AI made these movies sharper – critics say it ruined them
nytimes.com
nytimes.com
The culprit here is that the non-AI screenshots are taken from presumably 1080p non-HDR sources, while all the AI screenshots are taken from presumably 4K HDR sources. The "AI" images are all displayed in the completely wrong color space -- the dark+green/blue is exactly what HDR content looks like when played on software that doesn't correctly support decoding and displaying HDR content.
It's a shame that the creator of the comparison images doesn't know enough image processing to understand that you can't grab stills from HDR content from a player that doesn't properly support HDR.
On the other hand, the state of HDR support is a mess right now in software players. Playing HDR content in common players like VLC, QuickTime, IINA, and Infuse will give you significantly different results between all of them. So I can't actually blame the creator of the images 100%, because there isn't even a documented, standardized way to compare HDR to non-HDR content side-by-side, as far as I know (hence why each player maps the colors differently).
The NYT isn't free. I don't respect such shoddy feature-writing.
That should change over time: Blu-Ray is a format released about 20 years ago, and our streaming bandwidth (and encoding algorithms) keep improving.
Streaming isn't limited by the available bandwidth but by the margins the streaming service doesn't want to spend on it.
But every time a movie gets streamed, it costs the streamer twice as much if the file is twice as large. So streaming bandwidth will generally be minimized as much as possible to a "just barely good enough" quality level.
It's not about technology improving or how old Blu-rays are, it's about economics.
Compare eg Youtube videos today to Youtube videos 15 years ago.
I see no difference. Their 1080p is still the super-crummy 1080p they had 15 years ago.
In fact, if you pay for YouTube Premium you can get a separate "1080p Premium" with a higher bitrate. But you have to pay. Their free version hasn't gotten better at all. In fact, it's gone the other direction -- they've gotten more agressive about defaulting to 720p or 480p streams when they used to show 1080p by default.
Sure, a tiny handful of videos have 4K versions available, but that's only for a tiny proportion of streams. And YouTube even removes higher resolutions after a video has been published a bit. Which is why you see lots of YouTube VR videos described as 8K but there's no 8K stream anymore. There was when it was uploaded, but YouTube removed it.
So, no. Even if bandwidth gets cheaper, streaming twice the bits is still twice as expensive. Streaming services aren't going to stream more bits -- they're just going to take more profit, as long as it's still minimally good enough for the average viewer.
Oh, 1080p indeed been introduced in 2009, eg 15 years ago. I thought it was younger. However, just pretend I was talking about the next higher resolutions, or 60fps.
> [...] they're just going to take more profit, as long as it's still minimally good enough for the average viewer.
Competition will take care of that extra profit.
No it won't. I said "minimally good enough". Competition takes care of things up to "minimally good enough". And then it stops.
That's my point -- bandwidth getting cheaper is not leading streamers to increase the quality of their 1080p. Because not enough people care. Not enough people will switch services. Competition ceases because it's not a significant factor of differentiation.
(For the niche people that do care, that's why YouTube premium exists, but it's niche.)
Huh, why would it stop? People can freely switch between streaming services.
> Because not enough people care.
Yes, competition improves what people care about. That's a good thing. You don't want society to waste resources on stuff people don't care about.
Btw, your argument might demonstrate that video quality won't go up, if people don't care. Yes, I agree. But what does that have to do with profit?
People always eg care about price; and they will switch for price.
The cheaper bandwidth becomes, the more it makes sense for streaming services to give people what they want to pay for.
More generally, automatically second guessing the artistic choices of whoever originally color graded the film, for the sake of adding narratively gratuitous features like HDR or extremely high resolution, is a nasty thing to do. There might be moderately more dynamic range in the original physical film than there was in a digital copy; if so, by all means try to capture it in a new digitization while trying to match the visual impression of a theater projection to the extent practical. The "AI" features demonstrated here are incredibly tacky though.
People put in charge of making new versions of beloved artworks need to have good taste and exercise some discretion. Art from different times and places looks different due to changes in both the medium and the culture, and we should enjoy those differences rather than trying to erase them all.
If people want to AI-enhance movies for their own viewing pleasure, then great. Watch it in reversed color or with a face-swap for all I care. But "improving" the original by inventing detail that the director probably never wanted to show to begin with is a parlor trick devoid of any artistic merit.
I will submit that directors are not generally the right people to call the shots here. A director can never watch his own movie the way the audience does. They always see an unfulfilled vision, a compromise being made here and there because of technology being limited, and some like Cameron or Lucas will make misguided attempts to modernize to make up for it.
This disregards that Aliens will always be a goofy 80s movie and not a movie from them 2020s. Its look should be preserved, be that the film grain, or its color grade. At least, the few people who still buy optical media should be given the choice, as they are the most likely to care.
It contains very little of humour, all the cast and the director are taking it seriously. I'll buy Goonies as a goofy film, but Aliens? I remember someone glued to a wall and screaming in agony as a creature tears its way out of her chest. I remember the ratcheting tension of the motion-detectors getting more and more high pitched as nothing is happening but we're expecting it any moment now.
I know there's some don't-think-about-it plot-holes, but goofy?
Ok, fair. James Cameron can do whatever he wants with his films. They are his artistic output, and if he feels the technology available at the time didn't properly communicate that, he can certainly work to mitigate that however he wants. I assure you, this is not common among directors. If you invited Ridley Scott and Derek Vanlint to a meeting informing them that Alien was being re-released as an AI "enhanced" movie, I sure as fuck wouldn't want to be in that room.
> I will submit that directors are not generally the right people to call the shots here. A director can never watch his own movie the way the audience does.
Ha. I'm a Tech Artist and do VFX, so I have more visibility than most into how a lot of these decisions get made. I'm not trying to be a jerk, but of all the confidently incorrect declarations about art I've seen in tech fora, this has got to be in the top 5.
To some extent, the misconception is understandable from a consumer vantage point: without spending a lot of time creating images from nothing to have a specific impact on the viewer, you really don't understand how many decisions there are to make about minutiae, and how much it all affects how the work hits. Prompt-based AI image generators are a good example of this. Their making these decisions is the real boon for non-artists that want to make images, and also why they're non-starters for high-level professional deliverable work. Learning to use art tools to produce clean images is the easy part. The hard part is deciding at an extremely granular level what those images should contain. Since the models amalgamate those decisions from the millions of decisions that artists have already made in different images, that lets end users indefinitely create visually compelling images without having to know why they're compelling at a granular level. But the decisions at that granular level communicate a lot to the viewer, and even with the finer-grained control you can get with prompt manipulation, control nets, inpainting, loras, etc. it's still bumper cars vs driving. Famous artists are usually famous because their vision isn't just a mix of what other people have done before them-- it's got other frameworks, stimuli, references, connections and influences that are implemented in novel, complex, and very specific ways. Fighting against the amalgamated decisions of every artist that contributed to the model to achieve those results requires so much more effort than just making it from scratch. That's why generative AI in industry tools like Nuke don't integrate prompts or generate direct visuals-- they do things like make masks for compositing. A huge win, but 100% not creating images that go on screen. On a much more extreme level, Disney could not have trained a model with the entirety of early animation and pre-1937 non-animated visual art-- even their own films-- and come up with The Old Mill, no matter how much engineering they did.
In a film, almost all of those decisions made by an AI model when generating images are made, or at least finalized, by the director, often in conjunction with the DP and Production Designer. Most of them are obsessed with things like the slope and toe of film, grain, minute differences in very specific color representations, etc. etc. etc. and it affects nearly everything about the film-- set decisions, framing to get more or less of something, what time of day to shoot at, lighting, grading, etc. etc. etc. Saying directors shouldn't have complete control over the medium on which their movies are captured betrays a complete lack of understanding of making movies, both as an art form, and a practical process. If you want to learn more about how movies are made before you pontificate about the craft of making movies, I recommend looking at StudioBinder's youtube channel for a sub-cliff's-notes level introduction to how it's actually done.
State of the badass art.
This is the original sin of this process. They are tampering with the artistic object, it's no longer what the artists intended.
Some might say this is being pedantic, but the quality of the image has a huge impact on the feel of a movie. For some films, such as Blade Runner, the mood and feel created by the (dark, obscured) look of the film is easily half of the impact. Changing that film would be a crime against humanity, yet I imagine it's only a matter of time before it gets regraded so people can see the film "properly".
Hardly pedantic, the quality of the image is one of the key variables in cinema.
The first time I saw Blade Runner was on a CRT TV from VHS tape and it was still faithful to the original, which I've seen on the big screen and various digital formats too. Poor reproduction doesn't add something that isn't there in the original, it just degrades the image. That generally can't be controlled and is a lot different to intentionally changing the image.
It quite literally does
>That generally can't be controlled
Yes it can, just don't make any VHS versions
But then im also of the opinion that most HDR content is too damn dark anyway.
It's not, though -- it's entirely the problem with players. Not just software players but also TV's.
There are lots of TV shows now that are available in 1080p SDR and in 4K HDR. When you play them both on any player or TV, they should have the same brightness in "regular" scenes like two people talking in a room. They're meant to. HDR is only meant to make a handful of elements brighter -- glowing lightsabers, glints of sunlight, explosions. HDR is never meant to make any of the content darker.
Unfortunately, far too many video players and televisions (and projectors) map HDR terribly. The TV's aren't actually meaningfully brighter, but they want to advertise HDR support, so they map the wider HDR brightness range to their limited brightness, so that lightsabers etc. get the full brightness, but everything else is way too dark. This is the disaster that makes everyone think HDR content is too dark.
When the correct thing to do would be to display HDR content at full brightness so it matches SDR for normal scenes, and then things like lightsabers, glints of sunlight, etc. just get blown out to white. To admit the hardware has brightness limitations and should never have been advertised as HDR in the first place.
So the problem isn't with HDR content. The problem is with HDR players, TV's, and projectors. It's mostly on the hardware side, with manufacturers that want to advertise HDR support when that's a lie, because they're not actually producing hardware that is HDR-level bright enough.
I have to do some math to check if this matters or not, but HDR is able to make dark scenes much better too. SDR video doesn't have nearly enough bit depth to represent anything dark well (it's not even 8-bit) whereas 10bit HDR does. So I would expect to see more dark scenes in properly displayed HDR.
Also, in dynamic HDR you should be able to make a dark scene by encoding it brighter and using the metadata to darken it, though I don't remember if it actually works in practice or if anything does it.
Step ONE every time I buy a TV is to spend a half an hour or so using tons of calibration images to get it dialed in to a sensible setting, and it's shocking how much a lot of them need changed. Almost every TV comes calibrated to be used as a display unit in the store which means contrast is rammed through the roof, brightness is way up, they usually have about 15 different kinds of smoothing or frame interpolation going, "denoisiers," color filtering for some insane reason, it's fucking ridiculous.
This is honestly why most of the time I buy cheap ass off-brand TV's because there's just less bullshit to turn off when the TV has "fewer" features.
Even just talking about SDR for the moment, films and TV shows have a lot of work put in so that their color, brightness, sharpness and contrast are just right.
And then modern TV settings apply all sorts of filters that just destroy it.
It's bizarre. I do the same thing you do in terms of calibration and it takes forever. All just to get it to act like a dumb monitor for my Apple TV. I just want it to display the signal and nothing more.
So many people are watching TV and films with weird color, weird contrast, and weird swimmy frame interpolation. And that's before you get to HDR where it's weird dark on top of it all.
It genuinely makes me want there to be either a government regulatory body or else some kind of private-organization seal that guarantees you get a normal-color, normal-contrast, normal-brightness, normal-motion image as filmmakers and TV producers intend. You can have all the other options too, but either make them non-default, or provide a single-click "director's choice high-fidelity (TM)" option to undo all the garbage they add. Or something like that.
Because it's completely out of control.
It's not even possible to turn off in some - you have to install a "dumb" camera app.
Eh. Usually you can flip it to movie + warm(2) + dynamic contrast off + enhancements off (think noise reduction) and you’re 90% there. You can get a smidge better picture with calibration images, but most midrange+ TVs come decently calibrated out of the factory these days. Not really worth the effort unless you go all the way and use a colorimeter.
E.g. on my Epson projector, there's a "sharpness" setting that goes from 1 to 10, and IIRC the only way to turn off sharpening is to set it to 3. Because values 1 and 2 wind up applying a blur filter. It's not documented. You have to figure out the "3" value from trial and error with a calibration image.
Similarly you need to figure out whether each of your connected devices is outputting RGB 0-255 or RGB 16-235, and set the toggles for your TV to handle the correct input range.
And so forth. The calibration images aren't for figuring out perfect color accuracy with a colorimeter. They're for basic things like displaying smooth gradients from black to white, and are the blacks or whites being cut off? Is the gamma totally off when compared to a checkerboard? Are saturation gradients displaying normally or does it show the image is over-saturated? Is a series of alternating black and white lines displaying as simple lines, or is there a crunchy halo around them (from sharpening) or blurring?
So it's not "eh" at all. In my experience it takes 15 to 30 minutes, as you hunt around the submenus to try to find all the relevant settings, and try to figure out what each setting even means since the documentation is useless. Like, on a power-saving display, should the main color setting be "Natural" or "Cinema" or "Bright Cinema"? You're going to have to experiment.
That's the point I'm making. You can buy expensive HDR displays that actually display correctly where it's not dark.
They're just expensive, and so most people aren't buying those ones.
So I will repeat, the problem is not with HDR content. It's with displays advertising themselves as HDR when they're simply not.
Spatial audio is like this too.
However, like HDR, it can be difficult to avoid. My laptop screen supports HDR because the laptop is otherwise exactly what I want and it doesn't come in a non-HDR variant. My earbuds support spatial audio because I elected to get AirPods Pro for their other qualities and they don't come in a non-HDR variant.
"Vote with your wallet" isn't really an option with things like this.
Anyway I've listened to music mixed for Apple Music with "Spatial Audio" too and it sounds like crap compared to normal stereo mixes IMO so I guess my opinion stands.
Where do you configure "Spatialize Audio" globally? I can't find anything in the Bluetooth settings (where they've hidden every other AirPods config)
But, the net effect is still in the right direction, maybe.
Somehow the transition from SDR to HDR has been messed up in a major way, for which I can't think of any similar equivalent. E.g. there was never any "step backwards" in going from 480p to 1080p or 4K, or from stereo to 5.1, or from h.264 to h.265.
But the rollout of HDR has been bungled catastrophically. The content is technically fine but the hardware side has been a disaster.
If I were the movie studios, I would seriously think about releasing 4K only in SDR until television manufacturers start showing HDR at full brightness again, and release software updates to fix it in existing models.
Isn't the actual issue that SDR content tends to be mapped so that 0 is the darkest colour the TV can display and 1.0 is the brightest colour the TV can display?
When an HDR signal comes along, then of course the brightest element it can display isn't any brighter than 1.0 was in the SDR image. This has the effect of the scene being overall darker in order to give more headroom for brighter elements to be brighter.
But what is the other option? Have SDR content be dimmer than what the HDR monitor would otherwise be capable of displaying? I doubt you will get as many sales as your competition with that approach because the TV that renders SDR as bright as it can will always look better.
(I do understand that there are some caveats relating to the fact that max brightness is different depending on the amount of the screen covered in brightness, but I don't think it's actually relevant to this argument overall)
For “real” HDR you want at least 800 nits, ideally 1000. Otherwise you have to do what is stated upthread, latch max brightness to whatever your maximum nits is and make everything else very dark to keep the relative contrast intact.
I'm also not too sure about comparing it with some unidentified 'streaming' version. That's like comparing a high resolution digital audio file with a phone call.
Selective sharpening does have a tendency to overemphasize clear edges while leaving the 'unclear' parts untouched, this can give a bit of a plasticy effect especially on skin.
This is the main thing that has kept me from adopting HDR in my media library. I'd expect a feature like HDR would be progressive (if that's the right term that I'm going for), i.e., in a non-supported player, it would look exactly like the non-HDR version of the content, and simply adds more dynamic range when in an environment that supports it. Without that, I'm not going to grab a version of a media file that might look worse if I'm playing it in a non-ideal context.
Does anyone know why this isn't the case? Is it not technically possible for some reason I'm not thinking of, or was it really just a fumble of standards design?
AI does not change this.. It adds stuff that's not there, sure, it adds stuff that might be there, or could have been there, but it literally cannot know, so it just paints in plausibility.. Which is horrible and gross already in principle.
I imagine we're not far from the first conviction based on AI enhanced images.. give the model a blurry CCTV frame and a list of suspects and it will enhance the image until it looks like one of them, and since people are apparently this stupid, someone is going to be locked up for something they didn't do.
Back to movies, just, fuck no! There's no enhancements to make, the movie is _DONE_, it looks like it does, leave it.
> AI does not change this..
It kind of changes it in some cases. The information might be there, but superhuman pattern recognition might be needed to extract it.
And of course, in case factuality doesn't matter, the missing information can be generated and filled in. This obviously doesn't work when you are looking for a terrorist's license plate or when you want to watch an original performance in a movie.
But I think "information" in the context of film here refers to the indexical mark of light upon the image sensor, and in that case no. If it's not recorded, you can't extract it. And whatever you do put there is of little interest to the film buff to whom "image quality" means a more faithful reproduction of the negative that was seen in theaters.
>"zoom -> 9 pixels -> enhance -> clear picture -> zoom ->.." trope in movies
You can have e.g. a picture of a blurry piece of paper, that humans can't read, but I imagine software might be able to read it (with reasonable accuracy and consistency). The information might be recorded, but hidden.
You can’t reverse an operation that loses information with absolute certainty unless you are using other factors to constrain the possible inputs.
Ok, but there are other factors you can use to constrain the possible inputs.
For example with license plates. You know the possible letters and how they appear when "blurred" so you can zoom-enhance them.
And even STILL, you can't be SURE.
Let's imagine some license-plate system optimized to give the biggest hamming distance for visual recognision (for example, no O and o in, only one of them, and no i and 1, only one of them, and so on) to make it as good as possible.
Now, you take some blurry picture of a license-plate and ask the ai to figure out which it is, well, one of the symbols are beyond the threshold of what can be determined, and the ai applies whatever it's learned to conclude (correctly) that the only allowed symbol is X.. Now, thing is, the license-plate was a fake, and the unrecoverable symbol didn't conform to the rules of, it was actually a 1 printed there, but the AI tells it's an 'I' since that's the only allowed symbol.. It just made up stuff that was plausible..
You cannot extract what's not there. You can guess, you can come up with things that _COULD_ be there, but it makes no difference, it's not there.. It's the same with colorized vintage videos, we can argue that it'd not be wrong to assume this jacket was brown since we have lots of data on that model, but we _CAN_NOT_ know if that particular jacket was indeed, brown, it might have been any other color that made the same impression on the monochrome film. The information is _GONE_.
That's why I said "with reasonable accuracy and consistency". Human can't be SURE either. Nothing is ever SURE if we want to stretch it to absurdum.
My entire point is that computers can be better than people at a given visual recognition task. Therefore we might discover that some information is present in the data even though we previously thought that information was not recorded.
That's literally the entire argument. I'm not sure what you are opposing.
When you have moving film, you have 24fps from the same scene, a lens and a compression algorithm.
There is a higher chance for pixel a to be color y if over a span of x frames and compression artifact this pixel shows value z.
You will also have the chance to track details visible from frames/stills/pictures from seconds or even minutes ago if the actor just has the same cloth.
And you can estimate the face of the actor across the movie and create an inherant face.
Nonetheless, besides this type of upscaling, if an AI is trained to upscale based on probability of the real world and the movie is from the real world, it is still more than just random.
Btw. there have been plenty of movie makers making movies the way they did because thats the only way they were able to make it. Low light high end camera equipment is expensive.
And if you watch The Matrix without editing on a oled 60" and higher, it looks off because greenscreen details are more noticable than it has been before. Its aboslutly valid to revisit original material to adjust it to today (at least for me).
It’s better to use a light touch for techniques like this rather just blindly applying it to the entire film.
That's essentially how the brain works, too.
https://twitter.com/RazorwireRyan/status/1735753526167347232
I doubt he even looked, he's too busy with his blue monkeys these days. Most likely someone duped him on taking on the AI upscaling and he signed off on it without looking and the movie studio just shipped the output without QA to save time and money, because they're going to streaming not in cinemas.
The one on the right looks like you ran an edge-directed upscaler on it. Those things have distinct artifacts, and sometimes it looks like all curves turn into snakes. Or it can make new diagonal curves out of random noise.
Not knocking edge-directed upscalers though, they can work in real time and are very good for line-art graphics. You can even inject them into games that have never had that feature before.
Do you or kstrauser make films? Please enlightening me more on the nuances of personal preference?
Since the film does not run in theaters indefinitely and at the time VHS was a common format, among others.
Given film reciprocity and resolution related to period produced film stock and processing techniques, VHS and other reduced resolution formats from the same period, look better, to me! than rescans(4k, which super35 barely supports) and up-resing. Again IMHO because I guess people take this otherwise.
What do you mean 'again'? You never said that in your original comment.
Given film reciprocity and resolution related to period produced film stock and processing techniques, VHS and other reduced resolution formats from the same period, look better, to me!
What you're describing is just nostalgia, nothing more. That's fine, no need to rationalize it.
Why stop at VHS? Why not copy it a few times back and forth between tapes?
4k, which super35 barely supports
I agree, but regular blu ray is 2k
TV content like Star Trek the Next generation were definitely never meant to display in high definition, and such things can show in props etc.
Another famous example is Friends, also shot on 35mm film. It has the advantage over Star The Next Generation is that the props in Friends are actual, real life things, so if we now see the fine details of something, we just sea a teapot a little better. While the prop in Star Trek has a higher risk of looking like that duct taped PVC pipe it is rather than the SciFi ray gun it portrays.
I eventually foisted it off on some deliverymen for free. They didn’t believe me when I told them the weight. When they finally lifted it, it was gratifying to hear one grunt, “Damn. I guess it really is 200 lbs.”
https://www.matteason.co.uk/share/true-lies-colours.png
And the full-size original for comparison:
https://pbs.twimg.com/media/GBaikRhX0AEcjuy?format=jpg&name=...
(I just overlaid the original on the left over the upscaled version on the right and set the blending mode to Colour in Photoshop)
I bet even Kubrick would have been tempted if the situation came about.
Society collectively deciding for creators what is supposedly the right thing to do with or the right way to think about their work is so strange.
Torrent people do this too, and movies is not their dayjob, but their hobby. I guess some people prefer the machine denoised look.
If that takes away from the artistic nature of the film I understand the complaint, but I look forward to seeing this technology applied where the original reel has been damaged. In those cases we are already missing out on what the director intended.
Standard terminology would help us discuss what methods are acceptable for what purposes and what goes too far. And it has to be terminology that the public can understand, so they can make informed decisions.
Yeah - does anyone know if anyone is actually doing this? Like some sort of DLSS for video? I'd love to read about it.
I'm 99% confident that similar issues were raised with e.g. recolored films, HD upscales, etc.
The separation you're talking about is imaginary, the line doesn't exist. Any tool will affect the original meaning if it doesn't match the execution. Remastering is an art regardless of the tool, and it's always an interpretation of the original work. It's fine to like or dislike this interpretation.
But there is a line here. An editor that's using simple tools knows exactly what they're changing, and if they're using simple frame-global tools then they're not introducing anything that wasn't already there.
If you throw an AI at things, it will try to guess what things in the image are, and make detail adjustments based on that.
So that's three categories of edit, easily distinguished: human making frame-global changes, human deliberately changing/adding details, AI changing/adding details in a way that's basically impossible to fully supervise.
It sounds like they accept category 1 in remastering, even though it's not foolproof, and reject 2 and 3.
The problem is that remasters don't make a lot of money, so instead of a properly controlled faithful representation (or a good rethinking) it's typically a half-assed job with a couple filters run over the entire piece. Another issue is that you now have two possibly conflicting intents - one from the original and another from the remaster. ML haven't changed anything in here, it's always been like that.
Non-film buffs, i.e. most viewers, don't care about this.
See new Netflix show Ripley.
All shot in B&W, beautifully shot.
For example Tenet. Cool story, poor audio mix. (I don’t buy the explanation that Nolan had any reason other than expediency for this.) If we use “AI” to fix the audio after the fact, that’s a win in my book.
I’m not a film buff or a purist though. I watch movies with subtitles which is certainly not what the director had in mind, but that’s ok.
... consciously ...
Assuming competent cinematography, it will have an effect on the viewer whether they can analyze it or not.
Good storytelling (and probably blunt spectacle) is the only thing common to all viewers that can win them over. For mainstream media everything else is gloss that may have no effect.
Most people don't even have their sound/brightness/contrast well-adjusted. Some free-to-view services regularly air content with the wrong ratio (and I've seen people happily sit watching the wrong ratio seemingly oblivious to it).
Yes, media nuances can have an effect on the unwitting, but I suspect much doesn't even have opportunity to.
You're going too low level, I'm thinking of lighting and colour and intentional blur via adjusting focus.
> Good storytelling (and probably blunt spectacle) is the only thing common to all viewers that can win them over. For mainstream media everything else is gloss that may have no effect.
You really need to reverse spectacle and storytelling in this statement. How else can the box office be dominated by superhero movies that personally I ... just ... can't ... tell ... apart?
This just provides a new way to watch older movies should you choose to do so. Or not.
This is far from certain, unless "you" are willing to engage in piracy. It's often difficult or impossible to legitimately buy (or even rent) the original, unadulterated versions of older films.
It's a widespread issue with the emulation of old games that have been made for really low resolution/different ratio screens and slow hardware, especially early 3D/2D combinations like Final Fantasy, and those that relied on janky analog video outputs to draw their effects.
Another thing that's plainly obvious in TNG's remastered version is all the black cardboard placed over the display screens in the back of the bridge, to block glare from lights. In SDTV, this wasn't noticeable because the quality was so bad.
'Film buff' responses are common to every major change in technology and society. People highly invested in the old way have an understandably conservative reaction - wait! slow down! what happens to all these old values?! They look for and find flaws, confirming their fears (a confirmation bias) and supporting their argument to slow down.
They are right that some values will be lost; hopefully much more will be gained. The existance of flaws in beta / first generation applications doesn't correlate with future success.
Also, they unknowingly mislead by reasoning with what is also an old sales disinformation technique: List the positive values of Option A, compare them to Option B; B, being a different product, inevitably will differ from A's design and strengths and lose the comparison. The comparision misleads us because it omits B's concept and its strengths that are superior to A's; with a new technology, those strengths aren't even all known - in this case, we can see B's far superior resolution and cleaner image. We also don't know what creative, artistic uses people will come up with - for example, maybe it can be used to blend two very different kinds of films together.
These things happen with political and social issues too. It's just another way of saying the second step in what every innovator experiences: 'first they laugh at you, then they tell you it violates the orthodoxy, then they say they knew it all along'.
Maybe ... just maybe ... Step 2 is where it stops sometimes because it was a bad idea and did make the films worse.
The article is about AI upscaling "True Lies", which was shot on 35mm film. 35mm provides a very high level of detail -- about equivalent in resolution to a 4k digital picture. We're not talking about getting an old VHS tape to look decent on your TV here.
The differences in quality between 35mm film and 4k digital are really more qualitative than quantitative, such as dynamic range and film grain. But things like lighting and dynamic range are just as much directorial choices as script, story, any other aspect of a film. It's a visual medium, after all.
Is the goal to have all old movies have the same, flatly lit streaming "content" look that's so ubiquitous today?
I think the argument against "isn’t there space for people who just want to see the story but experience it with modern levels of image quality" is that such a space is a-historical -- It's a space for someone that doesn't want to engage with the fact that things were different in the (not even very distant) past, and (at the risk of sounding a bit pretentious) it breeds an intellectually lazy and small-minded culture.
Sorry if it sounds crass, but I feel the process of shooting the movie is less important than the story it is trying to tell.
https://en.wikipedia.org/wiki/Comparison_gallery_of_image_sc...
Nonetheless, i do believe that most film makers are actually want to make a film not work around contemprary limitations.
Then I watch Star Wars and I just cannot believe they did this back in the day.
On the other hand, I would probably pay for an AI that will 'undark' that new super artistic movies, because some of them, have worse lighting than the Blair With Project...
We'll see so much of this in the next few years, optimizing everything to the point of boring. Perfect pop songs, perfect movies, perfect novels and on it goes.
Who knows if those whales are even watching most of their own collection.
Where would, e.g., Koyaanisqatsi fit then?
The existence of IMAX should be a hint that there is value in very crisp visuals.
However, this stupid tiktok filter is NOT "very crisp".
This isn’t really a new debate. This is kinda the “it looks tacky to record a feature film at a high frame rate, Peter Jackson”. We have a cultural visual vernacular that guides how we expect certain media to look, and it just feels weird to have a movie from the last century remastered so that it looks like it was shot on the latest digital camera.
Whatever the case, my wife and I went to see The Lost Boys the other day and it was grainy and blurry compared to the super-sharp previews and I just wouldn't have it any other way.
Restoration is one thing. "Improving" it by smearing statistical filter juice over it is quite another.
I think there’s plenty of opportunity to exchange old videos, but I think it requires some human touch or deeper understanding of the movie to maintain its message.
The light and contrast and color is made a certain way on purpose, usually to convert whatever the scene is meant to convey. You can’t just mess with those things just so you add details.
The streaming copy is compressed so terribly that it contains at least 2x less information, and normally much worse. (A great UHD disk can be up to 120mbps, compared to 10-20 on streaming if you get lucky.) So when we’re trying to compare the minutiae in each frame, it makes no sense to compare to a copy that already had a lot of information removed via compression. Also, the 4k streaming copy is nearly always worse than the 1080p Blu-ray.
Dat me! But I shoot and process film on a home made camera, sometimes intentionally degrading the film, or processing without times, or mountain dew instead of water.
From what I’ve read the same team did all these Cameron remasters, and True Lies was the first one so they were still figuring some stuff out and it’s more flawed as a result. Haven’t seen it yet, though. But based on Aliens I’m interested in rewatching The Abyss.
I imagine in ten years you will be moving AI sliders to adjust brightness and contrast of the movie. Want to have everything super sharp? Art house? You can do that.
At the same time, directors may spend more time on the actual story and less on the picture.
I know the main complaint with upscaling older movies is that it can remove most of the film grain and that can give people a waxy, mannequin like look. But they really found a nice balance in Aliens.
I watched the non-UHD a few months ago and just recently saw the 4K version and the difference was pretty stark.
I think HD Blu-ray was good and I didn't have any complaints but the 4K version looks much better to me. I also watched it on a projector with an 120 inch screen which I think benefits a lot from the increased resolution.
Like anything, not all upscaled re-releases are of worthwhile quality.
"They want you to forget what a film looks like" | 310 points by zer00eyz | 3 months ago | 213 comments
Like, the problem with the closeup of Jamie Lee Curtis isn't the skin texture like suggested in the subtitle. It's that she is blue.
not enough to ruin all the newly produced stuff, they also need to ruin all the old stuff now...
https://web.archive.org/web/20240413093536if_/https://www.ny...
If you give shit-tier blurry DVD messes, it turns them into nicer blurry messes.
If you give it the usual bitrate starved streaming quality show, it looks almost as nice as a blueray.
If you give it a blueray... It basically comes out looking exactly the same. I'll usually still add its "darken lines" filter since I'm a fan of strong linework.
So yeah, making a movie sharper could very likely make it worse.
And then some people do masterpiece remasters like that Silent Hill where they removed the fog because modern systems can draw more. Also ruined the atmosphere.
> The recent Cameron restorations were based on new 4K scans of the original negative, none of which needed extensive repair of that kind. [...] The A.I. can artificially refocus an out-of-focus image, as well as make other creative tweaks. “You don’t want to crank the knob all the way because then it’ll look like garbage,” Burdick said. “But if we can make it look a little better, we might as well.”
The only movies which would require upscaling to 4K are those released between about mid-2000s to mid-2010s, the advent of native digital cinema, but filmed in 2K. Everything before was filmed in 35mm film, which can be scanned to 4K with information to spare; everything after is filmed in native digital 4K or more.
Moreover, upscaling which deal only with resolution has absolutely no need of AI. Any TV will decently upscale in _real-time_ a non-4K movie, and more sophisticated techniques can give basically indistinguishable results. 2017's _Alien: Covenant_ was voluntarily produced in 2K but released in 4K through upscaling and the image look just great.
Good to call this out, I think this is something that's really lost on people.
It really blows my mind that George Lucas, for all of his apparent obsessive concern about his films supposedly looking dated, chose to shoot Star Wars Episode 2 in 1080p in contrast to Episode I on 35mm film.
Imagine "Lord of the Rings - the full trilogy in three 8-hour movies, now with new AI-generated content that fills the gaps left from the 'extended' releases. I know people that will definitely renew their HBO subscriptions in order to watch them (if/when they re-release one movie per year)
On second thought, that could be hilarious. It would make all the smokers just look like very thoughtful people constantly bringing their fingers to their lips.
Almost looks like rotoscoping.
> Better to accept the infinite progress unfolding before us than to spend another moment angry or enraged.
Some would say this is infinitely regressive. From the twitter picture, it looks like a justified criticism, from the examples in the article, it's hard to tell whether there's a merited criticism or if it reflexive anti-AI'ism