Hacker News Reader

Comment by npunt | Hacker News Reader

TV resolution goes at the pace of available content, and everything goes at the pace of context (living room 8-12' away), which is presently stuck at roughly what the human eye can resolve at that distance, which is about 4k.

There isn't a world where a 5k or 6k tv ever makes sense. Monitor, yes. But not for TVs.

View on HN

> which is about 4k

4k does not exist because it's optimal for the human eye. It's dominant right now because anything higher carries a really significant bandwidth penalty, which would be hard to do for both streaming and optical media.

If we lived in a world where blu-ray disks could hold 1TB, and global internet speeds were 10x what they are now, 8k TVs would be the norm.

People can't even tell the difference between 4k and 8k. You need a microscope.

People have been saying "You can't tell the difference between X and Y" since the dawn of HD graphics in general.

"You can't tell the difference between HD and FHD." FHD and QDH. QHD and 4K. 4k and 8K. 60FPS and 120. 120 and 240. A higher resolution texture. Higher resolution audio. A higher resolution mesh. Etc.

Every single time someone makes the claim it makes no difference, yet we do it regardless.

For something that seemingly makes no difference, people sure seem to continue improving it regardless. And shockingly, consumers seem to really enjoy the products too.

Yeah, but you actually can't tell the difference

https://www.techhive.com/article/578376/8k-vs-4k-tvs-most-co...

Maybe. That test doesn't cover everything. Not all content, not all TV sizes, not all individuals, not all display technology, not all environments, not all compression tech, etc.

What about images that use HDR10+? Or Dolby Vision? What about completely un-compressed lossless video? What about common compressions like X264 and X265? What about on an 80-inch screen? What about in dark vs bright scenes? Or even environments? (Human eyes handle images differently at low vs high brightness)

I'm not saying the data is worthless because it's not, but I am saying that I think it's too sweeping to say it's completely un-noticable. Not enough data for that conclusion. Vision and display technology are both very complicated topics and there's always room for improvement IMO.

Pixel density has diminishing returns, we've known forever its not linear. Its the same reason why phones aren't all 4k by now, nobody can tell the difference.

Every single upgrade in resolution has been diminishing returns, and that hasn't stopped anyone from doing it anyways.

The only reason it will ever stop is if we hit major technological or even physical barriers. And even then, we'll most likely just move onto another technology that allows yet higher resolution.

Now you're gonna tell me I don't need more than 44,100 Hz for sound.

I don't know if that is meant to be tounge in cheek, but for most ears it is physically impossible to discern any broader audio spectrum.

Actually it is, but I'm stoned to death even when I share audible sound deltas between lossy and lossless codecs, so I'll just leave it with one sentence:

    Higher sampling rates and higher quality encodings result in a broader, more impressive and immersive sound stage primarily, and more details are secondary, and in some cases are elusive to run after.

"more impressive and immersive sound stage" is a nice way to say "placebo".

Lossy codecs, at too low bitrates, can be a problem, yes.

But every content can be recorded and played back as 48kHz 24-bit FLAC, which is perfect. No amount of money you spend can improve upon that.

Here we go.

So I have a setup: Akai AM2850 Amplifier, a pair of HECO Celan GT 302 speakers, a Yamaha CD-S300 CD player, which has USB input with iPod interface and MP3 capability.

A well mastered album (read: No brickwalling) like Wasting Light (Foo Fighters) or Brother in Arms (Dire Straits) will provide you a larger soundstage when you listen in front of this set from CD (or any lossless source, but let's keep everything in the same DAC), even compared with 320kbps MP3 (it's minimal, but audible).

This can be ruled as placebo alright, but audio science is hard science. It's subjective, and there's an aura of lies around it due to great deal of real snake oil in this.

Yet, I listen to same amplifier for 30 years, and I know how it behaves in any genre, any sound source and any input level, yet I can make sense of the subtle sound stage changes, because I'm so familiar with it.

Honestly, I played in large orchestras and whatnot, but when listened intently, the difference is there. It's difference between a good cup of Tchibo vs good cup of Davidoff coffee, but it's there.

I'm currently not at my home, but if you want I can provide you a fresh set of sound deltas (FLAC vs. 320kbps MP3) when I return.

I agree that higher encoding quality can provide the differences you describe, but higher sampling rates (beyond 48kHz, that is) cannot.

Personally, I try to acquire all music on CD or as FLAC from e.g. Qobuz or Bandcamp. But I see no reason to go beyond 24-bit 48kHz for any kind of audio unless I plan on slowing it down for e.g., a slow-mo video.

I have a 24bit WAV version of Radiohead's "OK Computer Oknotok" directly from the band, and sounds absolutely beautiful, though I don't remember whether it's 48K or 96K.

In any case, I really don't believe that I have a system can render 96K sound meaningfully different than 48K. I'm also not sure that I can tell the difference unless I try very intently, either.

I also an avid Bandcamp shopper. Some of the people make really beautiful music there.

Lastly, if I'm happy with the sound I get, I don't run after numbers. I use a Behringer Bass V-Amp as my bass processor, and even though it's far inferior on the paper, the result is not for my ears.

Heck, if Haggard can use this thing for live and studio, why can't I?

A sound stage is not quantifiable, subjective and probably not real. I've heard similar things said many times for a number of audio systems and encodings and in a blind test it is just not discernible IME.

It's discernible and measurable. It's called stereo separation and instrument placeability in mastering parlance. iZotope Ozone has a nice stereo imaging view to show the stereo separation in tracks.

When you mic a big orchestra (or any orchestra) clearly, you can capture the stereo image (which is generally done with non-instrument ambient microphones), too.

Mixing this ambience into the final stereo mix can increase the instrument placing when done correctly, but needs a good room with good acoustics. This is why we have Atmos and other tech. It allows you to virtually position channels in a room, and increases this separation, in theory. Nice when done subtly, ugly when done overboard.

Also, speakers' sound rendering has something with it. For example, the Celans I have can fill a room with positionally correct instruments fairly impressive. What's more impressive is, Creative Gigaworks speakers also capable of doing this while being desk speakers (yet they have kevlar cones and silk tweeters too).

There's a sensible upper limit to pay for audio systems. Mine is neither top of the line, nor too cheap to be true. It's a powerful, yet vintage all analog powerhouse, and I have it for 30 years or so.

After some point, even if the resolution provided by the system is justified by the money it requires, finding sources to saturate is impossible to find. So, when you find something you like for the first time, you stop there. I plan no changes in my audio system for example.

There are infinite number of variables in audio (system, source, room, placement, etc.), but sound resolution and soundstage are not placebo effects. If it was, a $4 headphone would sound the same wth $44 or even with $440 one, yet they don't.

Same for speakers, amps, etc. Until a certain point as I said.

When you hit a drum there are all sort of short duration high pitched components that make that chest thumping all the more impressive.

You don’t need more than 44,100 kHz thanks to the Nyquist-Shannon sampling theorem.

https://en.m.wikipedia.org/wiki/Nyquist–Shannon_sampling_the...

If you got an ideal filter maybe. Having margin to half the sampling frequency helps filter design.

Then again, the frequency response of the loud speaker is not flat so I dunno if the sampling filter cutoff frequency matters.

The real rabbit hole is when you start looking into the frequency response of your eardrums

Sure but I guess the goal is for the replayed sound to sound the same as the original would.

According to Wiki the absolute max an amazing(?) ear can hear is 28kHz. You can maybe hear aliasing of higher frequencies too?

I can't hear any difference between 44.1kHz and faster sampling, but we can't refute it on purely theoretical ground.

That's not what I'm getting at. For most purposes the 22KHz limit is a hard one. What I mean is that the eardrums do not have a linear response at any frequency.

Yeah but you can keep source material at 44.1k and oversample before the DAC.

And 24 bit is nice purely because you can regulate volume in software without losing fidelity (if DAC is good enough). But again, no such requirement on source, just the processing pipeline

Unless you're editing. 192kHz recordings slowed down to the audible spectrum are gorgeous.

Tbh, if you're editing, remixing, etc, 32-bit float at higher sample rates obviously make sense. But not for a final delivery, and not for your playback devices. Even when I edit 192kHz 32-bit float audio (e.g. for a slow-motion video sequence where I need to slow the audio to match) my playback setup still only supports 16-bit 48kHz (and I don't need more).

48khz is better for video due to the way it can be divided more evenly with popular video frame rates - and besides it’s the standard for audio alongside video so you might as well use it for everything.

So yes, for that reason alone you need more.

And if you'll edit, remix, and slow the audio down, you might want 96kHz or 192kHz, sure.

But for the final delivery, there's no need to ever go beyond 48kHz at 24-bit.

Why do you keep repeating 24 bits? It is way too much, even at 16 bits the dynamic range is far more than what you can reasonably hear (unless you want to damage your ears!) - especially if you are listening in a real world environment, and not in a completely silent room... Vinyl has a significantly worse dynamic range, yet it is loved by hi-end fans, sometimes even preferred to CD!

24 bits is useful for mixing various sources, but otherwise it is a waste of space.

You're absolutely right. Personally, my two most used playback setups cap out at 44.1kHz and 16-bit and that's good enough.

> 48khz is better for video due to the way it can be divided more evenly with popular video frame rates

Is the half sample per frame at 44,100 Hz and 24 fps really an issue? DVDs support 48kHz, but they only support 25 fps and 29.97 fps. You're going to run into so much NTSC content that I'm not sure even divisibility is a thing.

My 20+ years of being involved in broadcast/tv/video technology says otherwise about TV resolution goes along with content. It’s actually a weird chicken/egg situation, that involves more than screens or content such as delivery, price of content creation and customer acceptance.

Doesn’t the tv size matter? My little 42” 4k at 12’ can’t be the same amount of fidelity that a monster 83” at 12’ is doing.

A larger TV is meant to be viewed from a further distance though. At the suggested distance, your eyes shouldn't be able to resolve the individual pixels.

12 feet? The kids are working on strapping a pair of them to your face and calling it VR!

You're right, the angular size is the true determining factor. But most people don't have a large enough living room to sit 12' from their TV :)

e.g. I've got a 48" UHD TV at 2.5m (8') distance, and a 27" UHD monitor that I use at 60cm distance, and while the TV is beyond the resolution limit of my eyes, the PC monitor isn't.

Even with really good film scanners, 4K seems to be about the limit of what’s worthwhile. Even at 4K itself compared to 1080, film grain I’d much more evident (not a bad thing persay, but a thing).

You've missed screen size as a factor. TVs are getting larger by about an inch per year since flat screens took over, and that trend doesn't seem like it's stopping any time soon.

There's also the type of content and your personal visual acuity. If you're young, your eyesight is probably quite capable of resolving much finer details than normal 6/6 vision. I can tell the difference between 4K and 8K at a normal sitting distance on a 65" TV.

However... the difference isn't enough for me to care. I'm much more interested in brightness, colour, framerate and a load of other things over another resolution bump.

Content over 4k is readily available and even cheap to produce, but most of the good stuff is still being produced in 4k, so why is that???

Because 4k (and even 1080p) is good enough for most people?

Yeah, for TVs that sounds right, but not for desktop displays. I wonder if desktop display pixel density is only stagnated because the TV industry has so much influence on the momentum and the TV industry thinks 4k is good enough.

Bandwidth. ISPs already throttle streaming and have data caps.

I think that influence is pretty clear. In my neck of the woods, when pc monitors in 16:9 became a thing, the marketing copy was basically how great they were for watching movies.

Only now do some manufacturers start to put out taller monitors, but they're still rare, at least in my market.

That's why I'm happy to see new 8K TV release. I think it's overkill for 99% TV usage, but it's needed for PC display.

Also, if you compare same laptop available with a 1080p screen or a 4k screen, the former gets much better battery time.

Is it even needed on desktop displays? My experience has been that anything under 32” you need to use scaling to make anything readable on a 4K display. I’m current using a 43” 4K display as my main monitor, sitting not very far from to, I can’t distinguish individual pixels as it is, and frankly even this is uncomfortably large, I have gutters set up around the edges so that windows maximise to a comfortably viewable size.

Cost of delivery. Most stuff is being streamed these days, and nobody wants to have to pay for the cost of streaming 100gb files. Plus the cost of commercial videography equipment is insane. Why upgrade before you have to?

The better the capture, the longer the shelf life of that footage.

So, while the pipeline might not run at more than 4K, the source material ideally is at a higher resolution.

And you need to make sure it still looks good on lower quality displays.

Nobody is making movies that anyone would really care to preserve for the coming decades anymore anyways.

Besides which, we took a big step back when we switched from 35mm to digital. Digital has only recently reached par quality.

The BBC probably thought that when they recorded over old Dr. Who episodes. It’s up to our grandchildren to decide what they want to see and what they don’t.

As someone who used to run a photography company: total costs of producing 4k content (camera system, computer setup to handle file management and editing, time required to process and manage content, storage, headaches, etc) is absolutely enormous compared to 1080p. Higher resolution content like 8k seems like a nightmare for everyone but the consumer. I do think and hope we’ll get to the point where it’s not a big deal, but even today a brand new macbook pro will struggle to render a basic 1080p composite in After Effects. Even basic 1080p footage editing in Premiere Pro can overwhelm reasonably modern machines.

(Yes, I know that you do X on your Y machine without problem, but my point is that it’s easy to forget to account for all these little costs, and they pile up! Especially for professional workloads where people love to push the bar higher and higher.)

I know that you do X on your Y machine without problem, but my point is that it’s easy to forget to account for all these little costs, and they pile up!

Cloud-based After Effects render farms do help significantly though.

You still doesn't seem to get the point.

I think I do. The point is that you can build your video in Affect Effects locally at 1080p (or less) on a standard Macbook, and then render the video in the cloud at 8k really easily and at quite a low cost. You don't need to "do X on your Y machine". You can rent someone else's machine for that.

> and at quite a low cost

So, let's actually look at that. For a "low-budget" 8K pipeline you'd first need a camera that can shoot at 8K or more. Ideally you'd want more so you can crop-in or do stabilization in post. So you'd be shooting on the Blackmagic Ursa Mini Pro 12K (~$6k) [1].

Usually you're using 5-10% of the material that you shot in the final edit, for some movies that can go as low as 1% (e.g., Apocalypse Now [2]), but let's calculate with a 45min documentary and 7.5% material used.

That means you've now got 30 TB of raw material [3]. You'll obviously use proxies for editing, but at least for color grading and for delivery you'll need access to the full material.

So now you'll need to store 30 TB of raw material somewhere in the cloud accessible to the machine that's doing the delivery. Even assuming you've got a symmetric fiber connection so we can disregard potential traffic limits and transfer speeds for initially uploading the material, you'll still need to pay for the cloud storage.

Creative Cloud has a storage limit of 100 GB for individuals or 1 TB for business plans [4], which is obviously far too limiting, so you'd need to use something like LucidLink. As you'd be working with large video files, you'd need high-performance storage, so you'd have to calculate with their performance plan, which is another $80 per TB per month, so $2400 per month just for this one single project [5].

________________________

[1] https://www.bhphotovideo.com/c/product/1578059-REG/blackmagi...

[2] https://books.google.de/books/about/?id=wB7cAAAAMAAJ

[3] https://www.braw.info/capacity/

[4] https://www.adobe.com/creativecloud/plans.html

[5] https://www.lucidlink.com/pricing

And of course all of the above may be "peanuts" for professional studios, but I am not aware of any hardware/workflow/money/things you can throw at the problem to make the editing and production experience smooth. There are tons of inconveniences, practical barriers, bandwidth issues (and I'm not even talking about network/cloud bandwidth, that's a whole other thing-- even just disk IO, moving stuff around via USB-C, backing up stuff, etc... It's all super labour-intensive and annoying).

Well, if you're editing with proxies (which is really easy with Resolve or Media Composer), the editing experience is really smooth. But that doesn't help for grading or delivery, where you'll still need the full resolution files.

Even with Gen 4 NVMe storage you'll quickly hit bottlenecks at those resolutions.

You'd be surprised at how much buffering and loading is needed to just playback 1080p content (even directly from your local machine's built-in SSD) in the video editor before it's rendered. It's incredibly frustrating to do video work on "regular"/prosumer (macbook pro) hardware.

> After Effects

A lot of that is on Adobe more than anything else, mind you. Same with Premiere, too. These tasks can be a lot faster than Adobe software allows them to be. But there is a limit, of course

Asking as a person conpletely out of that world: Can DaVinci Resolve or Final Cut Pro do the same as After Effects and be faster?

I don't think these products are directly comparable. I think Final Cut Pro probably maps better to Adobe Premiere Pro... But even then, it's not a complete overlap. There are things you can easily do in Premiere that you can't easily do in FCP, and vice-versa. In general, I find that Premiere is a bit more powerful/flexible than FCP, but FCP is much better software and does what it does much smoother and much faster.

No, they're for completely different use cases. While it's technically possible to edit a video in After Effects, its main use case is compositing VFX and advanced motion graphics.

Agreed 100%, and I hear Final Cut Pro has much better performance.

But realistically, a lot of work happens on Adobe or other (Autodesk, Houdini, etc...) products that also have their own issues.

Also, from experience: The bitrate and bandwidth of streaming services is so bad that the actual resolution doesn't really matter.

A good high-quality 1080p export will have significantly higher perceived quality than a "4K" (UHD) video on YouTube, Netflix, Amazon Prime or Apple TV.

I've decided that I'll produce all content in 1080p at the current time, even though I'm recording video in DCI 4K oversampled from a 6K sensor. No viewer is ever going to see the difference anyway.

There are several studies which say that viewer emotional engagement caps out at 1080p, and the only thing that drives more engagement beyond 1080p is HDR/WCG and HFR.

Customers are also not as willing to pay for 4k as those invested in UHD would like to think. It's a "nice to have" and having big numbers makes people feel good, but it doesn't actually make them love the content more.

The demand for it would be from a minority.

A lot of people are watching content on their phones, for one thing.

I create some travel video content and up until last year, most of my clients weren't even fussed about 4K. I had been shooting at 5.1K for most of last year and then dropped to 4K this year because no one has needed it, but the storage and delivery costs are higher.

Everytime you increase the resolution all the requirements in term of cpu, ram, bandwith, storage explode for production, media distribution and end user explode.

It is not just replacing one screen for another on client's side.

Right, and then there’s HUDs!

> TV resolution goes at the pace of available content

there is also AI upscaling which can upscale content to higher resolution

Reply on news.ycombinator.com