How many Ks are Kenough?
An essay about megapixel myths, chasing Ks, and resolution.
Let's start with the basics.
What is resolution?
In a modern context, the resolution is the total pixel count of an image.
This is often expressed in different ways. Dimensions, millions of pixels, the number of non-interlaced, progressively displayed vertical pixels, or my least favorite... approximate number of horizontal pixels rounded to the nearest thousand (but sometimes there is a decimal to pad the marketing specs).
640x480 = Classic VGA pixel dimension goodness. 20 megapixels = 20 million total pixels. 4K = Approximately 4,000 horizontal pixels. 8K = A silly idea to sell more expensive TVs. 1080p = 1,080 vertical pixels with no line skipping. (a.k.a. 2K). 1440p = NOT 2K! There are 2,560 pixels, dammit. That's 2.5K!
We all use "resolution" as a shorthand for pixel dimensions. It's a bastardization of what the word originally meant, but that is just how the term evolved linguistically.
Sometimes you just have to accept that language changes and does frustrating things, and incorrect terms become correct when enough people use them incorrectly.
Irregardless, the traditional definition of resolution was just...
The amount of perceivable detail rendered in a given image.
Pixels were not always involved.
Resolution can be digital or analog. Lenses have a resolution. Film has a resolution. Vintage tintype photographs have a resolution. Tamagotchis have a resolution. Tattoos have a resolution. Your eyeballs have a resolution. Your brain has a resolution and it can be different than your eyeballs, which is a fun perceptual rabbit hole.
In this traditional sense, the amount of resolution was determined only by how much detail could be resolved.
Okay, what is detail?
Spatial detail is the amount of real, distinguishable visual information in an image. The fine textures, edges, tonal transitions, and shapes that can be cleanly resolved and perceived by the viewer.
Higher resolution is said to be "sharp". Typically indicated by clearly defined edges.
Lower resolution is said to be "soft". Soft does not mean blurry or out of focus. More like... blurry edges.
Sharp and soft are also dependent on magnification. Eventually, sharp turns into soft. Everything is out of focus if you believe in yourself and zoom hard enough.
In the pixel world, lower resolution may be softer or it may be pixelated.
If you take a digital photo and it looks soft, the lens didn't have enough resolution. If it looks pixelated, the sensor didn't have enough pixels to render what the lens captured.
And then there is sharpening, which is like a perceptual magic trick. The algorithm artificially enhances the contrast of edge detail. You aren't inherently increasing the detail, just tricking the brain's perception of it. Good sharpening is often unnoticeable, but the illusion breaks apart the more you magnify. And if you oversharpen, things start to look a bit chunky.
Sharpening is a perfect example of perceived detail vs measured detail.
For movies and photography, perception is what counts. A scientifically measured increase in resolution doesn’t matter if the viewer can’t see it under real-world viewing conditions. And while measured detail often enhances perceived detail, it’s not guaranteed. Which is why an image with less measured resolution can sometimes look sharper and more pleasing.
When it comes to visual art, perception is the truth. The viewer’s eye and brain decide, not the spec sheet.
If you use the traditional definition, it is entirely possible for a 3K image to have more resolution—more detail—than a 6K image.
Here is a 6K digital scan of 35mm film zoomed by several hundred percent.
And this is a frame from an older ARRI 3K digital cinema camera.
Zoomed out, the viewing experience would be nearly identical. They have equivalent perceptual resolution in that context. But at this extreme magnification, the 3K image has a tiny bit more detail. You can see more individual hairs in the eyebrows and lashes.
Twice the pixels but less resolution.
This is a clear example of how a pixel count does not give you a good sense of how much actual detail an image has.
Think of pixel counts as buckets for detail.
If you have a 2K bucket and a 4K bucket, they have the *potential* to hold that much detail. But not all pixels are created equal. Some pixels are junk. Some pixels contribute no detail to the image.
So if you fill the 4K bucket with 25% detail and 75% junk and the 2K bucket with 100% detail... the 2K bucket will have a higher resolution than the 4K one.
With live action footage, it is impossible to get a pixel perfect image. There are always going to be junk pixels. Noise, grain, blur, etc.
Sometimes junk pixels are a bad thing. If the lens sucks and it can't resolve fine edge detail, you get a bunch of soft, junk pixels.
But sometimes those junk pixels are a good thing. They can add texture and grit to the image.
Film grain nerds know what I'm talking about.
The best looking movies always have some junk in the trunk. And the loss of detail is often a worthy sacrifice to maintain a reasonable amount of said junk.
Real vs Fake: The Quest for Authentic Ks
Image quality often gets reduced to pixels and there are endless debates by the confidently incorrect about what factors result in a better visual. People argue that you can't see a difference between 1080p and 4K at normal viewing distances. Viewing distance is an important element of this discourse. But there are circumstances when you can actually see more detail in an image comparison—even if you can't differentiate individual pixels from your couch.
It's… complicated.
I've also seen many people grouse about how some movies aren't "real 4K.”
It is common for VFX/CGI to be rendered at 2K to save time and money and make the creation process less of a headache for the CG artists. Even powerful computers have trouble processing complicated 4K elements. But this process requires the live action elements to be downscaled to a "2K digital intermediate.” And then everything is scaled up to 4K for the theater and Blu-ray.
The Blu-ray slaps 4K UHD on the case and the marketing says "in beautiful 4K ultra high definition!" ...and this upsets people.
The movie is missing half of the original detail.
Right?
We are being cheated out of extra pixels by greedy movie studios.
Right?
People have made whole-ass websites to let people know if a movie is "authentically" 4K.
(For the love of... 1080p is 2K! That might be the quickest a website has gotten on my nerves.)
This obsession with pixel count doesn't make a lot of sense when you get a deep understanding of resolution and detail.
I am a photographer and I have been studying detail for several years. I'd like to try and explain what I've learned, and why "chasing Ks" does not always mean you will end up with a more detailed image. Megapixels and Ks are one small piece of the puzzle and once you get enough pixel density, other factors matter much more.
In ye olden times, pixel density was a huge deal. When things went from VGA to Super VGA, we couldn't believe our eyes. 640x480 to 800x600. That's 50% more pixels! It was so much clearer! And when 1 megapixel cameras went to 2 megapixels (and didn't require a floppy disk), we collectively threw our film cameras in the closet to collect dust.
Back then, any small jump in pixel dimensions made a huge difference. But once the HD era arrived, we reached a detail threshold. Adding more pixels brought diminishing returns for most viewing circumstances. But marketers latched on to the "bigger number = better" mantra, and the megapixel wars began. It was a simple metric to advertise and an easy way to make your product seem higher quality.
Just cram in more megapixels or Ks and count the money.
We have not been able to break free from these megapixel myths ever since.
Gamers figured out these diminishing returns a while ago. 1440p monitors are still insanely popular because they realized their video game experiences were made more immersive by other factors. Spending thousands of dollars to get a decent 4K experience doesn't make much sense. The small bump in perceptible detail isn't worth it.
So if this pixel density threshold has already been surpassed, what other factors affect our perception of viewing detail?
This is going to blow some minds, but as long as you meet that minimum pixel density required, one of the main contributors to how detailed we perceive an image is...
Lighting!
A well-lit 720p video can look better than 8K smartphone footage. Lighting defines form, contrast, and depth—it creates the shadows and highlights and gradations that make objects feel crisp and dimensional. If the light is flat or bland or lacks dynamics, the image looks mushy, no matter how many pixels you have.
Conversely, directional lighting with good contrast gives the impression of razor-sharp clarity even at lower pixel dimensions. Our eyes read contrast as detail, not pixel density.
Contrast is going from not-a-shadow to a shadow. And what is required to make shadows?
Light!
That’s why a properly lit film scene often looks clearer than a bland, evenly lit smartphone clip shot in “8K.”
Here is an image shot on a 200 MP smartphone.
It is decently resolution-y. You can see everything you need to see.
But here is a scene from E.T. shot in the 80s on film. Probably in the 5 megapixel range.
When that basic threshold of pixel density is met, good lighting with proper highlights, shadows, and contrast will feel "sharper" to our brain.
The next variable in this layered puzzle of detail…
The capture.
Lenses are one of the most important aspects of capturing lots of detail. A good lens on a 20 megapixel camera will look sharper than a tiny plastic lens on a 200 megapixel smartphone. And when you combine a good lens with good lighting, the perceived detail skyrockets even when viewing on a tiny phone screen.
In this photo of Otis, you can see the texture on his nose and every blade of grass. It started as an 18 MP image and has been reduced to ~2K. But the lens used to capture this image probably weighed more than a thousand smartphone lenses.
Good glass is key to capturing detail, as long as the bucket (the sensor or the film) is big enough to hold the detail the lens resolves. A bigger sensor and more megapixels increase the potential resolution, but the lens needs to have the optical quality to achieve that potential.
Bigger sensors require bigger lenses. Bigger lenses are easier to engineer, have more area to capture detail, and allow better light transmission. The less light transmission data is corrupted as it goes through the lens, the more detail it will retain.
Lenses have a variable analog resolution. They are sharper or softer at various apertures. If you shoot wide open, they might be softer. And if you shoot with a tiny aperture, you will get reduced resolution due to diffraction. If you want to capture the most detail possible, you need to find the "sweet spot" aperture for a given lens.
Even if your sensor is 8K or 200 MP, a bad lens may not be capturing that much information. And if the sensor is smaller, that may bottleneck the detail even more. There is less area to capture visual data. And if the image is noisier—another reduction in detail.
The variables involved seem endless.
Then there is film.
People love to give film magical properties. Some think film has infinite resolution because movies have been scanned at up to 12K. They believe this means film has more resolution than digital cameras.
I love film. Some of my best friends are film. And film often holds up well to modern display technology. But in terms of the amount of detail captured, digital will win most of the time.
Film has even more variable resolution than lenses. Film is sharper if the scene is brighter. It is mushy when there is not enough light. If the film speed is high, the larger grain will knock down the resolution. If the speed is low, it can capture a surprising amount of detail. Older movies with professional lighting and good lenses (for that era) are able to hold up pretty well to our high definition future. But when people say film is "more 4K' than modern movies with 2K upscaling, they are contributing to the confident incorrectness of this resolution discourse.
If the film is large like IMAX, it can resolve oodles of detail. Almost comparable to current high end digital cinema cameras. But only if the film speed is low, the lighting is bright, and the lens is sharp.
Be skeptical of IMAX claims of "18K" resolution. The IMAX nerds and Nolan acolytes on Reddit get very caught up in this claim.
Perhaps with 50 ISO film in a lab environment with bright lights and a scene with no motion. But, even then, most IMAX lenses are converted medium format designs from decades ago.
In real world settings, the spectrum of resolution from scene to scene can vary wildly. It can even vary across a single frame. Digital capture is always going to be more consistently sharp.
To bust another common myth... scanning film at 8K or 12K does not make the movie that resolution. And it is certainly not more detailed than modern digital cinema cameras. This high resolution scanning is mostly done to preserve the film grain structure. The actual, physical grain crystals do have an "infinite" atomic-scale resolution, but that doesn't mean the resulting image does. I'd be surprised if any film from the 80s even reached 4K levels of detail.
E.T. in 4K is never going to be sharper than any Marvel movie—even if upscaled from 2K for the effects.
This is 15 perf IMAX film scanned at 11K.
This is a large format 6K ARRI digital cinema camera (often used for Marvel films).
I would give the edge to the ARRI when magnified, but I would say under these circumstances these camera systems produce about the same level of detail. And since they are not pixel perfect images, their effective resolution is less than 6K.
Meaning the 18K IMAX marketing doesn't hold up well beyond the theoretical. And if 70mm film can't out-resolve the ARRI, 35mm scans aren't going to either. Especially if the film is 40 years old. This notion that scanned film has superior detail is inaccurate.
Under different circumstances, the IMAX might produce a bit more detail. Or it might produce less detail, but have more pleasing grain than the ARRI's digital noise.
A more pleasing image does not always mean it has more detail. It is entirely possible for IMAX film to look better and have less detail. It's a beautiful format.
My point is, both systems produce plenty of detail and making them compete in a pixel pissing contest doesn't really make sense. You can project the footage from either camera system on a 100-foot screen and get a lovely result.
What is... 4K?
I know that sounds like a dumb question, but I'm not sure everyone knows what they really mean when they say something is "4K." And we have certainly lost the plot when it comes to megapixels.
As I've demonstrated, the number of pixels gives you no information about the actual detail. It just tells you how many pixels there are. And remember, no live action footage has pixel perfect resolution. And every frame is filled with junk pixels that contribute nothing to the overall detail.
This is why lens and camera manufacturers use "line widths per picture height" (lw/ph) or "line pairs per millimeter" (lp/mm) to judge the actual resolution of a camera or lens. They shoot a test chart with parallel lines and determine how small and how close the lines can get before they just blur into gray.
This is a much more reliable resolution metric, but it is much more complicated and less marketable than Ks or megapixels. You can't assign one number to a lens or camera sensor. The resolution needs to be measured as an entire camera system and the combinations of lenses and sensors are endless. And each camera system can output multiple resolutions depending on settings and circumstances.
One of these cameras has a 200 MP (16.3K) sensor and the other has a 150 MP (14.2k) sensor.
One of them costs $1000 and the other costs over $30,000.
One of them has a sensor the size of a Froot Loop, the other has a sensor the size of a Post-it note.
One has a mass-produced, dime-sized plastic lens and the other has a handmade, fastidiously engineered lens with several pounds of optically perfect glass.
The Samsung smartphone has 200 MP.
The Phase One medium format camera has 150 MP.
The Phase One can capture up to ~2.5x the amount of detail, despite having 50 million fewer pixels. It can also capture much more dynamic range and color information, which can further enhance perception of detail.
Which means megapixels and Ks have become a near meaningless metric. You can't trust the advertised pixel count to give you a sense of the image quality or detail.
If smartphones disclosed their lw/ph numbers, their megapixel marketing schemes would crumble. Because most smartphones capture between 8 and 15 MP of actual detail despite marketing claims of 50 MP or 200 MP.
The actual measured detail captured is sometimes referred to as "effective megapixels" or "perceptual megapixels."
Smartphones reached their effective resolution limits years ago. They can get a small bump in detail if they make a larger sensor, but you are going from the size of a pinky fingernail to the size of a slightly larger pinky fingernail. They are limited by physics. There is only so much information that tiny plastic lens and that pinky nail sensor can resolve. So computational processing is their only option left to give them a significant edge in overall perceived sharpness.
So what is 4K? What is 200 megapixels?
Those numbers are just the size of the buckets. Nothing else.
I think the best way to think about it is... if you throw out the junk pixels, how many pixels are left that contribute to the detail of the image?
That is your true effective resolution.
For something to be effective 4K, it would need 0 junk pixels. Only a computer generation can achieve that.
So almost nothing is effectively 4K. Every time we say something is 4K, we really mean it is 4K-ish.
Our TVs are 4K buckets, but our content rarely fills them up to the top.
Okay, let's talk about upscaling and being cheated out of pixels.
2K CGI is not the devil people think it is.
This is one instance where the greedy studios aren't actually cheating you out of much. Perhaps their marketing isn't very transparent. And sometimes they slap a 4K label on something that isn't much of an improvement. But I'm afraid that "real or fake 4K" website was a huge waste of time.
If parts of the workflow are done at 2K to add visual effects, that doesn’t completely destroy the captured detail.
Capturing detail is more important than the final output pixel dimensions.
Remember the buckets?
If you shoot a film in 6K with a great lens, you can fill a 2K bucket and discard almost all of the junk pixels. You aren't losing tons of detail, you are concentrating it into a smaller space.
And since you discarded all of the junk and concentrated the detail, the 2K image is much more malleable to resizing with a marginal loss of fidelity.
Detail is a bit like caffeine. As long as the amount of caffeine remains about the same, you can change the container and still get a similar stimulating effect.
A can of Jolt Cola, a Big Gulp of Mountain Dew, or a tiny cup of double espresso will wake your ass up about the same.
I have prepared a demonstration.
I captured this image of an old box of light bulbs at 6K and 2K. Text is a torture test for perceived detail. The sharp transition from black to white is very easy for our brains to see.
It's hard to see much difference at this scale, so let's do some ridiculous magnification.
Here is the 6K image downscaled to 4K.
Here is the 2K image upscaled to 4K.
Here is the 6K image downscaled to 2K and then upscaled to 4K.
Let's magnify them even more in the same order.
This is using basic Photoshop scaling algorithms. I didn't do any extra sharpening or contrast enhancements.
To my eyes, the 6K>2K>4K image holds up pretty well. It's not quite as detailed as the 6K>4K image and it has some sharpening artifacts when extremely magnified, but at normal magnification, it is tough to see the difference. I would say the loss of detail is below 10%. At normal viewing distances, maybe even 0%.
Social media doesn't allow me to show you full-resolution samples, but the following crop represents a 4K section of the image on your screen. You can download these and view them at 100% zoom and this should give you an idea of the difference in normal viewing circumstances.
6K to 4K...
6K to 2K to 4K...
Zoomed out, the sharpening created by the upscaling algorithm might even give a bit of a boost in perceptual detail.
The lesson here is... the *quality* of the pixels matters more than the quantity. As long as you capture the detail originally, it will be maintained during these scaling processes with minor losses.
So if a Marvel movie is captured on a large format 6K ARRI camera with a giant modern lens, downscaled to 2K for VFX, and then upscaled to a 4K master, it's still going to maintain a lot of that original detail. Calling it "fake 4K" when only a small percentage of actual detail is lost is pure hyperbole. It fuels this chasing of Ks and bolsters the megapixel marketing scam.
You were not cheated out of extra detail.
Okay, but what about CGI?
Many visual effects are created at 2K and then upscaled. Surely we are getting cheated out of extra detail there.
Right?
Computer-generated elements are “pixel-perfect.” A 2K render is exact. It’s pure geometry and math, with no sensor noise or lens softness. That’s why older CGI films like Toy Story (rendered at under 1080p) still look surprisingly crisp when upscaled to 4K. Every pixel was intentional and none of them were junk.
When visual effects are integrated into live-action footage, they’re deliberately degraded to match the imperfections of a real camera. Noise, depth of field, motion blur, lens halation, and film grain are all added. If the VFX were razor-sharp, pixel-perfect 4K, they’d look fake as heck. Matching realism means reducing detail, not increasing it. You have to manually put the junk back in the trunk. And 2K VFX often have plenty of fidelity to mate with higher resolution footage.
If the CGI looks bad, that is an art direction or budgeting folly and not a lack of pixels.
So why do some 4K movie releases look more detailed and others look disappointing and soft?
Mastering, time, and money.
Proper mastering is where most of the “4K magic” actually happens. When you see a 4K Blu-ray that looks dramatically better than the old 1080p disc, that improvement rarely comes just from extra pixels. It’s because the studio created a new master. If they do the job right, they will go back to the original digital footage and reprocess it using more modern workflows. Or if it is film, they will go back to the original negatives, clean them, fix any damage, and use modern scanning processes to get the most fidelity possible to work with. They redo the color grade (hopefully matching the director's artistic intent), improve contrast, sharpen, and then encode it at a far higher bitrate.
I'm not saying having extra pixels is pointless. The higher pixel count does give them more latitude to make improvements. But the quality of the source material is always paramount. And the pixel count is only one variable of the overall quality.
If a 4K version looks disappointing, it is almost always because they cut corners in the remastering process. It rarely has anything to do with the simple act of upscaling.
Enter Artificial Intelligence.
AI upscaling is a process where trained AI models generate new detail pixels when enlarging an image. Instead of mathematical algorithms and sharpening, they literally invent new detail to increase the resolution of the image.
This AI upscaling has been a bit controversial and is already creating some horror show results.
I know it is easy to demonize all AI. And I agree that AI is often misused, corrupted, and hurtful to artists and the environment. But I also see promising potential from various uses of AI. There are scientific applications that could lead to improved medical treatments. It can help sort through vast amounts of physics data. And there are artistic tools that can reduce tedious, non-artistic tasks like advanced selections and rotoscoping. I personally think AI upscaling is a wonderful innovation. When used properly, it can help artists rather than trying to replace them.
I use it often for my photo restoration work, and I have had some stunning results fixing up old film photography.
I zapped all the specks and dust. I fixed the contrast and did the final color grade. But this was scanned at a low resolution and the photo was decades old. AI upscaling was able to create the detail needed so this image could be printed in larger formats or viewed on larger screens.
All forms of upscaling, including traditional algorithm-based methods, need to be prepared for success.
You cannot just hit the button and expect good results.
You need to guide it with artistry and skill. You need to pick the strength of the upscaling, the amount of sharpening, and you often need to do trial and error to arrive at the best result. You still need to adjust the exposure, saturation, and contrast. You need to repair any damage to the photo. The source material needs to be as high quality as possible to make sure the AI doesn't create a horror show when upscaling. And when it does go sideways, you need to fix it with traditional processing techniques. Sometimes you even need to selectively blend the upscaled layer with the original layer.
In my experience, AI upscaling never outputs a finished product. You still have to do a final polish to get an authentic result. If it removes all texture and grain from the photo, you need to add that back.
You can't just hit an "AI make picture look good" button and be done with it.
Push-button AI solutions create slop.
But AI tools made specifically for artists can do tedious, non-artistic tasks, or allow for results previously impossible with traditional tools.
AI upscaling does not save me time. It actually requires *more* time to achieve a good result. But it allows me to restore precious memories in a way that was not achievable before.
So when studios view AI upscaling as a push-button cost-saving measure and not a tool, that is when things go tits up.
Netflix and now YouTube want to automate this upscaling process. They do not care about the end result as long as they can market claims of higher fidelity.
What's even more frustrating is that these upscaling tools have subtlety built into them. You don't need extreme settings to get a good result. But people might not always notice the improvement if the upscaling is more tastefully done. And subtle is harder to sell.
You can absolutely get a beautiful 4K version from a high quality 2K source if you spend the time and money to master it properly.
I don't know if they used any AI upscaling tools, but even with traditional algorithms, you can see they put in the work to elevate the original 1080p digital footage. In this magnified image, you can see his face and beard have good texture and the sharpening artifacts are barely noticeable. You can't expect miracles from the very first 1080p digital cinema camera, and the lack of film grain is a little disappointing, but I think this was a well done effort.
But what happens when studios cut corners on the mastering process?
What happens when they skip the time and effort components and just hit the "AI make Arnie look smoooth" button?
They took the time to create a high quality 4K scan of the original film. The resolution of the source material was probably much higher than Attack of the Clones. They had beautiful film grain texture and a nice color grade to reference. And they hit the AI denoise button with the strength turned up to 11.
James killed the grain and the color. I don't know if he was trying to rush the process or if film grain killed a family member and this is his revenge, but this is a disappointing result.
AI did not fuck this up.
Human choices did.
I cite these examples to show that you can have more pixels and achieve a worse result. You can have a good upscaled result and a bad native 4K result. The quality of a remastered image, the detail you see in the final result, is affected much more by an unwillingness to spend time, money, and effort than by simply doubling the pixel count.
If we keep buying into this "more pixels = more better" mindset, then that's all studios will do. They'll double the pixels and take your money. Or they will say "scanned at native 4K' and turn Arnie into a rubber person.
You want the effort, not the pixels.
Viewing distance.
There is another ongoing raging argument in the home theater community about whether you can tell the difference between 1080p and 4K at typical viewing distances and screen sizes. The science says we can't distinguish pixels that small, but eagle-eyed videophiles *swear* they can see extra detail.
If the version of the movie is identical and the only variable changed is the pixel count, at normal viewing distances, your eyes cannot detect a change in spatial detail without vastly enlarging the screen or sitting absurdly close.
And there’s another factor some projector owners may not realize… even “4K” projectors rarely deliver effective 4K. High resolution lenses often cost thousands of dollars and are out of reach for most home theater owners. The image may measure 3840×2160 pixels, but the optical resolving power often lands somewhere between 2K and 3K.
So their systems may not even be capable of rendering the extra detail they claim to see.
But people *do* sometimes notice a difference in detail between a 1080p and a 4K version of a movie. It is not just a placebo effect. And this is probably the cause of these endless debates.
So... why would a 4K remaster look better than an older 1080p version if you can't actually distinguish more pixels at a given viewing distance?
Because someone cared.
What's even cooler is that a rising tide lifts all boats. A remaster done with care benefits all who view it, regardless of viewing distance or the resolution of their projector or display.
A high-quality 1080p encode of a well-done 4K remaster will still look better than a 1080p Blu-ray from 2009. The 4K format gives filmmakers more bandwidth and incentive to improve the overall picture quality. They have... more room for activities.
You are getting fewer junk pixels in your resolution bucket.
Even with smaller screens and farther viewing distances, your eyes see improvements from more colors, custom LUTs, better color grading, higher contrast, more dynamic range, brighter HDR, cleaner gradients, better texturing (film grain), selective sharpening, and less compression—all of which have far more impact on perceived image quality than pixel count alone.
So the discourse should not be "1080p vs 4K."
It should be "cheap vs properly funded remastering."
Artistic intent.
In the home theater community, we nerds tend to get bogged down in the technical. We are chasing perfection. We want the highest quality experience possible. We want to immerse ourselves in our favorite movies. We fill our rooms with expensive subwoofers and A/V equipment and argue about specifications until we are blue in the face.
But there is something that often gets lost in our technical discussions about bit rates, nit rates, and pixel counts.
I'm going to let you in on an industry secret.
Many of your favorite directors do not like 4K.
They love the ability to capture as much information as possible. But there is a huge difference between what is captured and the final output.
I mean, I want a 60 megapixel camera. I want to be able to crop. I want to be able to edit with precision. I want smooth gradients. Capturing a 60 megapixel image with a sharp lens is great. I like having all of that data. I like having a bigger bucket.
But all my photos are published at 2K. And I spend a lot of my editing time softening skin pores and vellus hairs because not all of that extra detail is flattering or realistic to human perception.
The human iris is beautiful and I love seeing every bit of detail my camera can deliver.
But the human nostril is a fuzzy cave full of boogies.
I can't get the iris detail without the fuzz and boogies, but I can reduce the boogies with a click of the mouse.
When working with video, it is much harder to be selective about detail. And a lot of directors do not always want the end result to be an ultra sharp, highly detailed image. So they use old lenses, diffusion filters, film grain and all sorts of other tricks to create texture and atmosphere. They might fill a room with haze or shoot a with a telephoto lens on a humid day. Air itself has variable transparency and can obscure detail.
THESE ARE ALL RESOLUTION REDUCING PRACTICES.
People often refer to Dune and how visually stunning it is. Some will partially attribute that to a pure 4K pipeline.
None of that 2K VFX upscaling bullshit.
No cheating us out of our extra pixels.
PURE 4K! YEAH! *rad guitar solo*
But, in reality, they took the super detailed digital footage, exposed it onto film, and then converted it back to digital. This gave it a film aesthetic, but it also reduced the sharpness of the digital image. It's very unlikely the Dune movies have anything near a pixel perfect 4K image.
And the movie looks *better* because of it.
Modern directors have also been using rehoused vintage lenses and old school lens designs to avoid clinical sharpness. Some have been favoring Russian lenses from the 1950s. They restore them and put them in a new casing and give them modern features.
You may have seen these vintage optical characteristics in shows like Shogun. Many scenes have pronounced barrel distortion and heavy vignetting.
Movies like The Northman are using Petzval lens designs from the forties.
The EIGHTEEN forties.
This gives them a trippy, swirly bokeh effect.
So while home theater nerds are chasing Ks, cinematographers spend more time reducing resolution than increasing it because ultra-crisp digital images tend to look fake, sterile, or “video-ish.”
Which begs the question…
How much does detail actually matter?
Why is the number of Ks or megapixels the most talked about aspect of image quality?
Having a big enough bucket to capture detail is important. I want to be clear on that. But it's one small factor in what makes a good image.
Do you all find yourself looking at a license plate in the background and complaining you can't read the numbers?
When you are engrossed in a movie, how much does a lack of detail affect your experience?
How often are you actually disappointed about the perceived sharpness?
Do you notice it casually, or do you have to resort to magnified A/B comparisons to actually appreciate the difference?
In my opinion, we have achieved plenty of pixel density. The detail wars are over and we all won. It's time to move on to things that matter more when it comes to picture quality.
We need to stop obsessing over "real or fake 4K" and asking for more competently crafted remasters.
We need to demand higher bitrate streaming so the actual detail intended by the filmmakers isn't muddied by compression artifacts. (I would totally choose a high bitrate 1080p stream over a super compressed 4K stream.)
We need to demand fewer "fix it in post" shortcuts. Build real sets. Go to real locations. Integrate CGI with practical elements.
We need to demand that Hollywood stop using bland lighting and desaturated color palettes to make the VFX cheaper and easier. More contrasty lighting. More saturated colors. Brighter HDR masters.
We need to give CG artists the time and resources needed to craft a beautiful result. Despite the Davy Jones memes, CGI has not regressed. Dune, Top Gun: Maverick, and Avatar prove it is better than ever. But art directors and CG artists are not given the time and resources they need to thrive.
We need to demand no lazy, push-button AI solutions for our favorite classic films.
We need to demand better artistic freedom like we see in movies like Dune.
Dune looks better because of artistry, not pixels.
Let's demand higher standards and abandon this chasing of Ks.
-----------------------
If you enjoyed this essay and want to delve further into the science of resolution, I highly recommend Steve Yedlin's resolution demo. Steve is a cinematographer/image scientist (Knives Out, The Last Jedi). He gives some wonderful visual demonstrations of how we perceive detail.
