead, in a digital sensor each pixel (or subpixel in the case of three-sensor cameras) can have 256 or more levels of intensity per pixel, so although the "apparent resolution" (this is, the number of line pairs per millimeter) is smaller, in practice, with "real world images" the resolution loss is not so extreme. Or that's my opinion.
Tmax 100, is nice and sharp, but it’s also very high contrast. Ilford 100, developed with tmax developer is much nicer, because the gamma curve is softer. More subtle and intricate detail if you expose it right, and develop it right.
This is a huge and complicated topic, and not sure if this vid adds much of value. And important to note: colour depth has a big impact on visual impact and enjoyment, witness the trend upwards from 8bit digital representation (whether in the camera or scanner) to 12bit and higher.
You tend to get the same argument when you compare analog sound vs digital. In both cases I tend to prefer analog but I use digital for the convenience of it
Has any film camera manufacturer even considered using an LCD as a shutter? Or would the polarizing filters on them affect the image negatively? That could help basically negate shutter movement distortion...
I think something like that has been patented decades ago, and shutters based on polarized filters have been used in many high-speed cameras, including the ones used for capturing nuclear explosions
chatgpt explain : Here’s a more detailed summary of the idea of using entropy in the conversion from analog film resolution to digital formats such as HDR10+, with some scientific insights and formulas: Entropy can be used to measure the amount of information when converting analog film resolution into digital formats like HDR10+. In information theory, Shannon entropy quantifies the average number of bits required to encode information, which is essential in image processing. For analog film, resolution is often measured in line pairs per millimeter (lp/mm). To estimate the number of pixels required in a digital format: P = L × H × (lp/mm)^2 where L and H are the film dimensions in millimeters. In HDR10+, each pixel uses 30 bits (10 bits per color channel). The total amount of information (in bits) for an HDR10+ image is: Total bits = P × 30 For example, a 35mm film with 50 lp/mm results in ~2.16 million pixels, needing: Total bits = 2.16 × 10^6 × 30 = 64.8 megabits Entropy plays a key role in determining the visual information and optimizing conversion from analog to digital formats. Higher entropy in film (due to grain and tonal variation) typically requires more digital data for accurate reproduction.
That 54MP is wrong based on the numbers there. The 9000 x 6000 dimensions each need to be multiplied by √2 to make sure the diagonal is capturing 250 pixels / mm. 12728 pixels x 8485 pixels = 108MP, not 54MP. Edit: OK they did briefly acknowledge that
My go to reference for film resolution is that Super35 film can produce 4K images. We've all probably watched movies shot on Super 35mm and displayed in 4K. Film has a very soft frequency rolloff, so what this comparison is for is less for maximum resolution of film and more for the vague perceived sharpness. The image is slightly softer than digital, but the higher frequency details are definitely there if it's shot on a sharp lens. Film will capture much finer details than that providing the lens is sharp enough and in focus, but I use that as a basic sharpness comparison. And this is fully anti-aliased. But it REALLY depends on getting a good scan. To get an image WITH NO ALIASING IN ALL CHANNELS out of a Full Frame Bayer pattern sensor, matching that resolution (4096 pixels per 24.89mm along a single dimension), you'd need a 187.MP sensor (double that pixel density then multiply it by √2 to get the diagonal*). That's higher resolution than i think most of the digital medium format options. You wouldn't need as many pixels with a hexagonal arrangement, but with bayer pattern sensors and an OLPF that abides by signal sampling theory that is what you'd need. For a more subjective comparison, you can take a decent quality (not cinema quality) full frame scan, and you compare it with a 60MP full frame photo, you should be seeing much sharper fine details in the digital photo, if your scan even picked up the fine details at all on the film photo at all. But the fine details in digital are not pleasant - they have a bunch of aliasing and noise reduction and de-Bayering artifacts. Manufacturers could apply an OLPF to fix this, but they don't, in favor of sharpness and maximum resolution, presumably for competitive reasons. * Arguably if I am comparing scanned film at that resolution I should use the diagonal for both, meaning not multiplying by √2, but: a) 4K images often contain aliasing either from the scan or the downscaling filter used to make the picture sharper, which I would argue is bad, but we are just using it as a resolution comparison here, and b) as I said film definitely can produce much higher maximum resolutions than this benchmark of vague perceived resolution anyway (as seen in MTFs on the datasheets), while digital has a hard limit on resolution. For reference without using the diagonal the full frame sensor would need to be 94MP, which is your absolute minimum full frame film resolution (well, probably not YOURS since most of us shooting film are using soft vintage lenses and not using cinema quality scans).
This is consistent with another UA-cam video I saw where a movie studio basically said there is no point in rescanning old movie film at resolutions greater than 4k as the detail isn’t in the film start with. So these remastered 4k movies will likely be the best we get (at least without AI).
Honestly the only way to even approach the sort of resolution that 35mm is theoretically capable of (digitally) would be shooting with modern Canon L glass at like f/8 and then drum scanning.
Absolutely lovely video but one note: The H6d-400c doesn't "fake" 400 megapixels. It shifts the sensor by by subpixel amounts mimicking the pixel density of a 400mp sensor. It also fixes the bayer problem of having 2 green pixels, for everyone 1red and 1 blue. By shifting the pixels around every color can be represented as 1:1:1 which makes for much better color reproduction and improved dynamic range like a film camera. Looking at a 400mp out of the Hasselblad vs the 100mp you can clearly see the increase in detail (not just sharpness)
It's "fake" in the sense that the sensor is only 100mp and effectively can only take a stacked array photo off of a tripod or other steady position. No one would take Apple seriously if they claimed the iPhone took 600mp panoramics (via stitching 12 photos in camera) so Hasselblads 400mp claim should be taken with the same grain of salt, IMO.
Pixel shift is a gimmic though. I’ve tried using it in perfectly still situations several times and it never produces a better image. And with the technical limitations of the process required it’s a useless tool. Better to stitch together 4 images at a more zoomed focal length
@@andytheturtle87 I mean Apple sells the new Iphone with a "48mp" main camera when in reality is a quadbayer sensor which is only 12mp in the bayer sense and it uses Ai Upscaling to fake the 48mp look. And people have seem to have accepted that in droves
@@tdawg719 I've tried pixel shift on a GFX 100 and my X-H2. And both have produced incredible shots well above the 100mp or 40mp respectively. I use it for art preservation and PCB shots. The trick is to use a Mac, Fuji's app on windows just flat out doesn't work well and even when it says it finished the image, there will be no different except in file size whipe it works much better on a Mac.
I compared the resolution of negative scanning quality between my 2 Fuji cameras - 16MP X-T1 vs the 24MP X-T30. There is easily detectable difference and I have dedicated the X-T30 body for that duty. I use a Nikon PB-5 bellows with PS-5 slide copying attachment. To that I coupled a Fuji-X to Nikon F adapter and a Zeiss 85mm f1.4 lens. Works like a charm and with live view and focus peaking you get spot on sharpness. Just need to have a good and uniform white light source.
Probably the most extreme example of lens softening I've seen is on the Lubitel 2, at f/4.5 it's surprisingly good if vignetting however f/16 and f/22 are very soft.
yeah film is uh... haven't touched one since the 90s >be me >shoot with my a77 using retro but great Minolta lenses all the time, and a minolta badge over the SONY logo for maximal swag ( flip screen folded inside cause i don't wanna hurt it accidentally) >hipsters around me believe I shoot film >praises me because on photoshop I do edid the contrasts and levels when the full auto exposure breaks the dynamic range >«See, that's real art, people who use a digital can't comprehend this, I want a print of that random shot!» >just get it printed on a plotter from a .psd this is as far as it gets, eventually for large prints i upscale it for the support and put a light random noise effect
Resolving power is an attribute of the lens. The appropriate term for film is accutance. It does not matter how many times this distinction is overlooked by UA-cam videos. It will always be contradicted..
Yes, same finding here. I scanned my old Fuji Velvia 50 on a Nikon LS5000 (4000dpi resolution = 20MP output) then carefully downsized it to a few versions and compared against the original scan. A 6 to 8MP image captures all the details of the original scan. I used prime Nikkor lenses at optimal apertures.
@@bernardd 1. all Nikon scanners have maximum resolution of 3600lpi regardless of "wow it 4000dpi optical" written on a box. 2. this resolution (3600) achievable only with high contrast test target, not a real photo 3. you say "prime nikkors" like it must mean something ) but no. vintage Nikkors are mediocre lenses for the most part. especially in terms of resolving power. fun fact - one of the best resolving Nikkors is one in their film scanners.
let's be honest - your opinion based on experience with poorly shot and scanned film where you did nothing to achieve max res. 99% opinions in "film community" are like "i scanned it with my Pooptek-9999KHD it''s 9999K optical res so where is my details ??" it is quite hard task actually to scan a film properly for it's max res. wet drum scanning is the way.
@@vasyapupken "Prime nikkor" simply means a fixed focal length lens (not a zoom). In the film days, most zooms used to be inferior optically to prime lenses. Modern lenses have made a huge jump in the past 20 years but I don't think a good "prime lens" from the 1990's shot at optimal aperture would be the limiting factor with 35mm slide film. That's an educated guess though. If anything, I'd say the Nikon scanner is the limiting factor, although again the LS5000 was one of the best non commercial scanners for slides, not counting drum scans. That's my feedback anyway. I'd love to find a way to extract more details so these comments are really interesting.
It seems that there is a similar analysis and debate in the digital vs. vinyl war. You can't directly translate between the two, but the reality is they're both perfectly acceptable formats that can be of high quality, so high the average person can't really appreaciate the difference. As you say, there is also a limit to human perception, regardless of the source or format; there are only so many cones and rods in your eye. I deal heavily in various media, and even though resolution is imporktant, there are other factors to consider. In digital cameras, I have found that the quality of the light sensor and lenses are more important than the number of megapixels it supposedly capable of recording. Anyway, thank you for the interesting analysis!
Hi ShyStudios. The material in the video is very intellectually stimulating indeed, regardless of the many comments. Kudos to you for not having censored any of them. Even if i am in no way an expert I'd like to add my two cents about the lens resolution as it seems to be common knowledge that most lenses reacht theyr maximum definition somewer between aperture 5 and 11 what means that at aperture 2 it is going to be worse than at aperture 5. This somewhat invalidates your statement about the line resolution of a lens being roughly 1600/aperture.
If film has so much resolution, where are all these hyper acute pictures? Why do 35mm lenses that were A-ok for film fall apart when used on digital camera, yet the reverse is not true? When you look at photo books that had 35mm or sometimes larger format color shots from the 80s or 90s, they seem technically archaic, they are obviously falling apart on double page spreads, for instance. These calculations may, in some way, be scientifically correct, but practically the resolution is nowhere near claimed. And it’s not a matter of using vintage equipment today, or of transferring shot film using digital scanning. if you go back and look at film shot when it was the state of the art, you will not be impressed by its technical qualities, certainly comparing same format to same format.
Don't forget that all those old photos were reproduced without any sharpening. Imagine if you took your 15 MP digital sir and made an inkjet print with all the sharpening turned down to zero I/e just printing the RAW file with no processing. That is digital's strength though - sharpening up fine detail even from a 5MP camera is child's play. Which get's us into the topic of "apparent resolution" aka "sharpness", which wasn't tackled in this video. If it had been, nowadays digital wins every time and without breaking a sweat. In fact I think many digital photographers tend to over sharpen their images, but I guess that comes down to taste. And I'm a film enthusiast, by the way! And you are right, digital sir's do expose deficiits in old (film era) lenses in a way that film never could.
@@emotown1 Nonsense, sharpening up images is very visible and not done with quality capture. Raw files are mostly quite soft but retain way more detail than argaic film pics. Besides 15 mp is nothing these days it`s more like 30 to 40 mp, sharpening is not needed.
Lenses for film doesn’t have to worry so much about hitting the sensor at a right angle. That’s much of the reason they don’t work well on digital. Photobooks made from digital look terrible. All the particularities stick out like a sore thumb. Waxy and missing micro detail, bayer artifacts which is mostly lower colour resolution and poorer tonality, harshness from edge enhancing during demosaicing and poorer dynamic range which results in highlight clipping.
@@Frisenette All you`re mentioning has been resolved a long time ago, you sound like a broken down record. Analog sucks in quality and reproduceability big time, analog is grainy, has very low dynamic range, poor color fidelity, is extremely cumbersome an expensive. In short, it belongs in a museum. That`s not my opinion, that`s everyone`s opinion. Every picture you see everywhere has been from a digital source how about that.
@@brugj03 these issues, and more, hasn’t been resolved. They are fundamental flaws of the current sensors and algorithms to low level process the sampled data out of them. Digital, as a viable means of photography has only been with us about twenty years. All the rest of photography is film. The most beautiful and profound work in the realm has been done in that medium and continues to be made. In museums in photo books, posters, small and large. Grain, in what shape and form it might take, is real. It’s like the texture of paper, like a fingerprint. It’s a structure from the physical world. And our brain recognizes it as such. Not so with interpolated mush, manufactured and guessed at detail or various kinds of aliasing and clipping.
Lenses are the biggest limitation. I think digital has driven lens technology improvements, past "quality" lenses obviously fall apart with modern sensors, so to get the sharpest digital pictures, they are making technically accurate lenses. I've got a few older lenses that were very expensive back in the heyday of film, and they do not acquit themselves well in front of a digital sensor. Fortunately we also have digital means to correct deviations, but the old adage of Garbage In Garbage Out still applies.
Your resolution vs F# formula is only true if ignoring lens aberration. At fast F# more light is gathered further away from optical axis and aberration will have much significant effect. In practice there is always an optimum F# for a specific lens, usually one stop slower than the fastest F#.
Well, given how superior the film is and how inferior modern digital sensors are according to this video, I kindly ask for the same scientific explanation why in real world film photos are looking so rough and unclear, while pictures even from a smartphone looks crisp and sharp.
Like he said at near the end of the video, modern cameras have better optics and most have some sort of stabalization, to help create a sharp image. Having said that, when I compare the enlargements I have of my 6x6 Mamiya (80mm lens) with large prints from my Nikon Z7 (85mm prime) the prints of the Mamiya are only comparable when they are shot with low ISO film, colour film has a grain size per colour layer and these layers are not alligned, so I think that is a mayor reason for the overal fuzzyness of clour film enlargements, when viewed close by. Also the optical quality of the lenses is very good nowadays: If you compare lens sharpness over time in history you can see the steps they made, better calculation methods, better milling and grinding and polishing, better coatings etc. I have two 35mm cameras from the 1930's one has no anti glare coating the other has, the contast in the photographs is very much better from the camera with the coated lens and this helps in percieved sharpness.
The video pretty clearly says that you shouldn't expect theoretical results in the real world due to optics, old cameras, not truly linear film planes, ect. It's also with noting that phone photos almost certainly have AI sharpening and denoise applied automatically, so it isn't a fair comparison unless you are viewing raw files from your phone.
@@andytheturtle87 Fair point. However, I often look at raws from my FF mirrorless camera and I've never seen the film photo come at least close to this level of detail
@@PointlessDrummer There is videos about it. No difference between 24mpx and 60mpx scan on 35mm film, you can't get more detail than 24mpx, probably even 12mpx
Back in the late 80s and 90s, analog cable had a grainy look which really was just noise from a weak signal. For reasons I never understood beyond the buried mainline into your home, they used RG59 coax cable to wire your home. It's okay for short hookups, but they strung this crap on the outside of the house under the eaves to get to different rooms. In the early 80s, they actually just brought the buried cable straight into the living room as only one room had cable. Grounding was doggy if they did it at all. They put a ground strap on a painted water facet. LOL. The picture was clear, except for static during thunderstorms. So, HD blu-rays of 35mm movies show the grain which remind me of analog cable noise. It's not exactly the same. Film grain is much finer, and I understand why people would want to preserve that. I also didn't get to go to the movies that much when I was a kid, so I guess it doesn't mean that much to me. We usually just waited for it to come out on VHS, and VHS couldn't resolve film grain. When I rip DVDs and BDs, I filter it out because it compresses better since video compression mostly updates changing areas of the picture. Do what you want. That's just me.
I really wasn't expecting such a thoroughly-researched, high-quality video! Most videos on this topic just repeat stuff you can find in a single google search; great job! I hope you make more film stuff :)
Science vs real, practical use. My experience has given me great looking 8x11 prints from a 5mp camera in the early days of digital and I'm more than satisfied today with the 20mp cameras I now use for printing tabloid size and much larger prints. I dont know where all this resolution science comes from but I doubt 35mm film has as much real world resolution as stated in this video, at least in comparison to what I've seen from my film days. It may be because I didn't have good gear when I used film way back when?
Wow that’s wild!! Thanks for doing the research and sharing. My happy with my APSC 40mp. Maybe someday I’ll hop into film. 😊 and I’m glad to know I won’t have to worry about quality that much.
A pixel will never be smaller than film grain, grain is non directional unlike pixels in a grid. You will always need more pixels than the two for anything at angles to match film. I once took a picture in Germany back in the 80’s that had a small plane visible flying in the distance. It was very small in the photo and looked very nondescript. I used a Contax RTS2 camera with Zeiss T* lenses and the film I used was Kodachrome with a very low ISO (25?). Anyway, I got curious so I looked at the original slide under a microscope and could make out the planes color scheme, the numbers on the fuselage and the word Osterreich on the wing underside telling me it was Austrian. At that magnification it was very grainy but still legible. I was gobsmacked at the time. Film really is incredible compared to digital. Of course it’s inconvenient, expensive and more difficult to use as well as more time consuming, but little can really compare when you need the max.
I have lots of film negatives from the 1950's when my dad had a 70mm camera. I have a black and white 70mm negative that I used a Cannon desktop scanner to scan the negative at 9600dpi. A car in the picture that is more than 100 feet off into the distance, you can enlarge the scan of the negative enough to easily read the license plate of this vehicle 100 or so feet away. I highly doubt you can do that with digital photography.
Why stop where the making proceeds: the technique of the enlarger in the darkroom! It is important to understand that this step also reduces the resolution because here also a lens is involved. So with historic analogue testreports showing that lenses with Technical Pan often have a max. of 130 line paires/mm (prime lens, center, 80 in the corner), the real result comes from multiplying the steps in the chain. At the end approx. 50 linepairs/mm remain, which explains the turnaround when this became available in digital photography. Moreover the incredible ISO-performance in digital photography is a revolution. I remember testing the "gamechanging" Konica 35 3200 film. Now what a mess it is compared to Sony A7Riii at 25600!
It gets even worse when it comes to optical magnification (in the darkroom and elsewhere) as the effective aperture (and potential imperfections in the optics) further lower the resolution of film. Most magnifications in the lab are done at f8.0 max, f11-f16 are more commonly used. Wider apertures are only used for focusing the image. Now factor in the magnification and you'll end up with really small effective apertures, barely enough to resolve the negative. It's like spreading butter on bread, getting a bigger piece of bread won't get you a thicker spread of butter.
Is it right? Is it wrong? Lots of discussion. But most of us can agree that film can absolutely hold its own in the current day. If you have a good lens and reputable film, the resolving power is quite surprising. I've been shooting digital for the last 15 years and my only film experience was disposables in the 90s. Recently tried film for that "grainy look", but quickly realized it's actually harder to achieve that than you might expect. Anything under 400iso is impressively sharp. Since then, I've almost exclusively shot film. It's just more fun to work with.
Resolution or not - I like the process of film, I like projecting my slides and be taken away by large projected and pixel-free image. I can't help but notice digital suffering from color gamut issues or what's the correct term. You can even see it in the sample you provided at 11:05 where colorful cloud highlights in digital are blown, whilst retained beautifully on film, and I find the digital performance in this regard strongly lacking or even ugly. And shadow detail too - shadows are naturally blue from the sky and it shows on film.
In the mid to late 90's I started getting some of my better photos scanned to the now defunct Kodak Photo CD's (NOTE: not those low res Kodak Picture CD's you goy at Walmart) They gave you a 6mp image and could fit 100 images per CD. The Pro Kodak Photo CD gave you a 25mp image and could only fit 10 photos per disc.
7:14 My guess is that you would start hitting the resolution limit of the lens you're using at that point. EDIT: Turns out that you mentioned this at the end of the video.
I spent 22 years in the photofinishing business. Analog of course, the argument might be is film better than digital, I shoot and appreciate both for different reasons. However if you scan film with a DSLR or a dedicated scanner you are not using film as it was meant to be. It was and is designed to be optically projected onto photographic papers. I worked in an enlargement lab and I was able to see absolutely amazing images shot on 35mm film and printed as enlargements. I am guilty of scanning my own film images with a DSLR because I love the process and the film look. As I said I also shoot digital with a few different cameras. I guess my point will be shoot with what you like, there is no wrong way to enjoy photography, to each his own.
I have a film to digital system that uses a Sony a6000 (24MP) camera, a bellows, a set of Nikon enlarger lenses, and a dichroic colorhead -- you get the idea. I'm able to see a bit of the grain structure of 35mm film as I convert it. But certainly not the amount of grain I'm able to see with the critical focus scope in my darkroom. I can certainly believe that if I double the pixel count in my camera I will begin to approach the information density of the film. Of course, that's nowhere near the number of pixels I'll need to properly do 120 film. The math is obvious.
But actually the whole system is diffraction limited. In case of 35mm it is somehwere around 40MP. In case of digital medium format (such as gfx) its diffraction limit resolution is around 75MP, and thus justifies the use of 100MP sensor. Even if film itself can have such high resolution, at the end of the day your camera system is diffraction limited due to the size of the film.
After scanning pics with my 24 mpx camera this past winter I really think the limiting power of the lens is the factor. I was scanning everything from 110 to 4x5. Back in the 35mm film days the lens were really pathetic especially wider angle ones used on scenic pics. Just really yuk. Give me my digital camera and modern lens any day. I was using Mamiya medium format too-I would say that they might be equiv to my 25mpx digital camera but one has to get the film developed plus scan it and clean up the dust etc. Just not worth the hassle.
A lot of the confusion in the comments here is basically answered by the last part of the video. Though anyone claiming that 35mm negatives only have 8-10MP of resolution doesn't really know what they're talking about. Many commercial film scanners targeted 4000DPI, and they did so for good reason: given the average resolving power of the film and lenses available, 4000DPI will resolve most of the detail in most 35mm negatives. That being said, there are exceptions. I have seen enormous prints from 35mm black and white negatives, maybe 4' x 3', that most people would guess were from 8x10 negatives. With very high resolution film, a lens with a high resolving power and a still subject even 35mm negatives are capable of achieving incredible detail. When we start talking about medium format and above you get into the territory where even a standard negative out-resolves even the best professional digital cameras, even on color negative film which tends to be the lowest resolution.
It always the same with these videos, posts, etc. Yeah you can technically scan a negative at an arbitrary resolution let's says 80 Megapixels but you are not obtaining any real detail. For color film typical resolution is around 6 Mp and AT BEST 12 Mp and even there we are talking slow film, a sharp lens stopped down and no camera and subject motion. I have to use at least ISO 3200 in my 15 years old D3s to start losing the same detail compared to an ISO 50 slide film! Don't get me wrong I love film, but resolution is not it's strong point 😅.
It’s important not to conflate the resolving power of film with perceived sharpness. The contrast of low to mid spatial frequencies play a critical role in perceived resolution, resolving power plays no role at all, as it’s typically beyond human visual acuity at regular viewing distances. Color negative typically employ development inhibitor releasing couplers to organically produce an unsharp mask on high contrast edges to boost acutance, despite this the cyan record often drops in modulation at very low spatial frequencies, so perceptually will appear lower resolution compared to say a digital sensor even if the resolving power of film is greater.
Interesting point of view but also, I think there is a practical and historical reason. Those of us who lived through the "painful" transition from analogue to digital photography saw the first digital cameras as clumsy, judging by the first results one could go to bed and sleep peacefully knowing that the film had a long life. But soon we began to see results from cameras that with 4MP already looked very good, even in large enlargements. And let's just say that with cameras like the D700 or Canon 5D there was definitely no point in hanging on to film. The resurgence of film has to do with reasons other than resolution power or colors. And I have the doubt if it is purely a whim of fashion or there is a certain issue of being fed up with the already disgusting manipulation that photography suffers in the hands of editing to impress.
@@mistergiovanni7183 I agree! People tend to go back for obsolete things like film, vynil, analogic watches, etc maybe because what we have now is "too perfect".
Resolution CAN be its strong point if that's what you're after. It takes specialty products and processes though. On the extreme end of HD analog, you can pack 30 portraits onto a drop of emulsion just one square mm big. A percieved resolution of around 1 - 1,5mp per square mm. No digital sensor can match that.
While an excellent explanation and thoroughly researched, I believe this would only hold true in theory. Even in instances were 35mm photo film was used to photograph test charts, true resolved resolution will be much less than you claim.
Great video - thank you. The "resolving power" of film vs digital is a question that many photographers have asked, but received different responses. This video does an excellent job answering the question.
I scan my negatives with a macro lens. 8x10 can resolve about 1000MP maybe a bit less at 800MP and 4x5 can resolve 150MP. Monochrome Ilford Delta. 35mm has very low resolution so I never use it. Medium format also has quite low resolution. You get mich higher resolution with a macro lens than any normal scanner (drum scanner is probably better)
Back at it again with another film video, film core individuals must rejoice🎉🎉.
The questions, the answers.
As well, the misunderstandings and omissions!
In the end: do what looks right and works for you!
Back in 1992, I had the opportunity to work in a professional photography studio for a short time. One of the experiences that stood out for me was taking a loop and inspecting the details on an 8x10 transparency. It was an architectural piece a photograph of a six-story building taken with a tilt lens. I was able to look under a magnifier at one of the windows of the building and I could see the details in the jewel lens of the ceiling trough lights inside the office space of that window. That was quite an amazing amount of resolution captured on film considering the entire frame encompassed the entire six-story building.
This is a complete misunderstanding of how film works. Silver halide crystals have quantized size. Not only do they tend to clump together to effectively form larger clusters, but smaller crystals have a negligible effect on the image. Color film is even worse as it uses dye dots, which are significantly larger than silver halide clusters. Furthermore, the line pair ratings are for contrast of 1000:1, or in other words 10 stops of light. That's close to the maximum dynamic range a piece of film can capture. A test chart cannot recreate such contrast - paper will never get that bright/dark. So you have to work with much lower numbers. In practice there is no way you can get remotely close to 54mp from a piece of 35mm film. Color film nets a equivalent resolution of 5-10mp, bw is around 10-20, both depending on the exact film stock. With modern sensors digital has significantly denser resolution than even the sharpest film stock. If you take fuji's 100mp medium format as a benchmark for digital you'd have to go to 4x5in to beat in in bw, and to 8x10 to beat it in color. Frankly this video is just misinformation. It is a complex topic, but that doesn't take away from the fact that this is wrong.
I don't intend to be argumentative, but am curious about some of your assertions. A contrast ratio of 1000:1 is 10 stops, agreed. Also agreed that paper will have trouble with that dynamic range. But where is your data for 5-10 MP for 35mm color film and 10-20 MP for black and white?
I know that Fujifilm calls its GFX series "medium format." But, the sensor size is about 44x33mm, certainly larger than 35mm but quite a bit smaller than 6x4.5cm. Furthermore, the Bayer interlace pattern means that about 3 pixels are required for each color pixel rendered in the final medium. This general principle (multiple digital pixels required on the sensor for each rendered) applies to digital sensors regardless of the RGB sensor pattern used.
As you say, this is a complex topic, but one that I am fascinated by. There are lots of reasons to prefer digital over film (and vice versa), but I'm thinking that resolving power is not the main reason for the preference.
Thoughts?
@@nvrumi In most documentation on film a secondary line value is included, at contrast 1.6:1. If I remember correctly it was enforced by the ISO since the 1000:1 value is basically never relevant. If you take those values (there are some in the documentation provided in this very video) you'll end up with resolution similar to what I described. Of course different film stocks are going to vary. I read a paper recently which compared some of them; if I can find a link I will attach it below.
The bayer pattern does make this somewhat more complex. In this case it is more helpful to look at luminance and color resolution separately. The luminance resolution is preserved really well even through debayering, while color gets a bit more lost. The multi layer approach of color film conserves both types of resolution equally (or foveon sensors). Applied to numbers we could assume that a sensor with 24mp has its full resolution for luminance, but only about a quarter for color (or half in each dimension), whereas color film will have the same 5-10mp for both. In practice color resolution is less relevant, as we see that less well than luminance differences, but on a technical level yes, digital suffers from debayering. These numbers are only estimates though.
I don't want to hate on film - it's a lovely medium. It is fascination to divulge on the technology, but in practice it's just about what look one prefers. In technical applications digital is significantly superior nowadays, but photography is and remains art. Personally I shoot digital for various reasons and I don't care for the look of 35mm film, but medium format film I adore (and larger) - maybe I'll get into that some time ;)
@@nvrumi I am unable to provide the link since youtube keeps autodeleting my comment.
@@ChrisThe1 I was just reading Tim Vitale's white paper on this topic (and remembered that I'd read it before) and reviewing what Ken Rockwell (love him or hate him...) wrote as well.
In the end, this discussion seems like it misses the target (no offense to you, but in general) that digital photography and film photography are just different beasts. The sort of "resolution comparison" is the wrong question because I don't think there is a final answer because of the inherent differences.
Both mediums (media?) produce images that can be very pleasing, or informative, or alluring, or choose an adjective. Particularly if film is wet printed (which is very different than digital printing of scans) there is an even greater difference between the two results.
I think people want to say one is "better" than the other. But both are art and so much of art is subjective.
I shoot both digital and film, although I am shooting more film now (35mm and medium format -- the latter is a blast). I love the convenience and instant gratification of digital. But I prefer the appearance of film (mostly) and I know this is going to lead me back to the wet darkroom again.
Maybe this is nothing more than I grew up with film. I don't know...
But I do enjoy the discussion with reasonable people.
Thanks...
@@nvrumi Totally agree, they are both just very different and incredible in their own way. Although I haven't shot film in quite a while I have very nice memories of the wet printing process. Thanks for the discussion :)
I shot 35mm film in the 90's into the early 2000's. I gave up on film when i got a 10mp Nikon D200. Scanning film with dust, scratches, curved negatives etc. at home was not worth the effort any longer. For me, the benefit digital gave me was the freedom to shoot more, experiment more, and not needing to worry about the price of film + processing and hours of time scanning. I get the idea of people wanting to shoot analog end to end, developing and printing at home, but I honestly don't get shooting film to have someone process it and then scan it to then edit digitally.
Each to their own (preferences.) I was an early digital adopter, my first camera was a 2.1mp Oly and USED memory cost over a $1 a MEGA pixel. Film? Since 1953 since my dad was a professional photographer. Learned how to develop and print as a kid and teen.
I upgraded cameras as the technology got better and I could afford them. That ended with my 8mp Konica-Minolta A2. That was about seven years ago and that's when I said to myself. "I miss film. It was a lot more fun." That includes making my own B&W developers from scratch and playing with my own formulas. I've also developed color neg film since the 1980's. Difficulty and challenges is why I like film.
OTOH, inexpensive phones can do things within their SW that inexpensive digital cameras can't, based on recent research. (I was hoping to find a fixed lens, hand held camera under $500 to do what my old phone can do.) I won't go into what having super tiny pixels but a lot o them (64+mp!!) allows to be done by pixel binning, combining, and comparing. Instant single shot HDR. Expanded, full dynamic range. Things like that, right in my pocket.
Not a perfect metaphor, but digital is like flying, film is like a good road trip. One gets you there w/o drama (hopefully!), the other puts you into the sights and smells and interacting with the world.
@@frequentlycynical642 That's a horrible metaphor. I can just as easily switch each other for road or air travel and neither format makes that type of difference. I can use by plane or car as the less drama method depending on what I want to say, but more importantly, neither film nor digital gets you *more* into "the sights and smells and interacting with the world."
@@JH-pt6ih Jeezuz, man, cool off. If you do your own chemicals and processing, you are much more involved than you can ever be in digital. If you will reread my comment, I have decades of experience with both. Do you? And it said it wasn't the best metaphor, so don't pile on me with your anger. All digital cameras are auto-everything. Kind of like letting a pilot fly you. While there were a few autofocus film cameras, they were horrible and never sold well. The most auto film ever got was auto exposure.
But mostly, you don't have to prove yourself right to me or anyone else. Nor do I. Just let go and go have some fun, OK?
@@frequentlycynical642 53-42 = 11 for film, i started @13 w/brownie 127
@@JH-pt6ih Get BTFO'd homie. What he's saying is digital photos can be fired off a hundred times a minute, all perfectly calculated, all properly exposed. Film means you slow down, take your time, enjoy the process. It's a different mindset.
Trump2024
You're really only going to get 54MP out of those color films if you're shooting a black-and-white test chart (i.e., that's where you get that 1000:1 contrast ratio that gives you the 125lp/mm number). With actual real-world color subjects, the resolution's going to be more like that 50lp/mm number, which works out to only about 8MP, which is well below what modern digitals are rated at.
Of course, all of this is really a red herring, since resolution is one of the least important aspects to what makes a good photo. As long as it's higher than about 4MP, things like subject, composition, lighting, etc are going to be far more important to making a good photo than just number of pixels. A well-composed picture of an interesting subject taken with an iPhone 5 is going to be a better photo than a test chart shot on a 16x20 view camera, and a well-composed picture of an interesting subject taken with a Pentax 110 is going to be a better photo than a test chart shot with 150MP Phase One IQ4 full-645-frame digital. The only sort of photography where pure resolution matters if you're taking documentary photo scans of other artworks or historical artifacts etc, which is what films like Adox are designed for. For anything else, you wanna go with whatever setup works better for you artistically.
Actually, you won't even get that contrast in a test chart. 1000:1 is 10 stops, close to the maximum dynamic range of film. Paper's dynamic range isn't close to that. Matte, as is used in most test charts is anywhere from 4 to 6 stops.
By the same marker, this would only apply to monochrome digital cameras or foveon types. Most digital sensors need a grid of 4 pixels to accurately record colour, while film does this through layers stacked on top of each other.
@allankcrain That's also my finding. My old Fuji Velvia 50 slides (scanned with a LS5000) have about the same amount of details as a 6 to 8MP digital image.
I'm in the process of scanning my family's photo negatives. There are at least 400+ rolls waiting to be scanned. Most are just typical family stuff, taken with anything from cheap compact camera to pretty decent SLR - but no award winning pictures here obviously.
After conducting some test I've settled down the resolution to about 8.6Mp (3,600*2,400) - also in glorious JPG format too. This should be more than enough in this context. Scanning at 56 Mp / in RAW format would not only be waaaay too overkill, but extremely impractical so I won't bother considering that. Maybe just for some particular / important images, but not all of those 14,000 images - for sure.
Yap, just wanted to post similar. I have scanned thousands of negatives on Nikon and Acer scanners, and my conclusion that at the best resolution is max 12mp of color negatives 100asa. B&W ILFORD 400asa which we use to shoot while I was photojournalist is far worse than that.
Purely theoretically this might be correct. But real life useable resolution is much - MUCH lower. I digitised my film archive with DSLR scanning. First of all I have to admit - my film equipment was pretty average. Not bad, but not great either. I had 12 and 24 mpix DSLRs at that point and a vintage slide duplicator. As a reference for the slide duplicator capabilities I used film defects - scratches. I focused to achive best sharpness on them. After trying different ways to focus this gave me the best results. Contents of the film were much less sharp than these scratches and I couldn't achieve anymore sharpness improvement. I tried both my DSLRs and found that even 12 mpix gave me more resolution than I can actually extract from film without diving into counting individual film grains. I admit that if all the analog process if focused for maximum resolution - you can achieve noticeably better result, but that's not how an average film works. So for real life usage digital surpassed film long ago.
"Film Grain, Resolution and Fundamental Film Particles" by Tim Vitale pdf compares various color films, states avg line res 64 lp/mm that makes about 14 Mp for 35mm film, but that's line res, not color res, dye clouds are quite large.
In Kodak Gold 200 35mm, I didn't see details worth 14 Mp. But it's important to understand that not every camera lens resolves too much, many good lenses only resolve something like 20-30 Mp, some cheap zooms cannot resolve even 14 Mp and only few lenses can resolve more than 50 Mp. Also for high-quality printing we often need only 2-8 Mp (that covers sizes from 15x10cm to 20x30 cm), for larger prints it's logical to drop the ppi because with greater print size viewing distance grows so 8 Mp is really all we need, no point to chase megapixels in digital or in film.
P.S. Because of the Bayer's matrix in digital sensors you need more digital pixels to capture finest details accurately. In other words, 14 Mp digital pixels directly from digital sensor won't have 64 lp/mm due to aliasing/debayering issues. So while 8Mp is all we need, it really matters if these final 8Mp were obtained from 10Mp sensor or from 24Mp sensor.
lol. so you say that if a mount cheap zoom on a 50Mp camera i will not see 1 pixel details ?
@@vasyapupken that's what I say, yes.
Lenses do not have infinite resolution, especially cheap zooms.
For example Canon EF 24-85mm f/3.5-4.5 USM and Canon EF 55-200mm f/4.5-5.6 II USM when mounted on 50MP Canon 5DSR resolve details worth 12 MP max and ONLY if we close aperture to f/8 and ignore the fact that corners are blurry, have fringes etc.
But Canon EF 85mm f/1.4L IS USM on the same camera is perfectly sharp even at f/1.4 with measured sharpness: 45 MP.
This theoretical analysis is very different from real-world photography, scanning and printing. I’ve been a graphic designer for 40 years, and have been using scanned photographs in design software for over 30 years. A high-res 8x10 scan from a $250,000 drum scanner of roughly 100MB was not nearly as good an image as I get from my LUMIX S5, which is 24MP. The files open up at about 115MB in Photoshop. The detail in normal professional 35mm or even 120 film cannot produce real-world results of a modern 24MP full frame sensor. Even my Canon 7D and 5D III produced better files than I ever saw from the service bureau scans. When I shoot film today I have to really reset my expectations for resolution and detail or I’m very disappointed. Still, there is a beautiful roll-off from specular highlight to the regular highlights on film that you don’t usually get with a digital sensor.
I agree, and I think the problem comes from overinterpreting MTF curves (which admittedly is easily done). I was a commercial photographer in the 1970s, and there is absolutely NO WAY 35 mm film could yield anything approaching the quality of hundreds of megapixels. We NEVER used 35 mm film for commercial work, and even medium format struggled to make the mark. We had to use large format, at least 5x4" but often 10x8", for the kind of work that is now routinely done with 50-100 MP digital cameras.
@skfineshriber Yeah I'm trying to figure out this you tubers results on thinking film gives more detail. I remember learning thst 9 to 12 megapixels is a goid range for the digital equivalent to film. Because film depends alot on how good the lens is. Cheap lens = Maybe 9 Megapixels. Pro gear abd lenses = 12 maybe 15 megapixrls if your lucky
@@SchardtCinematic and it depends on which film you’re talking about. 👍
@skfineshriber Yep exactly. So many diffrent grains to choose from as well as film speeds
@@SchardtCinematic Exactly, and in real life no-one used ISO 20 film. 50 or 100 for studio work, both in b/w and colour - and 200 or 400 for handheld, on-site jobs. Yes, ISO 20 for copy work and the like, but never for anything that had depth or moved. The thing about real photography is that it's a practical exercise, not a theoretical one.
1:11 thank you for adding a Cirno for scale, helped immensely!
This essay has one major flaw: If we take that Fuji 200, it only has 100% resolving power up to about 20lp/mm. A 1.3MP digital camera can match that resolving power. (There is some fudging with demosaicing, but let's assume a monochrome camera). Past 20lp/mm, the accuracy starts to drop. Meanwhile a digital camera is 100% accurate well up to the point where the film drops to 50% response. A 12MP camera is "better than film" up to that point. A 45MP camera will have a 100% response at the point where that film has dropped to 20%.
So it's disingenuous to say that Fuji 200 film has a resolution of 54MP. It has a resolution of probably more like 4MP. Sure, it can still show difference between those values in some sense at a finer and finer scale, but a modern DSLR will wipe the floor with it at clarity, right up until the sensor's own MTF curve takes an abrupt downward turn into noise, somewhere around 115lp/mm.
A 45MP camera will not have 100% accuracy because, even if it does not have a low-pass filter, the MTF will be smoothed out by the objective lens.
The same happens with the 1.3MP and 12MP cameras.
Good contribution!
Am I misreading, or are you suggesting that digital imaging systems are immune to resolving power limitations of lenses?
But given a scan of sufficient resolution, you can boost the film-typical poor contrast at the finest detail level that you describe as "less accuracy" and recover more of the inherent resolution. The same thing that a photographer did in the old days when printing on "hard" contrasty paper to make the details crisper and less soft, only much better controllable now in the digital domain. From a low-res scan however that throws everything that isn't 100% contrast (in your example, 20lp/mm) away, no finer details than 20lp/mm can ever be recovered. Astronomers (as a typical example) demonstrate every day that lens flaws and focus problems can also be reverse-engineered digitally if the input is sufficiently finely resolved, to get images that fully exploit, sometimes even exceed, the original specifications of the equipment and materials used. Yes, raw full-res film scans will always look "soft" and unfocused because of the gradual dropoff of that curve. Mirroring its characteristic (aka sharpening) in the digital domain can however bring back close to 100% of detail at any point along that curve that is distinctly above 0%.
Sure, a 1.3MP camera can match that but it won't do anything more, while the film will. 4MP is still a super low number to assign to a film, even one as boring at Fuji 200. I shot some recently, using a 88 year old lens that is losing its coating, and had it scanned on a Noritsu at 30MP as TIFFs. I see
There's another aspect of this comparison that is of great importance, and that is the spatial spectrum of the noise (grain or electronic) in an image. Film images will always have much larger noise energy out at the standard limiting resolution where 1000:1 contrast resolution target is reproduced with 20% MTF.
By the way, the correct term is "modulation transfer function."
The comparison between digital and film sharpness (not resolution) was addressed in the early 1950s by Otto Schade at RCA. He developed the technique of comparing the squared area under the MTF curves of the different media to find a numerical estimate of the apparent sharpness. There has been some argument since whether it's the squared area, just the area, or something in between that best estimates the visual sharpness, but the main principal of examining the area under the MTF curve. has remained. Essentially, higher contrast at mid frequencies (as in electronic cameras)can produce about the same apparent sharpness as higher limiting resolution with a lower mid frequency contrast and more gradual rolloff (film).
Nicely done. I have been a photographer for a long time and can say that this mirrors my findings very closely.
Excellent video. I am a boomer, working in the movie industry for a few decades now, so I really had to deal with film. My experience is that scanning 35mm film higher than 4k makes no sense because with the usual film stocks (Fuji, Kodac) you already have like 4-ish pixels per grain. But keep in mind 35mm film runs vertical instead of horizontal in photography, that means our negatives are closer to MFT than to Full Frame. On the other hand even though our eyes can't see much more than 2k, scanning a 35mm in 2k is still decently sharp but you can't recreate the original corn structure due to a lack of resolution, so you can spot it as digital right away. Watch the balcony scenes of De Niro/Amy Brenneman in "Heat", that was scanned in 2k and you can tell right away that it is a greenscreen scene, not because of bad vfx but because of a lack in resolution.
I just retired after over 40 years in Hollywood in post (colorist), visual effects, film restoration and HDR R&D. I just looked at the scene you offered in Heat. The wide shot just isn't a great comp. Possibly too much spill on them from the chroma screen. If they'd feathered the matte a bit it may have taken away some of the edge harshness. The singles are better. It's not a resolution issue. Being released in 1995 those comps were done optically. They probably looked more natural in the release print. I once had an issue with a feature from the early 2000s that had 8 bit digitally composed opticals that had been made into film elements and cut into the O neg. The entire feature was scanned 4K for a home video release and as one would expect, the scans revealed the extreme disparate quality of the negative versus the opticals. They probably would have looked as bad in 2K. I had to jump through a lot of hoops to match them to the rest of the show. The photochemical process covers a lot of issues that scanners unforgivingly reveal. When restoring old features, we had a process for completely removing grain and restoring it to whatever degree the client asked for. Our process could boost resolution of the image by sampling images before and after to add more detail. This was possible since the grain position is different in each frame of film. The finished deliverables were a digital intermediate film element and a home video version. The point is, we could get away with all of this stuff since we were using motion picture images. When you scan a still, you're just sampling the grain of the film...which is itself, the image. Unless you have a large format negative, at higher scanning resolutions you're just getting a better look at the edges of the grain. Sorry to be so long winded.
As a professional photographer, I shot film for 40 years until starting digital photography in 2000. My experience was, in the early days of scanning film on a Kodak 35mm film scanner, (about the year 2000) that if I made a print from a 35mm color negative onto Type C paper, in the darkroom, lets say for an 11x14 print, I would get a beautiful, smooth toned, relatively grainless image. If I scanned that 35mm negative, and then made a print from that digital file, on a color printer, I found that the scanning would bring out the grain. Instead of a smooth grainless image, I got a printed image that was full of grain. I am guessing there was something in the wet process and the combination of the color negative film and the Kodak Type C paper, that made the grain so much less. When the Nikon D2x digital camera arrived, and I did some aerial photographs with it, I discovered that I got a a beautiful, smooth toned image, compared to shooting negative color film and scanning that! I hope this adds some information to this discussion.
as you may know, a scanner is a one pixel wide sensor, which acts like a point light source, bringing out the grain structure.
I no longer do detailed posts on this kind of thing, but a lot is not covered here: RMS values for positive film, specialty films which required meticulous processing- notably Tech Pan film in HC-110, and Gigabit film in its dedicated chemistry. Use of pure azo dyes in now defunct Kodachrome 25. Tricolor (trichromatic) process making 3 identical exposures in black and white through a 29, 47, and 61 filter, then converting them to a single color print. A LOT of variables. Few people know about Linhof tested lenses. At one time, Linhof had agreements with Zeiss, Rodenstock, and Schneider to buy lenses, but then to only keep a % of them which tested best. They found 30% variations in the SAME make and model of lens, as sample variation. Personally, the best lens I have ever seen was a Voigtlander Apo-Lanthar in 105mm, although some process lenses are probably better. That Voigt was a different company than the lenses you see today, which are quite good, but made by Cosina.
Great video!
It’s worth noting that in audio, a higher sample rate than 44.1kHz doesn’t actually result in any greater perceived audio quality. Only an increase in the maximum frequency which is able to be captured, but as those higher frequencies are outside the range of human hearing they go completely unnoticed. This can be useful if those higher frequencies are going to be shifted down into audible range but otherwise it won’t make any difference aside from larger file sizes and greater CPU usage.
As a musician that performs with audio. There’s a difference in feel
@@EntonationStudiosno. It isn’t. I guarantee you, if you would do A/B testing you would not be able to distinguish 44.1 KHz vs anything higher
@@TheScrubsfanforever Come to Cincinnati Ohio, I’ll bet you dinner 😂
@@TheScrubsfanforever I’ll expand: I do audio engineering part time on top of being a musician and I spend a lot of time listening to high res audio through studio gear, so it’s possible my ears are attuned to it
@@EntonationStudios ah nice. The Anecdotal evidence. Fun fact: I studied that shit 😄
You only need higher frequency’s if you a) don’t use a low pass filter, or b) if you somehow slow the audio down
There is probably no human on earth that can hear frequency’s higher than 22Khz, maybe a few newborns
Nyquist says you must have GREATER THAN twice the frequency. If the rate = the frequency you get a constant value that is arbitrary. As you approach the Nyquist rate, the number of samples you need tends toward infinity.
Thank you for this very interesting video. My experience with the Fuji GFX line is that while the sensor is not as large as a 6x4.5 film image, it is substantially larger than a "full frame" digital sensor, and produces fantastic, highly detailed images at a very reasonable price.
This is the clearest and most concise explanation on the subject I've ever heard. And I've heard a LOT in my day. Even if film didn't meet or exceed digital in the technical sense, I'd still shoot it for the beauty and fun of the process.
@@jaimeduncan6167 You're right. It would be interesting to see a photograph of checkerboards to illustrate this point. Basically if you compare a checkerboard with pure white and black squares (TOC 1000) to one with light grey and dark grey squares (TOC 1.6), and you have amazing lighting, if you move the checkerboards away from you as you take a bunch of pictures, you will be able to clearly see the black/white checkerboard from much further away. The grey checkerboard will turn to mush as soon as its size at the film plane reaches 0.25mm whereas the black/white one will do that at 0.08mm.
So despite using the same exact film, the same exact size checkerboards, the same exact image, you will be able to clearly see the black/white checkerboard from over 3x further away than the other one.
@@jaimeduncan6167 He mentions that these measurements are theoretical and ideal, and the reality is dependent on all the systems between the media itself (film or sensor) and the viewer (optics, scanning, enlarging, monitors etc). At any rate, it doesn't really matter, it's just a mental exercise. We choose film or digital for subjective reasons
Check out drum scanning and how it works.
Interesting video. But from a practical viewpoint, you will often get a sharper image from a digital camera. With film. It depends on what film, how it is developed and scanned. Most of the time we don't use drum scanners. Too expensive. And film is best at low ISO. Where digital will often give great results at 1600 and higher speeds. I remember my first 5 megapixel camera. I had doubts about the quality. I had a model come in, shot a few photos and printed an 11x14 inch print on an Epson printer. It looked great. As good as from my Nikon film camera using color negative film. My current Fuji 26 megapixel camera makes great 2x3 Foot prints. Much better than my old 35mm film cameras. Why shoot film? It has an artistic quality. I even like the grain I get. It seems more authentic. But technically digital passed 35mm film up years ago. And is competitive with medium format film and even approaches 4x5 film cameras, at times.
One thing to not forget is the scanning process when using a scanner like a Nikon 9000ED will yield colour information for all three colours in the RGB spectrum and at 64Bit including infrared information. That’s 16bit per colour and all colours present in one pixel instead of a interpolated Bayer pixel which offers 16bit max for one colour channel in one pixel, but most often 14bit unless a highend medium format sensor. It’s just like how Foveon sensors capture more details per pixel.
And still those sensors blow away the crappy chemical spec based celluloid carrier, film is based on.
How could it be otherwise, sharpness in film is just luck, because of the perfect spreading of the chemical particals needed which is off course never the case.
And it shows there`s always some crappyness about it, and grain lots of grain and looking at it is uneasy on the eye.
And we`re not even talking dynamic range.
Microfilms were used a lot in the past by libraries or mechanical workshops to store a lot of manuals and information. The resolution was incredibly high. I wonder if it compares to the resolution of small image sensors on phones for example.
Back when digital cameras literally were first coming out, I read a pretty well researched article in one of the photography magazines. Unfortunately, i can't remember which one or exactly when. They said that 35mm 100 iso film would be the equivalent of 24 megapixels. And this was back when 24 megapixels was a pipe dream.
Their calculations were based on the actual visual information that a film image can contain, counting only the level of resolution that actually contributes to what the human eye can discern as part of the original scene, and what optical equipment can convey. In other words, a 24 megapixel image at 8x10 would look indistinguishable from a 35mm 100 ISO 8x10 film print... For most people in most normal conditions.
Those are the numbers that I have been going by. My Sony A7R2 is 42 MP, and the a6700 i want is 26 MP. So I'm happy.
I am a photographer who started in 1975, we used very fine grain low ISO/ASA B&W film and that was extremely fine grain. Shooting reversal/slide film gave very fine detail and we used every format upto 5x4 in the field (sometimes 10x8) . My last proper use of reversal film was to photograph camouflaged military equipment at varying distances and height from a helicopter to be used for aerial recon training.
When you send in film to develop in a shop, most of them only send you 12mp images, and even then I often can see softness in the pixels.
Looking at that MTF curve, only at 20lpmm can it resolve 100% contrast, but 2 digital pixels can resolve 100% contrast all the time. Accounting for lines not alined with the pixel grid let's just say 4 pixels required for the equivalent of 20lpmm. This gives us a digital resolution of 2880x2160 or 6mp, a much more reasonable number.
I mean just look at the curve at 120lpmm it has 20% contrast, that doesn't really count. I will say under the most ideal circumstances, the best 135mm film photo I've seen, can be reasonably scanned to 24mp. I've never seen a 50+mp scan that doesn't have serious sharpness issues at the pixel level.
This answers a lot for me: I only recently got into a serious digital camera a couple years ago. I played with it for a while, but just was not happy with the images, and soon set it aside.
Back in the day, I used to shoot Panatomic-X which was ASA 32! And, I got amazing images that I could easily enlarge to 16x20 without any serious degradation in the image. When I would shoot it on 120, you could project it across a room and not see anything that looked ‘too-far!’
Anyway, I’ve learned that you have to settle for limits, and be happy with the current state-of-the-art; otherwise you’re pretty well stuck with accepting what you get out of your phone… 😢
Very thorough examination of film resolution.
That video taught me so much. Thanks for bringing this to me!
My dad was a commercial photographer. He dragged me into his darkroom enough times to make me realize I would prefer cleaning up after elephants.
Is it really you, son!
Spent quite some time testing our and other manufacturer's digital dental intra-oral sensors and film for our competitor comparison when I worked in dental radiology ... we had a (very expensive) test pattern to shoot giving us the "resolution" proof
In audio Nyquist theorem assumes that there is no destructive effect from having to brickwall filter aliases above the chosen bandwidth - but in the real world there is. In the same real world, there is no way that a 35mm film has the resolving power of an MF digital image.
Intuitively, nobody would would assume that "lines/mm" refers to both the white and black lines.
That's one of the reasons I made this haha I was confused too and no one had a clear answer.
These numbers are very interesting, but I think also very wrong. My father had a specialized high quality photo lab ("photo finisher") for professional photographers and customers who needed enlargements for trade fair show wall-sized pictures. We actually did have one photographer who shot on 8x10" - and we were the only lab in hundreds of kilometers that even had an enlarger who could take these huge slides. I say slides because this was the reason the photog shot this big: to show his work to the customer without any other party being able to mess up. So even in the 1990s there really was no good reason to shoot film this large. Bread and butter was 4x5" and 5x7" for high end and most was shot in medium format.
None of those still shoot film professionally. Even 10 years ago most went digital in 24x36 and when the first medium format cameras/backs came out large format was almost instantly dead.
So digital already surpassed film in quality and I would say by a large margin. I guess the misconception revolves around the contrast: Of course you can resolve B&W test charts fine, but real world pictures are a differnt story. They don't have contrast 1000:1 usually.
This was supremely understandable, while also being sufficiently technical. Prior to watching this, I'd not considered how the 'resolving power' of a film format could be affected by micro-vibrations caused by moving camera gear (shutter, etc).
One of the best content I have got on the internet on this topic
And the most informative, arguments against it in the comment box
Thanks a lot
This settled a lot of debate
If you get down to the molecular level, film is digital, a silver crystal can be activated or not activated. while a digital sensor is analogue, it measures the amount of light activating a photodiode. but since pixels are much larger than film grain, and the photodiode is immediatly converted to a digital signal it makes sense to call it digital.
Just to confuse the conversation. lol
Don't confuse me any more! 😂😭
And why anyone with experience scanning will modulate the scan resolution dependent on the stock/processing, scan delta 3200 pushed at 5400DPI and it's literally black and white, there are no greys.
You're confusing film grain with individual silver halide crystals in the film emulsion. What you see as 'grain' is the aggregate of thousands of crystals. So, no, grain is not digital.
@@emotown1 A silver halide molecule has an electron exited to a higher energy shell when hit by a photon. This silver halide is either in an excited state or not. it becomes soluble when developed, or not. On or Off. (yes this is simplified, but is essentially true. I went to college for photography back in the film days)
Don't confuse grain for detail. You typically need multiple grains to create a detail that is meaningful.
Anywhere from 9 to 12 Megapixels is considered 35mm Film Quality. Like you said you really can't define film as Megapixels due to there being so many diffrent grades of film. Plus it also depends on the quality of the lens you used.
It’s a relief to know that even a camera phone these days will give a highly satisfying image and pro-sumer cameras are good enough for any professional, and pro-level cameras are more about personal taste and work flow than image quality.
Excellent thank you - this is surely should be the definitive reference for this subject on UA-cam.
Awesome job, love the video and your way simple but very informative to speak about the technical aspects, it’s fascinating!
There's something that I have never seen addressed in these videos, and is the fact that, AFAIK, each grain in a film can only be either black or white (or, if it's a color film, either the color or black), which means that any light/color gradation must be done by dithering. So yes, you can have very small lines for very high-contrast elements, but as soon as you have normal-contrast elements, the image dithers and you loss a lot of resolution (which, BTW, explains the two values of lines per millimeter). Instead, in a digital sensor each pixel (or subpixel in the case of three-sensor cameras) can have 256 or more levels of intensity per pixel, so although the "apparent resolution" (this is, the number of line pairs per millimeter) is smaller, in practice, with "real world images" the resolution loss is not so extreme.
Or that's my opinion.
this was cool and very interesting, thanks!
Great video! Just a side note, Adox recently (2024) discontinued their CMS 20 II 4x5 film. The 4x5 would theoretically have an equivalent of 32 GP.
Tmax 100, is nice and sharp, but it’s also very high contrast. Ilford 100, developed with tmax developer is much nicer, because the gamma curve is softer. More subtle and intricate detail if you expose it right, and develop it right.
This is a huge and complicated topic, and not sure if this vid adds much of value. And important to note: colour depth has a big impact on visual impact and enjoyment, witness the trend upwards from 8bit digital representation (whether in the camera or scanner) to 12bit and higher.
You tend to get the same argument when you compare analog sound vs digital. In both cases I tend to prefer analog but I use digital for the convenience of it
Has any film camera manufacturer even considered using an LCD as a shutter? Or would the polarizing filters on them affect the image negatively? That could help basically negate shutter movement distortion...
Two pol filters would cut the light a lot. But it has been done on lab equipment.
I think something like that has been patented decades ago, and shutters based on polarized filters have been used in many high-speed cameras, including the ones used for capturing nuclear explosions
chatgpt explain :
Here’s a more detailed summary of the idea of using entropy in the conversion from analog film resolution to digital formats such as HDR10+, with some scientific insights and formulas:
Entropy can be used to measure the amount of information when converting analog film resolution into digital formats like HDR10+. In information theory, Shannon entropy quantifies the average number of bits required to encode information, which is essential in image processing.
For analog film, resolution is often measured in line pairs per millimeter (lp/mm). To estimate the number of pixels required in a digital format:
P = L × H × (lp/mm)^2
where L and H are the film dimensions in millimeters.
In HDR10+, each pixel uses 30 bits (10 bits per color channel). The total amount of information (in bits) for an HDR10+ image is:
Total bits = P × 30
For example, a 35mm film with 50 lp/mm results in ~2.16 million pixels, needing:
Total bits = 2.16 × 10^6 × 30 = 64.8 megabits
Entropy plays a key role in determining the visual information and optimizing conversion from analog to digital formats. Higher entropy in film (due to grain and tonal variation) typically requires more digital data for accurate reproduction.
Thank you! I’ve been looking for this exact information for years! Cheers!
That 54MP is wrong based on the numbers there. The 9000 x 6000 dimensions each need to be multiplied by √2 to make sure the diagonal is capturing 250 pixels / mm. 12728 pixels x 8485 pixels = 108MP, not 54MP.
Edit: OK they did briefly acknowledge that
I loved my Omega B22XL enlarger, the first one shown in the video.
My go to reference for film resolution is that Super35 film can produce 4K images. We've all probably watched movies shot on Super 35mm and displayed in 4K. Film has a very soft frequency rolloff, so what this comparison is for is less for maximum resolution of film and more for the vague perceived sharpness. The image is slightly softer than digital, but the higher frequency details are definitely there if it's shot on a sharp lens. Film will capture much finer details than that providing the lens is sharp enough and in focus, but I use that as a basic sharpness comparison. And this is fully anti-aliased. But it REALLY depends on getting a good scan.
To get an image WITH NO ALIASING IN ALL CHANNELS out of a Full Frame Bayer pattern sensor, matching that resolution (4096 pixels per 24.89mm along a single dimension), you'd need a 187.MP sensor (double that pixel density then multiply it by √2 to get the diagonal*). That's higher resolution than i think most of the digital medium format options. You wouldn't need as many pixels with a hexagonal arrangement, but with bayer pattern sensors and an OLPF that abides by signal sampling theory that is what you'd need.
For a more subjective comparison, you can take a decent quality (not cinema quality) full frame scan, and you compare it with a 60MP full frame photo, you should be seeing much sharper fine details in the digital photo, if your scan even picked up the fine details at all on the film photo at all. But the fine details in digital are not pleasant - they have a bunch of aliasing and noise reduction and de-Bayering artifacts. Manufacturers could apply an OLPF to fix this, but they don't, in favor of sharpness and maximum resolution, presumably for competitive reasons.
* Arguably if I am comparing scanned film at that resolution I should use the diagonal for both, meaning not multiplying by √2, but: a) 4K images often contain aliasing either from the scan or the downscaling filter used to make the picture sharper, which I would argue is bad, but we are just using it as a resolution comparison here, and b) as I said film definitely can produce much higher maximum resolutions than this benchmark of vague perceived resolution anyway (as seen in MTFs on the datasheets), while digital has a hard limit on resolution. For reference without using the diagonal the full frame sensor would need to be 94MP, which is your absolute minimum full frame film resolution (well, probably not YOURS since most of us shooting film are using soft vintage lenses and not using cinema quality scans).
This is consistent with another UA-cam video I saw where a movie studio basically said there is no point in rescanning old movie film at resolutions greater than 4k as the detail isn’t in the film start with. So these remastered 4k movies will likely be the best we get (at least without AI).
Thank you. A really great explanation!
Honestly the only way to even approach the sort of resolution that 35mm is theoretically capable of (digitally) would be shooting with modern Canon L glass at like f/8 and then drum scanning.
Валерий Дмитриевич, спасибо! Берегите себя и семью!!💚💚
Excellent video 😮!
Absolutely lovely video but one note: The H6d-400c doesn't "fake" 400 megapixels. It shifts the sensor by by subpixel amounts mimicking the pixel density of a 400mp sensor. It also fixes the bayer problem of having 2 green pixels, for everyone 1red and 1 blue. By shifting the pixels around every color can be represented as 1:1:1 which makes for much better color reproduction and improved dynamic range like a film camera.
Looking at a 400mp out of the Hasselblad vs the 100mp you can clearly see the increase in detail (not just sharpness)
It's "fake" in the sense that the sensor is only 100mp and effectively can only take a stacked array photo off of a tripod or other steady position. No one would take Apple seriously if they claimed the iPhone took 600mp panoramics (via stitching 12 photos in camera) so Hasselblads 400mp claim should be taken with the same grain of salt, IMO.
Pixel shift is a gimmic though. I’ve tried using it in perfectly still situations several times and it never produces a better image. And with the technical limitations of the process required it’s a useless tool. Better to stitch together 4 images at a more zoomed focal length
@@andytheturtle87 I mean Apple sells the new Iphone with a "48mp" main camera when in reality is a quadbayer sensor which is only 12mp in the bayer sense and it uses Ai Upscaling to fake the 48mp look. And people have seem to have accepted that in droves
@@tdawg719 I've tried pixel shift on a GFX 100 and my X-H2. And both have produced incredible shots well above the 100mp or 40mp respectively. I use it for art preservation and PCB shots.
The trick is to use a Mac, Fuji's app on windows just flat out doesn't work well and even when it says it finished the image, there will be no different except in file size whipe it works much better on a Mac.
@@tdawg719 My comment to you got deleted for some reason
I compared the resolution of negative scanning quality between my 2 Fuji cameras - 16MP X-T1 vs the 24MP X-T30. There is easily detectable difference and I have dedicated the X-T30 body for that duty. I use a Nikon PB-5 bellows with PS-5 slide copying attachment. To that I coupled a Fuji-X to Nikon F adapter and a Zeiss 85mm f1.4 lens. Works like a charm and with live view and focus peaking you get spot on sharpness. Just need to have a good and uniform white light source.
Probably the most extreme example of lens softening I've seen is on the Lubitel 2, at f/4.5 it's surprisingly good if vignetting however f/16 and f/22 are very soft.
yeah film is uh... haven't touched one since the 90s
>be me
>shoot with my a77 using retro but great Minolta lenses all the time, and a minolta badge over the SONY logo for maximal swag ( flip screen folded inside cause i don't wanna hurt it accidentally)
>hipsters around me believe I shoot film
>praises me because on photoshop I do edid the contrasts and levels when the full auto exposure breaks the dynamic range
>«See, that's real art, people who use a digital can't comprehend this, I want a print of that random shot!»
>just get it printed on a plotter from a .psd
this is as far as it gets, eventually for large prints i upscale it for the support and put a light random noise effect
do old minolta lenses work with Alpha cameras?
@@uwuuwu3628 every minolta AF mount does. except on the more recent 2010s Sony with E-mount which are for more compact gear anyway
I watched a documentary about old films being scanned. They said there wasnt any point in scanning over 4k.
I'll assume that the AA filter on most digital cameras would negatively affect sharpness/resolution when compared to scanned film.
Resolving power is an attribute of the lens. The appropriate term for film is accutance. It does not matter how many times this distinction is overlooked by UA-cam videos. It will always be contradicted..
Let’s be honest, 35mm is 8-12 megapixels of resolving power, even with fine grained film, and maybe 18mpx with tmax. Good technical video however.
As I wrote on another comment, what is the source of your data? I'd like to see it. (Really!)
Yes, same finding here. I scanned my old Fuji Velvia 50 on a Nikon LS5000 (4000dpi resolution = 20MP output) then carefully downsized it to a few versions and compared against the original scan. A 6 to 8MP image captures all the details of the original scan. I used prime Nikkor lenses at optimal apertures.
@@bernardd
1. all Nikon scanners have maximum resolution of 3600lpi regardless of "wow it 4000dpi optical" written on a box.
2. this resolution (3600) achievable only with high contrast test target, not a real photo
3. you say "prime nikkors" like it must mean something ) but no. vintage Nikkors are mediocre lenses for the most part. especially in terms of resolving power.
fun fact - one of the best resolving Nikkors is one in their film scanners.
let's be honest - your opinion based on experience with poorly shot and scanned film where you did nothing to achieve max res.
99% opinions in "film community" are like "i scanned it with my Pooptek-9999KHD it''s 9999K optical res so where is my details ??"
it is quite hard task actually to scan a film properly for it's max res. wet drum scanning is the way.
@@vasyapupken "Prime nikkor" simply means a fixed focal length lens (not a zoom). In the film days, most zooms used to be inferior optically to prime lenses. Modern lenses have made a huge jump in the past 20 years but I don't think a good "prime lens" from the 1990's shot at optimal aperture would be the limiting factor with 35mm slide film. That's an educated guess though. If anything, I'd say the Nikon scanner is the limiting factor, although again the LS5000 was one of the best non commercial scanners for slides, not counting drum scans. That's my feedback anyway. I'd love to find a way to extract more details so these comments are really interesting.
It seems that there is a similar analysis and debate in the digital vs. vinyl war. You can't directly translate between the two, but the reality is they're both perfectly acceptable formats that can be of high quality, so high the average person can't really appreaciate the difference. As you say, there is also a limit to human perception, regardless of the source or format; there are only so many cones and rods in your eye. I deal heavily in various media, and even though resolution is imporktant, there are other factors to consider. In digital cameras, I have found that the quality of the light sensor and lenses are more important than the number of megapixels it supposedly capable of recording. Anyway, thank you for the interesting analysis!
Hi ShyStudios. The material in the video is very intellectually stimulating indeed, regardless of the many comments. Kudos to you for not having censored any of them.
Even if i am in no way an expert I'd like to add my two cents about the lens resolution as it seems to be common knowledge that most lenses reacht theyr maximum definition somewer between aperture 5 and 11 what means that at aperture 2 it is going to be worse than at aperture 5. This somewhat invalidates your statement about the line resolution of a lens being roughly 1600/aperture.
If film has so much resolution, where are all these hyper acute pictures? Why do 35mm lenses that were A-ok for film fall apart when used on digital camera, yet the reverse is not true? When you look at photo books that had 35mm or sometimes larger format color shots from the 80s or 90s, they seem technically archaic, they are obviously falling apart on double page spreads, for instance.
These calculations may, in some way, be scientifically correct, but practically the resolution is nowhere near claimed. And it’s not a matter of using vintage equipment today, or of transferring shot film using digital scanning. if you go back and look at film shot when it was the state of the art, you will not be impressed by its technical qualities, certainly comparing same format to same format.
Don't forget that all those old photos were reproduced without any sharpening. Imagine if you took your 15 MP digital sir and made an inkjet print with all the sharpening turned down to zero I/e just printing the RAW file with no processing. That is digital's strength though - sharpening up fine detail even from a 5MP camera is child's play. Which get's us into the topic of "apparent resolution" aka "sharpness", which wasn't tackled in this video. If it had been, nowadays digital wins every time and without breaking a sweat. In fact I think many digital photographers tend to over sharpen their images, but I guess that comes down to taste. And I'm a film enthusiast, by the way! And you are right, digital sir's do expose deficiits in old (film era) lenses in a way that film never could.
@@emotown1 Nonsense, sharpening up images is very visible and not done with quality capture. Raw files are mostly quite soft but retain way more detail than argaic film pics.
Besides 15 mp is nothing these days it`s more like 30 to 40 mp, sharpening is not needed.
Lenses for film doesn’t have to worry so much about hitting the sensor at a right angle. That’s much of the reason they don’t work well on digital.
Photobooks made from digital look terrible. All the particularities stick out like a sore thumb.
Waxy and missing micro detail, bayer artifacts which is mostly lower colour resolution and poorer tonality, harshness from edge enhancing during demosaicing and poorer dynamic range which results in highlight clipping.
@@Frisenette All you`re mentioning has been resolved a long time ago, you sound like a broken down record.
Analog sucks in quality and reproduceability big time, analog is grainy, has very low dynamic range, poor color fidelity, is extremely cumbersome an expensive.
In short, it belongs in a museum.
That`s not my opinion, that`s everyone`s opinion. Every picture you see everywhere has been from a digital source how about that.
@@brugj03 these issues, and more, hasn’t been resolved.
They are fundamental flaws of the current sensors and algorithms to low level process the sampled data out of them.
Digital, as a viable means of photography has only been with us about twenty years. All the rest of photography is film.
The most beautiful and profound work in the realm has been done in that medium and continues to be made. In museums in photo books, posters, small and large.
Grain, in what shape and form it might take, is real. It’s like the texture of paper, like a fingerprint. It’s a structure from the physical world. And our brain recognizes it as such.
Not so with interpolated mush, manufactured and guessed at detail or various kinds of aliasing and clipping.
Lenses are the biggest limitation. I think digital has driven lens technology improvements, past "quality" lenses obviously fall apart with modern sensors, so to get the sharpest digital pictures, they are making technically accurate lenses. I've got a few older lenses that were very expensive back in the heyday of film, and they do not acquit themselves well in front of a digital sensor. Fortunately we also have digital means to correct deviations, but the old adage of Garbage In Garbage Out still applies.
Your resolution vs F# formula is only true if ignoring lens aberration. At fast F# more light is gathered further away from optical axis and aberration will have much significant effect. In practice there is always an optimum F# for a specific lens, usually one stop slower than the fastest F#.
Well, given how superior the film is and how inferior modern digital sensors are according to this video, I kindly ask for the same scientific explanation why in real world film photos are looking so rough and unclear, while pictures even from a smartphone looks crisp and sharp.
The best retort I’ve heard. Digital is for reproduction, for capturing the image. Film is for expression, for capturing the mood.
@@Sharpened_Spoon Yeah, yeah, similar to this phrase from my bike friend: "4 wheels ride the body, 2 wheels ride the soul" :)
Like he said at near the end of the video, modern cameras have better optics and most have some sort of stabalization, to help create a sharp image. Having said that, when I compare the enlargements I have of my 6x6 Mamiya (80mm lens) with large prints from my Nikon Z7 (85mm prime) the prints of the Mamiya are only comparable when they are shot with low ISO film, colour film has a grain size per colour layer and these layers are not alligned, so I think that is a mayor reason for the overal fuzzyness of clour film enlargements, when viewed close by. Also the optical quality of the lenses is very good nowadays: If you compare lens sharpness over time in history you can see the steps they made, better calculation methods, better milling and grinding and polishing, better coatings etc. I have two 35mm cameras from the 1930's one has no anti glare coating the other has, the contast in the photographs is very much better from the camera with the coated lens and this helps in percieved sharpness.
The video pretty clearly says that you shouldn't expect theoretical results in the real world due to optics, old cameras, not truly linear film planes, ect. It's also with noting that phone photos almost certainly have AI sharpening and denoise applied automatically, so it isn't a fair comparison unless you are viewing raw files from your phone.
@@andytheturtle87 Fair point. However, I often look at raws from my FF mirrorless camera and I've never seen the film photo come at least close to this level of detail
Yeah 138MPx... then you scan your film and it's no better than 12mpx at max...
why?
@@PointlessDrummer There is videos about it. No difference between 24mpx and 60mpx scan on 35mm film, you can't get more detail than 24mpx, probably even 12mpx
Back in the late 80s and 90s, analog cable had a grainy look which really was just noise from a weak signal. For reasons I never understood beyond the buried mainline into your home, they used RG59 coax cable to wire your home. It's okay for short hookups, but they strung this crap on the outside of the house under the eaves to get to different rooms. In the early 80s, they actually just brought the buried cable straight into the living room as only one room had cable. Grounding was doggy if they did it at all. They put a ground strap on a painted water facet. LOL. The picture was clear, except for static during thunderstorms.
So, HD blu-rays of 35mm movies show the grain which remind me of analog cable noise. It's not exactly the same. Film grain is much finer, and I understand why people would want to preserve that. I also didn't get to go to the movies that much when I was a kid, so I guess it doesn't mean that much to me. We usually just waited for it to come out on VHS, and VHS couldn't resolve film grain. When I rip DVDs and BDs, I filter it out because it compresses better since video compression mostly updates changing areas of the picture. Do what you want. That's just me.
I really wasn't expecting such a thoroughly-researched, high-quality video! Most videos on this topic just repeat stuff you can find in a single google search; great job! I hope you make more film stuff :)
Science vs real, practical use. My experience has given me great looking 8x11 prints from a 5mp camera in the early days of digital and I'm more than satisfied today with the 20mp cameras I now use for printing tabloid size and much larger prints. I dont know where all this resolution science comes from but I doubt 35mm film has as much real world resolution as stated in this video, at least in comparison to what I've seen from my film days. It may be because I didn't have good gear when I used film way back when?
I still want that Fuji GFX 100.
Wow that’s wild!! Thanks for doing the research and sharing. My happy with my APSC 40mp. Maybe someday I’ll hop into film. 😊 and I’m glad to know I won’t have to worry about quality that much.
Very interesting. Thanks, I learned something
I think that results are more important than theory. A comparison, depending on the method, will be the final word.
A pixel will never be smaller than film grain, grain is non directional unlike pixels in a grid. You will always need more pixels than the two for anything at angles to match film.
I once took a picture in Germany back in the 80’s that had a small plane visible flying in the distance. It was very small in the photo and looked very nondescript. I used a Contax RTS2 camera with Zeiss T* lenses and the film I used was Kodachrome with a very low ISO (25?). Anyway, I got curious so I looked at the original slide under a microscope and could make out the planes color scheme, the numbers on the fuselage and the word Osterreich on the wing underside telling me it was Austrian. At that magnification it was very grainy but still legible. I was gobsmacked at the time.
Film really is incredible compared to digital. Of course it’s inconvenient, expensive and more difficult to use as well as more time consuming, but little can really compare when you need the max.
I have lots of film negatives from the 1950's when my dad had a 70mm camera. I have a black and white 70mm negative that I used a Cannon desktop scanner to scan the negative at 9600dpi. A car in the picture that is more than 100 feet off into the distance, you can enlarge the scan of the negative enough to easily read the license plate of this vehicle 100 or so feet away.
I highly doubt you can do that with digital photography.
Why stop where the making proceeds: the technique of the enlarger in the darkroom! It is important to understand that this step also reduces the resolution because here also a lens is involved. So with historic analogue testreports showing that lenses with Technical Pan often have a max. of 130 line paires/mm (prime lens, center, 80 in the corner), the real result comes from multiplying the steps in the chain. At the end approx. 50 linepairs/mm remain, which explains the turnaround when this became available in digital photography. Moreover the incredible ISO-performance in digital photography is a revolution. I remember testing the "gamechanging" Konica 35 3200 film. Now what a mess it is compared to Sony A7Riii at 25600!
This is why film is so beautiful. I miss it so much.
Which stock and exposure affects the outcome a lot, but for 35mm I’m going to guess somewhere in the range of 3K - 5.5K.
It gets even worse when it comes to optical magnification (in the darkroom and elsewhere) as the effective aperture (and potential imperfections in the optics) further lower the resolution of film. Most magnifications in the lab are done at f8.0 max, f11-f16 are more commonly used. Wider apertures are only used for focusing the image. Now factor in the magnification and you'll end up with really small effective apertures, barely enough to resolve the negative. It's like spreading butter on bread, getting a bigger piece of bread won't get you a thicker spread of butter.
Is it right? Is it wrong? Lots of discussion. But most of us can agree that film can absolutely hold its own in the current day. If you have a good lens and reputable film, the resolving power is quite surprising.
I've been shooting digital for the last 15 years and my only film experience was disposables in the 90s. Recently tried film for that "grainy look", but quickly realized it's actually harder to achieve that than you might expect. Anything under 400iso is impressively sharp. Since then, I've almost exclusively shot film. It's just more fun to work with.
Yeah Bro, excellent !! Keep going.
Resolution or not - I like the process of film, I like projecting my slides and be taken away by large projected and pixel-free image.
I can't help but notice digital suffering from color gamut issues or what's the correct term. You can even see it in the sample you provided at 11:05 where colorful cloud highlights in digital are blown, whilst retained beautifully on film, and I find the digital performance in this regard strongly lacking or even ugly. And shadow detail too - shadows are naturally blue from the sky and it shows on film.
In the mid to late 90's I started getting some of my better photos scanned to the now defunct Kodak Photo CD's (NOTE: not those low res Kodak Picture CD's you goy at Walmart) They gave you a 6mp image and could fit 100 images per CD. The Pro Kodak Photo CD gave you a 25mp image and could only fit 10 photos per disc.
7:14 My guess is that you would start hitting the resolution limit of the lens you're using at that point.
EDIT: Turns out that you mentioned this at the end of the video.
At 5:09 min.: Dots are not pixels. Pixels Per Inch (PPI) for resolution, not Dots Per Inch. DPI is for the ink dots in a printer!
I spent 22 years in the photofinishing business. Analog of course, the argument might be is film better than digital, I shoot and appreciate both for different reasons. However if you scan film with a DSLR or a dedicated scanner you are not using film as it was meant to be. It was and is designed to be optically projected onto photographic papers. I worked in an enlargement lab and I was able to see absolutely amazing images shot on 35mm film and printed as enlargements. I am guilty of scanning my own film images with a DSLR because I love the process and the film look. As I said I also shoot digital with a few different cameras. I guess my point will be shoot with what you like, there is no wrong way to enjoy photography, to each his own.
I have a film to digital system that uses a Sony a6000 (24MP) camera, a bellows, a set of Nikon enlarger lenses, and a dichroic colorhead -- you get the idea. I'm able to see a bit of the grain structure of 35mm film as I convert it. But certainly not the amount of grain I'm able to see with the critical focus scope in my darkroom. I can certainly believe that if I double the pixel count in my camera I will begin to approach the information density of the film. Of course, that's nowhere near the number of pixels I'll need to properly do 120 film. The math is obvious.
But actually the whole system is diffraction limited. In case of 35mm it is somehwere around 40MP. In case of digital medium format (such as gfx) its diffraction limit resolution is around 75MP, and thus justifies the use of 100MP sensor. Even if film itself can have such high resolution, at the end of the day your camera system is diffraction limited due to the size of the film.
After scanning pics with my 24 mpx camera this past winter I really think the limiting power of the lens is the factor. I was scanning everything from 110 to 4x5. Back in the 35mm film days the lens were really pathetic especially wider angle ones used on scenic pics. Just really yuk. Give me my digital camera and modern lens any day. I was using Mamiya medium format too-I would say that they might be equiv to my 25mpx digital camera but one has to get the film developed plus scan it and clean up the dust etc. Just not worth the hassle.
A lot of the confusion in the comments here is basically answered by the last part of the video. Though anyone claiming that 35mm negatives only have 8-10MP of resolution doesn't really know what they're talking about. Many commercial film scanners targeted 4000DPI, and they did so for good reason: given the average resolving power of the film and lenses available, 4000DPI will resolve most of the detail in most 35mm negatives. That being said, there are exceptions. I have seen enormous prints from 35mm black and white negatives, maybe 4' x 3', that most people would guess were from 8x10 negatives. With very high resolution film, a lens with a high resolving power and a still subject even 35mm negatives are capable of achieving incredible detail. When we start talking about medium format and above you get into the territory where even a standard negative out-resolves even the best professional digital cameras, even on color negative film which tends to be the lowest resolution.
It always the same with these videos, posts, etc. Yeah you can technically scan a negative at an arbitrary resolution let's says 80 Megapixels but you are not obtaining any real detail. For color film typical resolution is around 6 Mp and AT BEST 12 Mp and even there we are talking slow film, a sharp lens stopped down and no camera and subject motion. I have to use at least ISO 3200 in my 15 years old D3s to start losing the same detail compared to an ISO 50 slide film! Don't get me wrong I love film, but resolution is not it's strong point 😅.
It’s important not to conflate the resolving power of film with perceived sharpness. The contrast of low to mid spatial frequencies play a critical role in perceived resolution, resolving power plays no role at all, as it’s typically beyond human visual acuity at regular viewing distances. Color negative typically employ development inhibitor releasing couplers to organically produce an unsharp mask on high contrast edges to boost acutance, despite this the cyan record often drops in modulation at very low spatial frequencies, so perceptually will appear lower resolution compared to say a digital sensor even if the resolving power of film is greater.
Interesting point of view but also, I think there is a practical and historical reason.
Those of us who lived through the "painful" transition from analogue to digital photography saw the first digital cameras as clumsy, judging by the first results one could go to bed and sleep peacefully knowing that the film had a long life.
But soon we began to see results from cameras that with 4MP already looked very good, even in large enlargements. And let's just say that with cameras like the D700 or Canon 5D there was definitely no point in hanging on to film. The resurgence of film has to do with reasons other than resolution power or colors. And I have the doubt if it is purely a whim of fashion or there is a certain issue of being fed up with the already disgusting manipulation that photography suffers in the hands of editing to impress.
@@mistergiovanni7183 I agree! People tend to go back for obsolete things like film, vynil, analogic watches, etc maybe because what we have now is "too perfect".
Resolution CAN be its strong point if that's what you're after. It takes specialty products and processes though. On the extreme end of HD analog, you can pack 30 portraits onto a drop of emulsion just one square mm big. A percieved resolution of around 1 - 1,5mp per square mm.
No digital sensor can match that.
While an excellent explanation and thoroughly researched, I believe this would only hold true in theory. Even in instances were 35mm photo film was used to photograph test charts, true resolved resolution will be much less than you claim.
And you grap this out of what thin air?
Great video - thank you. The "resolving power" of film vs digital is a question that many photographers have asked, but received different responses. This video does an excellent job answering the question.
7:30
Well ACKSHUALLY the relation is not exponential, it scales linearly with the size of the film frame.
This was chad level. Thank you very much
I scan my negatives with a macro lens. 8x10 can resolve about 1000MP maybe a bit less at 800MP and 4x5 can resolve 150MP. Monochrome Ilford Delta. 35mm has very low resolution so I never use it. Medium format also has quite low resolution. You get mich higher resolution with a macro lens than any normal scanner (drum scanner is probably better)