It's pretty amazing the digital-equivalent resolution that actual film has (depends on the film size obviously but if you were to compare average film size to a average 10+ year digital camera it's a huge difference)
It is pretty amazing how well those sources have held up as other technology advanced.
Based on some googling, various sources indicate that 35mm film has a usable resolution somewhere between 4K and 8K video, so we're arguably reaching the limits of what we can extract from it (without consideration for "AI upscaling" and such).
Cinema grade digital cameras, like the RED V-Raptor[0] (MKBHD behind the scenes[1]), can now shoot 8K footage at 120+ fps with 17+ stops of dynamic range. As far as I can tell, those specs are objectively more capable than what you can get with traditional 35mm film. It has taken quite awhile for digital to outclass film across the board, but I think we're at that point now, and the results from these cameras are spectacular[2].
At this point, it's probably a question of how much storage you want to use and whether you have enough light in each scene to shoot at high frame rates like that. (120 is an even multiple of both 24 and 30, so you can always produce 'cinematic' frame rates just by throwing away other frames, without any stuttering, but then you have the option to remaster into higher frame rates in the future if low frame rates fall out of fashion, and you can easily add slow motion effects in post, as long as the final frame rate is intended to be less than 120.)
I'm far from a videography expert, but it is something I find interesting.
You cannot just get "cinematic" frames by using every fifth frame. The shutter speed at 120 fps is much higher than shooting film at 24 fps hence a lot less motion blur that is part of the "cinematic look".
The shutter speed and frame rate are not necessarily linked. Unless you’re using an ND filter (or an undesirably high f-stop) specifically because you want to be able to have more motion blur, most well-lit scenes would surely be blown out at 1/24 of a second (or the usual 1/48 or 1/60), even at the lowest ISO, so digital cameras compensate by shooting individual frames at higher shutter speeds, reducing or eliminating motion blur anyways.
Under optimal conditions, maybe you’re right… but I would personally lump the desire for tons of motion blur in with the nostalgia that causes people to use 24 fps in the first place.
It’s not like people originally wanted to shoot at a noticeably low frame rates… it’s just what they had to do. Then it became a standard that resisted change. Now people artificially restrict themselves to be bug-for-bug compatible with old technology. In fact, a lot of silent films were shot at 16 fps. Why does no one clamor for the return of 16 fps? Arguably, 16 fps is 33% more cinematic!
There are plenty of reasons that I’m not a professional cinematographer… but for the same reason that no one would prefer to watch a film captured in 10 fps, it follows logically that 24 fps is not actually “better” than higher frame rates. It’s just what people have been taught to see as better through experience when they contrast traditional, high budget films shot at 24 fps with low budget TV shows that were broadcast at 60 fps. It’s probably going to be decades before people unlearn this low frame rate preference, but I predict people a hundred years from now will be far less impressed with 24 fps footage than some people today.
I have plenty of other unpopular opinions available too. :P
Regardless, it doesn’t seem beyond belief to imagine that someone could combine the 5 frames of 120fps -> 24fps into individual “long exposure” shots that produce a similar motion blur effect as a single frame taken with a slower shutter. The necessary data is (mostly) all there, if someone took advantage of it. A well-proven technique similar to this is used in astrophotography to create artificially longer exposures, but it is combined with an alignment step to avoid the motion blur of the Earth spinning relative to the stars, which is why astrophotographers don't just extend the length of the exposure, and why they bother with combining multiple exposures. Obviously, applying this technique to create motion blur would mean skipping the alignment step, at a minimum, but this is probably one of those things that would be relatively simple for a properly trained neural network to do a good job with smoothing out, to avoid the gaps of motion blur between the frames that are available... each of which would likely be individually shot with a shutter speed faster than 1/120 anyways.
One downside of the incredible resolution of the film and the resulting transfer is that you can see that one or more of the cameras was perpetually out of focus through the first couple of seasons.
As a counter-point I think it looks worse in many ways. You can see the screen make-up jarringly clearly. And the special effects were added after filming, so they have had to be redone.
Spot on. The film can also be cleaned, restabilized, and restored using certain chemical agents but it's a risky procedure so most of the remastering is done post-process digitally after importing it using better scanners like you mentioned.
Film grains will always set a limit for how much resolution we extract from old film but it's really high. Current methods yield 4K to 8K from 35mm film but odds are with better interpolation technology that understands the interaction of light with the random spacing of the film grains, we could probably get something on the verge of 16K. No one has 16K TVs though, so it's a rather pointless exercise right now.
Still makes me curious, how much information is actually lost versus just computationally/physically obscured..