Resolution of 35mm film and 70mm IMAX film?

[Luba C]

In digital terms, what's is the theoretical resolution of 35mm film and 70mm IMAX film? Just curious . . . I know the best we can see is 4k because the bottleneck is the digital projector which is now only 4k.

 

[Graeme Nattress]

In projection of print terms, 35mm is sub 4k, and dependent upon actual print and projector circumstances sub-HD. A typical cinema projection of a distribution film print will offer no better resolution than an HD digital projection, probably less.

Graeme

 

[David Mullen ASC]

It's easier to just talk about the resolution on the original negative and avoid bringing in the resolution of various printing and projection methods, etc. Red has tested Super-35 and generally found it to be, I forgot, 3.2K or 3.5K, something like that.

You could therefore say that if a 24mm wide piece of film negative resolves 3.2K, for example, a 36mm wide piece of film (VistaVision) would resolve 4.8K, and a 52mm wide piece of film (5-perf 65mm Super Panavision / Todd-AO) would resolve 6.9K, and a 70mm wide piece of film (15-perf 65mm IMAX) would resolve 9.3K. However, this ignores some real-world issues like the fact that older medium-format lenses used on large format movie cameras have a lower MTF compared to modern 35mm cine optics (because the larger negatives don't need lenses with high MTF's because if you have more millimeters overall, you don't need to resolve as many lines per millimeter...)

If you really want to be crude, you could say that you lose maybe half the resolution of the negative once it is printed through dupe elements and thrown onto a theater screen, which is why 2K projection seems on par with the best 35mm print projection, and 4K projection would be similar to 70mm projection, but it therefore also means that IMAX digital projection should be at least 6K...

Now don't confuse measurable resolution with optimal scanning, mastering, and archiving resolution -- if 35mm film really resolves 3.2K, then in reality you really should be scanning it at more like 4K to 6K to avoid aliasing, which is why most people round things off to 4K as being ideal for posting 35mm photography... but perhaps 6K would be better for scanning, and then you should finish at 4K.

 

[Marc Wielage]

Kodak always told me that original 35mm camera negative after 2000-2001 had equivalent resolution (based on line pairs per millimeter and MTF) to right around 6K. But they also admitted that, after going through a traditional IP -> IN -> release print stage in the lab, it goes down to 2K. Given the weave and lack of pin-registration in most projectors, plus the less-than-optimum focus, you could easily wind up with well below 1080 resolution.

I agree with Mr. Mullen that 4K is optimal for the real world. Note also that many (if not most) Imax films involve a 35mm uprez process, bumping everything up to about 6K for film out. They're also using CRT recorders for Imax recording, which have their own issues.

Real Imax productions, like the 65mm sequences shot for Dark Knight, are very hard to beat, especially when you're talking a Vision Premiere print struck right off the camera negative. That's about as good as it gets for photochemical acquisition, better than any existing digital camera or projector on earth. It's a good goal to shoot for.

 

[Dan Hudgins]

Image vs. media

Image vs. media

You can check the Celco site for the scanner resolutions:

http://www.celco.com/FormatResolutionTableKodakCineon.asp

Quote: [STANDARD 1.85
.864 X .467

1.85

3656 X 1976(2)

1828 X 988]

Since digital images can resolve about 0.707 their pixel count you get at best a visual resolution without aliasing of:

1828x0.707 = 1292

988x0.707 = 698

Which correspond to tests done that show that viewing of 35mm film in theaters results in about 1280x720 +/- 25% or so visible resolution from test charts read off the movie screen.

The peak resolution of the film does not matter since the resolution of film varies with exposure and density and can be much lower than Kodak's specs would make you think in the highlights and shadows since they use methods that do not reflect what can be seen in projection contrast as projected.

If you take the image area and use a typical f/1.7 projection lens having 20lp/mm at high MTF you get:

.864 X .467 = 21.9mm x 11.8mm

21.9 * 40 = 876 lines per image width (*1.4 = 1226 pixels needed)

11.8 * 40 = 472 lines per image height (*1.4 = 660 pixels needed)

Those values are close to what you can see off a 4th generation print at high MTF under good conditions.

In today's theaters there is no projectionist to keep the projector in focus during the film projection as there was in the past and 35mm film projection with an f/1.7 lens needs its focus adjusted every 8 minutes or so to keep the maximum lens resolution, otherwise the focus can drift even lower than 20lp/mm at high MTF.

35mm film is a bit better than SD, which is why VistaVision and 65/70mm were developed.

The image area for IMAX is given on this page,

http://www.celco.com/FormatResolutionTable4K.asp

Quote: [65MM 15 PERF (IMAX/OMNIMAX)
2.772 X 2.072

1.34

4096 X 3060

2048 X 1530 ]

2.772 X 2.072 = 70.4mm x 52.6mm

70.4 * 40 = 2816 lines per image width (*1.4 = 3942 pixels needed)

52.6 * 40 = 2104 lines per image height (*1.4 = 2945 pixels needed)

Those would be lines of actual visible resolution at high MTF not scanner pixels that don't relate to the image itself, when the print is made by contact or in a good optical printer. Both contact and optical printing have their own problems as only someone who has worked with them can know. Its much harder to print large film sizes because its harder to hold the film flat enough over that larger area.

==

In previous times the projection lens was a 4 element design, not a 6 or 7 element one like used today, and would have even lower than 20pl/mm resolution at high MTF in the corners of the frame because of the curved field. Older theaters were longer and had smaller screens, but the 4:3 about ratio projection had more image height than today's wide screen prints 1.85:1. Some theaters still use older projection lens types which acted as a low pass filter to reduce the grain in the films of the past.

Using figures that represent less than 2% MTF as "resolution" give an impression of much more visible resolution than one would expect to see. Using a MTF value of 30% is a more useful value I think as far as being sure that the audience can see details of that contrast.

Another factor is that moving images printed film to film did not have a fixed raster like digital cameras and projectors, so the eye averages grains from frame to frame and so when you watch movie film projected at speed your eyes pick out details in the grain and blur as the frames weave by, so you can see a bit more detail in moving images than looking at a freeze frame.

If someone tells you that their projection lens can project 180lp/mm at high MTF off a film print onto a movie screen for more than a few seconds without adjusting the focus they should be considered of suspect experience. The film cannot stay in the focal plane within those limits.

The dye bleed in film stocks at higher exposure levels can reduce the 80% MTF resolution to 10lp/mm or less. The resolution of film varies with exposure, density, processing, and contrast of adjacent image points (edge effects). If you step contact print without an air gap you get Newton's rings (I think Kodak just changed the surface finish on their color print stocks to reduce the Newton's rings, very smooth print stocks show those more.)

To convert film resolution lines per values to raster pixels needed you need to multiply by at least 1.414, such as scanning at 6K to try to capture 4K off the film because of the anti-aliasing filters needed in a scanner, and the aliasing issue in general with all digital images using a grid pattern.

 

[Robert Niessner]

Well, I am all for digital, but I have to say that I am not very happy with 2k projections. Sitting at the ideal position in a cinema, which is normally in one of the middle rows, a 2k projection is - to my eye - not sharp enough. As soon as titles blend in it gets even worse. The problem is not the 2k footage but the 2k projection. 2k footage would benefit well from beeing projected in 4k, smoothing out the stair steppings and decreasing pixels on the screen. I am really looking forward when cinemas start adopting better projectors.

Back to the question of the topic starter: At the end of the nineties I worked in highend scanner development and saw an awful lot of scans of 35mm. Our scanner had an optical resolution of 5080 dpi, resulting in a 4360 x 3720 px image (Super 35 2.35:1). But you could see the film grain very well at that magnification. From that experience I would say that 35 mm is not able to give you any more resolution beyond 4k.
The highest resolution I ever got was from a 4 x 5 still image which was shot under perfect studio conditions - very fine grain and details up to the full resolution the scanner was able to deliver.

 

[Keegan Flynn]

Well, I am all for digital, but I have to say that I am not very happy with 2k projections. Sitting at the ideal position in a cinema, which is normally in one of the middle rows, a 2k projection is - to my eye - not sharp enough. As soon as titles blend in it gets even worse. The problem is not the 2k footage but the 2k projection. 2k footage would benefit well from beeing projected in 4k, smoothing out the stair steppings and decreasing pixels on the screen. I am really looking forward when cinemas start adopting better projectors.

Good point. I've seen plenty of 2K screenings at my local cinema (not sure what kind of projector they use), but another one near me just upgraded to the Sony 4K projectors (that Sony is giving away like candy) and the difference is definitely noticeable, even though both are running 2K DCPs. I'd kill to be able to send some true 4K footage through those things.

 

[Peter II]

Color Gamut of Film vs. 4K

Color Gamut of Film vs. 4K

Hi Guys,

I know this is an old thread, but wanted to thank you for your insight. Some of you are like technical film gods to me.

One question for which I cannot find the answer is the color gamut of film vs. the new 4k WCG (Rec. 2020). Any insight would be appreciated.

Thanks

 

[Nick Morrison]

It's easier to just talk about the resolution on the original negative and avoid bringing in the resolution of various printing and projection methods, etc. Red has tested Super-35 and generally found it to be, I forgot, 3.2K or 3.5K, something like that.

David, don't you think this makes Arri's decision to make their Alexa sensor resolve 3.4K all the more interesting? The way it lines up with film is pretty uncanny, and I suppose no surprise - they've always been pretty candid about trying to make a sensor that emulated film as closely as possible...

 

[Phil Holland]

Somehow a film scanning related thread that I haven't replied to

One question for which I cannot find the answer is the color gamut of film vs. the new 4k WCG (Rec. 2020). Any insight would be appreciated.

Film is a relatively limited pallet, though truly contains a lovely aesthetic influenced by it's unique properties. Back in the day we were indeed scanning film in 16-bit, but it would be a hard argument based on measured data that it captures more color than modern digital cinema cameras, by a pretty noticeable margin. Dynamic Range comes into play a bit too, but digital systems are also now past what film could possibly capture. You can certain get a good scan of film and grade to HDR if need be, but it's not the best solution IMO.

I've written it elsewhere on REDuser, but Kodak's hard number on the latest stock is 13 Total Captured Stops of Dynamic Range. In rather controlled scenerios you can likely squeeze out 15 stops, but nobody shoots that way for sure.

In the RED ecosystem, particularly from Dragon on we have been outresolving, capturing higher dynamic range, more captured color, and certainly more light sensitivity than motion picture film. For other manufacturers, sans resolution, they have approximately hit or exceeded the film mark as well.

David, don't you think this makes Arri's decision to make their Alexa sensor resolve 3.4K all the more interesting? The way it lines up with film is pretty uncanny, and I suppose no surprise - they've always been pretty candid about trying to make a sensor that emulated film as closely as possible...

Nothing at all uncanny about it Nick. Back in I think December of 2000, 1080p became a digitally broadcast standard and it was pretty clear to the industry that's where High Def was going on the consumer level. The focus on 2K and 1080p became a very influencing directive. Outside of that a great deal of experimentation and exploration was occurring with digital and analog. Film Scanning was a no-brainer, particularly early on with the VFX side of things. In the mid-90's the 4K film scanner was invented and it's what I was fortunate to work on until better solutions became available. Particularly from Northlight and Imagica with their 6K and 8K scanners respectively.

We sided with Northlight for Full Aperture Super 35mm scanning and focused on the 6K oversampling after a great deal of testing. The advantages of 6K scanning produced a grain structure that was higher resolution than a pixel and also helped minimize artifacts as well as producing a higher quality output to 4K and 2K when downsampled.

As noted above, a release print of a film often was hitting around 2K and certainly wasn't what was captured or even seen in dailies when all cylinders were firing well and you had a print hit the general LAD target.

Arri's focus early on was to create a system that could appropriately create a compelling 2K finish. In the earlier days of what become known as Digital Cinema Cameras, in 2005 Arri produced the D-20 with a capture resolution of 2880x2160. Same happened again 2008 with the D-21 with the same resolution. And yet again in 2010 with the Alexa, all fairly focused on that 2880x2160 resolution. Eventually new models and Open Gate became a thing to allow up to 3414x2198.

This in itself is where RED saw the opportunity. In looking at Arri's offerings, the 1080p Panavision Genesis, HD Thomson Viper, and various other systems; there was a notable cap between what Super 35mm Motion Picture Film was "as a medium" versus what exactly was happening with Digital Cinema Cameras at the time. There were a few other solutions at the time, but none gained commercial or real industry adoption and most who had experienced film on a post production side of higher resolution scanning and even laser recording "knew what was there".


Back to film resolution. As you know, generally it is based in general on the size of the format size. In reference to the film resolution above from Celco, my measured data from the 2000s lines up pretty damn tightly with their's. When you use the formatCompare Tool at PHFX | tools and select a given film format it is giving you accurate data on well shot/well scanned/well produced potential resolution of each format.

The two important ones for 35mm are Super 35mm Full Ap (about 4096x3072 total capture resolution) and Super 35mm Academy (3620x2632).

Beyond that, things don't scaled too directly to the larger 70mm 15-perf systems which should have the potential of about 11586x8658 resolution. However, the difficulty is the vibration when filming at speed combined with the large surface area of the medium. It typically takes a noticeable resolution hit, though still has the advantage of smaller grain particles. Typically 70mm is actually at maximum scanned at 11K today. Most of the time it resolves somewhere between 5.7-6.5K. That's actually all good news as it still leads to a stunning finished image.


Just looping around to some general theory, the concept of starting with more and refining down or even losing information along the way to the finish has always been a mainstay in film production which is initially why RED captivated me so much early on. In truth with MX and Mysterium-X reveal so much of the potential, but Dragon really tipped that over the edge for me. We're now on a newer generation of sensor tech like the large format multi-resolution/multi-tasker Monstro, Dragon still exists because it is that good, Helium which on a format level outresolves most S35 systems, and Gemini which sees into the darkness. In some ways you can look at RED's offerings as simllar to different speed films or available format sizes. The 8K Monstro VV for me is so enticing as you can shoot and produce proper VistaVision, Super 35mm, and even Super 16mm all in one camera. However, at this point I have been mostly seduced by VistaVision as the format is attractive for my work and still love shooting Super 35mm when the calls come.


I'll reiterate, film is absolutely beautiful towards what it does best. But as a capture medium we have certainly surpassed it. Especially on a pure image control level as we are not dealing with print variation, chemical baths, etc....

 

[Nick Morrison]

Somehow a film scanning related thread that I haven't replied to



Film is a relatively limited pallet, though truly contains a lovely aesthetic influenced by it's unique properties. Back in the day we were indeed scanning film in 16-bit, but it would be a hard argument based on measured data that it captures more color than modern digital cinema cameras, by a pretty noticeable margin. Dynamic Range comes into play a bit too, but digital systems are also now past what film could possibly capture. You can certain get a good scan of film and grade to HDR if need be, but it's not the best solution IMO.

I've written it elsewhere on REDuser, but Kodak's hard number on the latest stock is 13 Total Captured Stops of Dynamic Range. In rather controlled scenerios you can likely squeeze out 15 stops, but nobody shoots that way for sure.

In the RED ecosystem, particularly from Dragon on we have been outresolving, capturing higher dynamic range, more captured color, and certainly more light sensitivity than motion picture film. For other manufacturers, sans resolution, they have approximately hit or exceeded the film mark as well.




Nothing at all uncanny about it Nick. Back in I think December of 2000, 1080p became a digitally broadcast standard and it was pretty clear to the industry that's where High Def was going on the consumer level. The focus on 2K and 1080p became a very influencing directive. Outside of that a great deal of experimentation and exploration was occurring with digital and analog. Film Scanning was a no-brainer, particularly early on with the VFX side of things. In the mid-90's the 4K film scanner was invented and it's what I was fortunate to work on until better solutions became available. Particularly from Northlight and Imagica with their 6K and 8K scanners respectively.

We sided with Northlight for Full Aperture Super 35mm scanning and focused on the 6K oversampling after a great deal of testing. The advantages of 6K scanning produced a grain structure that was higher resolution than a pixel and also helped minimize artifacts as well as producing a higher quality output to 4K and 2K when downsampled.

As noted above, a release print of a film often was hitting around 2K and certainly wasn't what was captured or even seen in dailies when all cylinders were firing well and you had a print hit the general LAD target.

Arri's focus early on was to create a system that could appropriately create a compelling 2K finish. In the earlier days of what become known as Digital Cinema Cameras, in 2005 Arri produced the D-20 with a capture resolution of 2880x2160. Same happened again 2008 with the D-21 with the same resolution. And yet again in 2010 with the Alexa, all fairly focused on that 2880x2160 resolution. Eventually new models and Open Gate became a thing to allow up to 3414x2198.

This in itself is where RED saw the opportunity. In looking at Arri's offerings, the 1080p Panavision Genesis, HD Thomson Viper, and various other systems; there was a notable cap between what Super 35mm Motion Picture Film was "as a medium" versus what exactly was happening with Digital Cinema Cameras at the time. There were a few other solutions at the time, but none gained commercial or real industry adoption and most who had experienced film on a post production side of higher resolution scanning and even laser recording "knew what was there".


Back to film resolution. As you know, generally it is based in general on the size of the format size. In reference to the film resolution above from Celco, my measured data from the 2000s lines up pretty damn tightly with their's. When you use the formatCompare Tool at PHFX | tools and select a given film format it is giving you accurate data on well shot/well scanned/well produced potential resolution of each format.

The two important ones for 35mm are Super 35mm Full Ap (about 4096x3072 total capture resolution) and Super 35mm Academy (3620x2632).

Beyond that, things don't scaled too directly to the larger 70mm 15-perf systems which should have the potential of about 11586x8658 resolution. However, the difficulty is the vibration when filming at speed combined with the large surface area of the medium. It typically takes a noticeable resolution hit, though still has the advantage of smaller grain particles. Typically 70mm is actually at maximum scanned at 11K today. Most of the time it resolves somewhere between 5.7-6.5K. That's actually all good news as it still leads to a stunning finished image.


Just looping around to some general theory, the concept of starting with more and refining down or even losing information along the way to the finish has always been a mainstay in film production which is initially why RED captivated me so much early on. In truth with MX and Mysterium-X reveal so much of the potential, but Dragon really tipped that over the edge for me. We're now on a newer generation of sensor tech like the large format multi-resolution/multi-tasker Monstro, Dragon still exists because it is that good, Helium which on a format level outresolves most S35 systems, and Gemini which sees into the darkness. In some ways you can look at RED's offerings as simllar to different speed films or available format sizes. The 8K Monstro VV for me is so enticing as you can shoot and produce proper VistaVision, Super 35mm, and even Super 16mm all in one camera. However, at this point I have been mostly seduced by VistaVision as the format is attractive for my work and still love shooting Super 35mm when the calls come.


I'll reiterate, film is absolutely beautiful towards what it does best. But as a capture medium we have certainly surpassed it. Especially on a pure image control level as we are not dealing with print variation, chemical baths, etc....

Awesome, thanks for the breakdown Phil. Appreciate it.

 

[Peter II]

Dang Phil, your NYC flyover is just gorgeous! https://www.youtube.com/watch?time_continue=94&v=UN3uF3990Q0

Also, thanks for your input on the color and dynamic range of film. As a consumer, I'm obsessed with "original-ness" and hope to someday own my favorite films in a format that most closely resembles the original source material.

Years ago, just as 4k was just beginning to enter public consciousness, I found a few (now questionable) articles on the resolution of film. In one frequently cited page, Ken Rockwell suggests that you'd need a scanner of about 175 Megapixels to resolve the full detail of the highest-quality film (See here: https://kenrockwell.com/tech/film-resolution.htm). Mr. Rockwell makes a compelling argument for me as a novice, but I think you guys know more than he does. Can you tell me where he's gone wrong?

Thanks,

Peter

 

[M Most]

Every post I see here is missing the very significant fact that film has the same resolution across all 3 color channels due to its physical construction (3 layers, all with roughly the same number of photosensitive grains). Film scanners also yield that because the better ones do 3 passes of every frame. Bayer sensor cameras - be they Arri, Red, Panasonic, Canon, or any other manufacturer - have the limitation of splitting the available pixel sites among 3 channels, with 2 of the 3 getting half the number of sites as the third. So the notion that "4K" in a film scan is the same as "4K" from a Bayer pattern sensor is at the very least inaccurate, and at the very most just plain wrong. Granted, the current 8K Red sensors have more than that, but they don't yield "8K resolution" across all three channels under any circumstances.

 

[Nick Morrison]

Every post I see here is missing the very significant fact that film has the same resolution across all 3 color channels due to its physical construction (3 layers, all with roughly the same number of photosensitive grains). Film scanners also yield that because the better ones do 3 passes of every frame. Bayer sensor cameras - be they Arri, Red, Panasonic, Canon, or any other manufacturer - have the limitation of splitting the available pixel sites among 3 channels, with 2 of the 3 getting half the number of sites as the third. So the notion that "4K" in a film scan is the same as "4K" from a Bayer pattern sensor is at the very least inaccurate, and at the very most just plain wrong. Granted, the current 8K Red sensors have more than that, but they don't yield "8K resolution" across all three channels under any circumstances.

Great thing to keep in mind - thanks.

 

[Graeme Nattress]

Every post I see here is missing the very significant fact that film has the same resolution across all 3 color channels due to its physical construction (3 layers, all with roughly the same number of photosensitive grains). Film scanners also yield that because the better ones do 3 passes of every frame. Bayer sensor cameras - be they Arri, Red, Panasonic, Canon, or any other manufacturer - have the limitation of splitting the available pixel sites among 3 channels, with 2 of the 3 getting half the number of sites as the third. So the notion that "4K" in a film scan is the same as "4K" from a Bayer pattern sensor is at the very least inaccurate, and at the very most just plain wrong. Granted, the current 8K Red sensors have more than that, but they don't yield "8K resolution" across all three channels under any circumstances.

But a 4k scan of 35mm film isn't the same as a 4k rendered CGI image.... MTF of 35mm motion picture negative gives maybe 2100 l/ph for 10% MTF or maybe max limiting resolution of 2400l/ph which for 24mm x 18mm film gives you a "k" of 3.2k max - so the luma resolution will be rather similar to that of a 4k bayer pattern sensor with OLPF.

And of course, that's why 4k for 4k only lasted until better could be achieved....

Graeme

 

[Marc Wielage]

But a 4k scan of 35mm film isn't the same as a 4k rendered CGI image.... MTF of 35mm motion picture negative gives maybe 2100 l/ph for 10% MTF or maybe max limiting resolution of 2400l/ph which for 24mm x 18mm film gives you a "k" of 3.2k max - so the luma resolution will be rather similar to that of a 4k bayer pattern sensor with OLPF.

Kodak color scientist Mitch Bogdanowicz told me about 15 years ago that Eastman 5219 Vision 3 stock could capture the equivalent of 6K, but he didn't give me specifics on how they measured it. He also admitted that a lot would depend on registration, contrast, lenses, and other factors, and that that was a "best case" scenario right off the negative.

 

[Graeme Nattress]

I was going with: http://www.motionfx.gr/files/35mm_resolution_english.pdf for the numbers.

Graeme

 

[M Most]

I won't bother to point out that this paper was written over 16 years ago, was intended to measure performance of an all-film origination through projection chain, and based on film stocks that haven't been manufactured since around 2006.

Wait a minute - I think I just did...... :sarcasm:

 

[Phil Holland]

Every post I see here is missing the very significant fact that film has the same resolution across all 3 color channels due to its physical construction (3 layers, all with roughly the same number of photosensitive grains). Film scanners also yield that because the better ones do 3 passes of every frame. Bayer sensor cameras - be they Arri, Red, Panasonic, Canon, or any other manufacturer - have the limitation of splitting the available pixel sites among 3 channels, with 2 of the 3 getting half the number of sites as the third. So the notion that "4K" in a film scan is the same as "4K" from a Bayer pattern sensor is at the very least inaccurate, and at the very most just plain wrong. Granted, the current 8K Red sensors have more than that, but they don't yield "8K resolution" across all three channels under any circumstances.

This was a few years back, I think around Q4 2015. During Dragon's initial release Kodak was circulating a graphic used by a rather big filmmaker, who I won't name, of motion picture film scanned at 10,000X magnification via I believe an electron microscope displayed in monochrome and compared it against a digital sensor also scanned at 10,000X.

This was used as a tool to explain to filmmakers as to why digital can never top film. Which is sort of a fun graphic showing the grid pattern of the sensor, but literally ignores how a digital image is constructed. The same for film honestly.

I replied with a 10,000X scan of film in color which not only shows a great deal of the silver halide crystals do not contain color, but rather monochromatic information, as well as the fact that their are gaps between the particles within the gelatin substrate.

You can actually even observe this via a standard microscope with higher quality optics, which we did to examine grain patterns between slower and faster stocks, particularly that moment when Vision 3 dropped.

Whenever I write regarding film scanning, it's development, and where it's at in modern times it's always about capturing a better grain particle that's ideally larger than a pixel as it benefits from that. As to what is contained on a smaller level, you can see often where film doesn't contain much information when going down the 6-3μm level.

The Bayer Pattern argument has been made at the studio level, but this is the one thing that absolutely nobody on the film side of things should ever focus on as there are several sensors that clearly out perform the technical merits of film and that lovely 10,000X graphic reveals exactly where this shows.

Which is why when I discuss film, and hell if I was marketing over at the big K, I'd focus much more on the aesthetic and unique properties it brings rather than focusing on digital never performing as good. It's been a few years since that's been a solid stance.

And just to bump my old 2014 tests, which not-ironically was the last time I needed to do anything like this:
Red Dragon ISO Texture Versus Film Grain

 

[Anton Shavlik]

35mm having more than 4k of resolution implies that 16mm has more than 2k, but it very apparently does not. I'm hoping to test exactly the resolution of Kodak's 4 motion picture stocks on 16mm later this year in preparation for a project.

Kodak's own tech sheets show that the stocks have fairly different MTFs with 50D being rather low and 200T being above 100 in some frequencies, I guess meaning some grain patterns develop more contrast than they should. It's a pretty complex question.

Here's an image of a 16mm scan at 4k as well as 2k uprezzed to 4k and honestly you're just getting better MTF on the grain, but the dye clouds are already appearing to be out resolved.