Jump to content
Daniel D. Teoli Jr.

If cell phones are going to 8K why are film scanners so behind?

Recommended Posts

I read cell phones are going to 8k video shortly. With all the $$ film scanners charge why isn't 8K or even 12K the standard?? 

Share this post


Link to post
Share on other sites

Also it's unlikely cell phones would actually resolve 8K. The file might have 8k pixels - but they probably won't have 8k worth of visible detail. The lens and the compression being the limiting factor.

Also at the moment there isn't a D-cinema projector that's above 4K. So a 8 to 12k scan couldn't be screened - unless you lasered it out to IMAX stock and screened it in one of the 10 or so 15/70 screens.

Personally I have to be very close to the screen to see a difference between 2k and 4k. Above that your into diminishing returns. 

Share this post


Link to post
Share on other sites
2 hours ago, Stuart Brereton said:

Because there's no point in scanning 35mm at 8 or 12k. After about 6k, there's no more information to be had.

There's plenty more on film than you think, and it depends on your goal. If the goal is digital preservation, higher resolution is always better. The whole point is to resolve the grain, because the grain is the picture. You may not be able to resolve more than the lens on the camera and lens were able to when the film was shot, but that's not really the point. In order to have the most to work with, especially with film that's degrading fast and won't be around in 10 more years, you want the highest resolution you can get. The idea is to future proof the scan as much as possible so when crazy algorithms for restoration are available in 10 years you have a suitable source to work with. Also, oversampling is important. if you need an 8k scan for display on an 8k screen, you need to scan at a higher res to downsample to that 8k for best results. 

To Daniel's question: There are a few scanners out there that can do those resolutions (Lasergraphics Director 10k, some of the GoldenEye scanners, and a couple others I think). We're building a scanner right now that will do 14k for 70mm and roughly 10k for 35mm. Our ScanStation has a 6.5k sensor in it. 

That said, 8k on a cell phone is nothing in comparison to the data rates required to do a proper film scan. You're comparing apples and oranges (a cell phone is capturing crappy 8-bit highly compressed files, but a film scanner is capturing uncompressed). The bandwidth requirements are beyond what even the most high end computers can handle right now, so any scanner with resolutions like that is going to be limited to a few frames per second tops and any post work is going to be done on lower res proxies.

  • Like 1

Share this post


Link to post
Share on other sites
1 hour ago, Perry Paolantonio said:

There's plenty more on film than you think, and it depends on your goal.

The whole point is to resolve the grain, because the grain is the picture.

I believe that 4k is capable of capturing all the information up to the 80 lp/mm MTF that Kodak publishes for its neg stocks. Throw in some over-sampling to be safe, and you've got 6k.

As you say, you can continue to scan at ever higher resolutions, but whether you are gaining any useful information is open to question.

  • Upvote 1

Share this post


Link to post
Share on other sites

In theory if a film frame resolves 4K of actual detail, it would have to be scanned at 8K to avoid aliasing according to Nyquist. But the reality is that most film frames resolve less detail and a lot of the time, that detail isn’t prone to aliasing. I think scanning at 6K for a 4K finish would probably be optimum in most cases.

  • Upvote 1

Share this post


Link to post
Share on other sites
1 hour ago, Stuart Brereton said:

I believe that 4k is capable of capturing all the information up to the 80 lp/mm MTF that Kodak publishes for its neg stocks. Throw in some over-sampling to be safe, and you've got 6k.

As you say, you can continue to scan at ever higher resolutions, but whether you are gaining any useful information is open to question.

As I see it there are three reasons to scan at higher resolutions than you need:

1) Oversampling, nyquist, etc.

2) Increased fidelity. Don't think about the picture on the film, think about the film. Most people can't hear the difference between 44.1kHz audio sampling adn 48kHz. Yet every self respecting audio mastering house will work at 96k or higher. Why? Because audio is analog (and so is film), and more samples gets you closer to the original analog signal. Do you need it immediately? maybe not. But 10, 20 years from now, it may come in very handy. And if the film doesn't exist anymore, it'd be pretty foolish not to have scanned at the highest resolution you could, at the time. 

3) It's only a matter of time before 8k becomes a theatrical reality and possibly even takes over home theater (I think that's debatable, but who knows). If you scan at 4k, you have to do a lot of upsampling to get it to 8k. If you scan at 8k, you already have that ready to go when needed. 

i"m not saying everyone needs to do this, but these are totally reasonable arguments for scanning at higher resolutions. People need to think beyond their immediate needs, especially as the costs to do this kind of scanning comes down. A couple years ago, doing a 4k feature film scan for $5000 was unthinkable. Yet, we do it all the time.

Times change, resolutions increase, costs come down, computers get faster and will be able to handle it easily soon enough. 

 

Share this post


Link to post
Share on other sites

A $10k Sony Fx9 down samples from 6K to 4K.. giving a "true" 4K..  .. all this 8k stuff was to sell TV,s.. that failed ..now its to sell phones.. !!  ridiculous  microscopic pixels. . we need high quality content not more pixels .. marketing droids  and bench top ,pixel peeking "visual consultants " ..  be gone I say !..

Share this post


Link to post
Share on other sites
7 hours ago, Perry Paolantonio said:

2) Increased fidelity. Don't think about the picture on the film, think about the film. Most people can't hear the difference between 44.1kHz audio sampling adn 48kHz. Yet every self respecting audio mastering house will work at 96k or higher. Why? Because audio is analog (and so is film), and more samples gets you closer to the original analog signal. . 

 

Your argument here seems to be about the need for higher resolution capture devices, not scanners. Audio engineers use 96kHz to capture sound from the original source, to try to preserve all of its subtleties. With film, that initial capture of information was done when the film was exposed, and is limited by the resolving power of the film. Once you’ve scanned it at a high enough resolution to capture that information, there’s nothing more to be had. You can’t create information where there is none. Ok, an 8k scan means you won’t have to upsample it in the future, but in a sense, you’ve already done that by stretching 4K of information to fill an 8k file.

You’re right that technology will quickly become fast enough and cheap enough that there will be no reason not to scan at higher resolutions, but that doesn’t mean it will be necessary.

Share this post


Link to post
Share on other sites
18 hours ago, Stuart Brereton said:

Because there's no point in scanning 35mm at 8 or 12k. After about 6k, there's no more information to be had.

Don't think that is true.

Share this post


Link to post
Share on other sites
6 hours ago, Stuart Brereton said:

Your argument here seems to be about the need for higher resolution capture devices, not scanners. Audio engineers use 96kHz to capture sound from the original source, to try to preserve all of its subtleties. With film, that initial capture of information was done when the film was exposed, and is limited by the resolving power of the film. Once you’ve scanned it at a high enough resolution to capture that information, there’s nothing more to be had. You can’t create information where there is none. Ok, an 8k scan means you won’t have to upsample it in the future, but in a sense, you’ve already done that by stretching 4K of information to fill an 8k file.

You’re right that technology will quickly become fast enough and cheap enough that there will be no reason not to scan at higher resolutions, but that doesn’t mean it will be necessary.

The issue with digital is that what once was amazing is now last gen. I remember watching avatar going ‘wow this looks amazing’. I rewatched it the other night thinking ‘it sort of looks more and more like video game graphics’. I remember with 720p, it was revolutionary. Why would anyone need more than that? Now a 720p visually looks poorer in quality. As the human eye evolves, and it is evolving. The difference between 8k and 4K will grow. The difference between 4K and 1080p will grow. The thing with digital is that if you watch a 480p video. The pixels are a harsh lapse of quality. They are tiny box’s, no real smoothness to them. Film however even though arguably doesn’t have more than 4K worth of information. The sort of lapse of quality is smooth. It isn’t janky like digital. Scanning at 8k does future proof it. Yes technically you’d be upressing it by a fair bit but like I said it’s different. One of the reasons why I love film. 

This however is complete opinion with very little fact. More personal taste than anything else.

Share this post


Link to post
Share on other sites
1 hour ago, Daniel D. Teoli Jr. said:

Don't think that is true.

Daniel, Kodak's own technical literature states that their motion picture stocks can resolve 80 lp/mm. That translates to a maximum of about 4k of information. If you oversample, say to 6k, you should be able to capture all the available information in that negative.

Share this post


Link to post
Share on other sites
25 minutes ago, Gabriel Devereux said:

The issue with digital is that what once was amazing is now last gen. I remember watching avatar going ‘wow this looks amazing’. I rewatched it the other night thinking ‘it sort of looks more and more like video game graphics’. I remember with 720p, it was revolutionary. Why would anyone need more than that? Now a 720p visually looks poorer in quality. As the human eye evolves, and it is evolving. The difference between 8k and 4K will grow. The difference between 4K and 1080p will grow. The thing with digital is that if you watch a 480p video. The pixels are a harsh lapse of quality. They are tiny box’s, no real smoothness to them. Film however even though arguably doesn’t have more than 4K worth of information. The sort of lapse of quality is smooth. It isn’t janky like digital. Scanning at 8k does future proof it. Yes technically you’d be upressing it by a fair bit but like I said it’s different. One of the reasons why I love film. 

This however is complete opinion with very little fact. More personal taste than anything else.

You're right that there are big differences in quality between 480p, 720p and 1080p. However, continued increases in resolution yield diminishing returns. Many people find it hard to tell the difference between 2k and 4k. In fact, 4k seems to be the sweet spot for human perception. Higher resolutions than this are visually indistinguishable.

Have a look at these videos from Steve Yedlin, ASC, where he compares identical footage, film and digital, shot at different resolutions.

http://yedlin.net/ResDemo/

  • Upvote 1

Share this post


Link to post
Share on other sites
21 hours ago, Perry Paolantonio said:

...

Most people can't hear the difference between 44.1kHz audio sampling adn 48kHz. Yet every self respecting audio mastering house will work at 96k or higher. Why? Because audio is analog (and so is film), and more samples gets you closer to the original analog signal. Do you need it immediately? maybe not.

 

 

The problem with the audio analogy is that there's a very specific reason for having chosen 44.1 / 48 kHz, and that's due to the upper limit of human hearing; Nyquist sampling theorem as it pertains to audio says you can capture the entire audible sound spectrum by sampling 2x the highest audible frequency (~ 20kHz, and that goes steadily downhill from your 20s onwards!). A pure fundamental tone (sine wave) can be perfectly described with two bits of data, the peak and the valley. And digital to analog converters are able to "perfectly" reproduce a sine wave from just those two points, there's no audible aliasing or "stepping".

And for the record (no pun intended), many recording studios record in 44.1kHz, though modern DAWs and plug-ins internally process at higher resolutions, to avoid accumulating rounding errors - just ilke bit-depth in colour correction apps.

(There are actually valid reasons to record at 96kHz but they're more to do with latency, not the sound itself.)

Share this post


Link to post
Share on other sites
On 2/7/2020 at 10:16 AM, Daniel D. Teoli Jr. said:

I read cell phones are going to 8k video shortly. With all the $$ film scanners charge why isn't 8K or even 12K the standard?? 

I mean it's quite simple, 8k cell phones are going to use extreme data and chroma compression. The whole point of a film scanner is that you want to scan it so you can correct it later. So compression needs to be very low and you need to have an imager that can not only capture at a great deal of dynamic range, but do so at a very low noise floor. It's a very tricky thing and as a consequence, very costly. There are 10k scanners out there in the wild, but they're grossly expensive because of this reason. If Blackmagic were to use their new 6k imager in the Cintel II, it would be a game changer because it has a low noise floor and decent dynamic range, thanks to a double preamp system, very much like the Alexa. It can capture blacks and highlights on two different preamps, thus negating the need for multiple passes. Currently the HDR version of the scanner needs to do multiple passes. 

I think a $30k 6k scanner that does real-time 12 bit 444 HDR capture is all anyone needs. Higher then that is silly for anything else but large format. 

 

Share this post


Link to post
Share on other sites

Manufacturers of digital sensors will continue to make higher resolution units with better specs.

And scanner manufacturers will put those sensors in the scanners.

Look how fast everyone with a Scan Station 5K dumped that (noisy) sensor for the 6.5K Sony Pregius one.

 

Share this post


Link to post
Share on other sites
On 2/8/2020 at 5:28 PM, Daniel Lo Presti said:

 

The problem with the audio analogy is that there's a very specific reason for having chosen 44.1 / 48 kHz, and that's due to the upper limit of human hearing;

Of course. That said, I can't hear the difference between 44.1 and 48, but I know audio mastering engineers who can. However, we're not talking about straight recording and then listening to that recording. Just like with film scanning, you're capturing it and then further manipulating it. And for that, more samples are critical.

With an image, you may be doing a lot more than resizing. Let's say you want to do grain reduction (shudder), you're going to get better results if you have better definition of the grain, therefore you're less likely to affect the underlying image this way. Or if you're doing restoration work - the algorithms will work better with sharper images. We've been doing digital restoration for 15 years and I can tell you that the quality of the restoration work increases with resolution. We just get better results.

The point most people miss is that while the film/lens/camera system can only resolve so much, and that's true, that doesn't mean there's some hard and fast pixel count that corresponds to a given film gauge. That's a gross oversimplification of the problem. For one thing, pixel count is only one small factor in the overall image (I can make you a pretty crappy 8k image that's still 8k, but doesn't look as sharp as 4k). The quality of the optical system in the scanner, the quality of the sensor, noise in the sensor, the quality of the scanner's light source, post-scan image processing, and dozens of other factors play into the quality of a scan.

And the scan is not the final step in the post production workflow, yet it's treated that way. You're never scanning the film and then looking at it immediately, you're always doing more to it afterwards: Grading, compositing, restoring, etc.

Does 8k television make sense? Probably not, unless you're one of those people with a 10 foot screen and you're sitting 4 feet from it. Then, the pixel density makes a difference. But the reality is that display resolutions keep increasing, so if nothing else, film scans should keep up, if only to avoid upsampling later to fit those higher resolutions. But there are plenty of good reasons to scan at a higher res than you think might be necessary. 

 

 

 

Share this post


Link to post
Share on other sites
23 hours ago, Perry Paolantonio said:

Of course. That said, I can't hear the difference between 44.1 and 48, but I know audio mastering engineers who can. 

With sampling it's not just the nyquist theory that defines perfect reproduction.  All sampling systems require a low pass filter to prevent aliasing. With audio you want a low pass filter that preserves audio upto 20khz , and on 44.1Khz you'd need 100% cut at 22Khz - that filter is very hard to build and probably would distort the sound in some way. 48Khz means you need 100% cut at 24kHz - thats a much more gentle filter. 

So the difference in sound quality is not just due to the sample rate - but the type of low pass filter needed. 

In audio the main point of oversampling is dealing with issues of aliasing that would be audible, even when most people can't hear frequencies above 18khz 

  • Like 1

Share this post


Link to post
Share on other sites

Sound designer chiming in on the audio thing:

The big reasons people will record in 96khz is future proofing and flexibility. Slowing down sound effects that are 96khz retain far more of the high-end frequencies as opposed to slowing down a 44.1khz sound effect (or song for sampling, whichever). The needing to hear in 96k dips more into the audiophile realm as opposed to the engineering realm.

I've never heard of putting a low pass filter when mastering modern-sounding audio because every program just does that automatically, was there a time when it had to be done manually? I can't even hear anything past 17khz and have never gotten a mixing complaint.

  • Like 1

Share this post


Link to post
Share on other sites

Also I would say any Bayer mask CFA sensor will have limitations built into it in terms of color fidelity, the dyes used on a silicon mask are far from perfect and far from perfectly applied to the individual photosites.

IMO a "real" RGB scan from a monochrome sensor with dichroics or a RGB led sequential exposure will continue to be the standard by which scans are judged and why film will continue to have a special look with different color fidelity than math made CFA cameras no matter how "good" they may be.

  • Upvote 1

Share this post


Link to post
Share on other sites
On 2/12/2020 at 4:58 AM, Robert Houllahan said:

Also I would say any Bayer mask CFA sensor will have limitations built into it in terms of color fidelity, the dyes used on a silicon mask are far from perfect and far from perfectly applied to the individual photosites.

IMO a "real" RGB scan from a monochrome sensor with dichroics or a RGB led sequential exposure will continue to be the standard by which scans are judged and why film will continue to have a special look with different color fidelity than math made CFA cameras no matter how "good" they may be.

Indeed, I remember how angry peeps on Reduser got when someone posted a test that showed how much better the F23 looked then the RED One. Simply impossible! It's only HD, 4K must be better because its erm 4K

  • Upvote 2

Share this post


Link to post
Share on other sites
On 2/11/2020 at 3:07 PM, Max Field said:

I've never heard of putting a low pass filter when mastering modern-sounding audio because every program just does that automatically, was there a time when it had to be done manually? I can't even hear anything past 17khz and have never gotten a mixing complaint.

The low pass filter is usually built into the Analogue to Digital converter. The aliasing distortion would be audible on high frequency sounds otherwise.

You could probably get away without a low pass filter on a 96Khz audio system as few mics can record audio frequencies above 25-30khz they would act as natural low pass filters

Share this post


Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.



  • Serious Gear



    Paralinx LLC



    Abel Cine



    Gamma Ray Digital Inc



    Rig Wheels Passport



    G-Force Grips



    Broadcast Solutions Inc



    Metropolis Post



    FJS International



    Tai Audio



    Media Blackout - Custom Cables and AKS



    Just Cinema Gear



    Glidecam



    The Original Slider



    Ritter Battery



    CineLab



    Visual Products



    New Pro Video - New and Used Equipment



    Wooden Camera


    Cinematography Books and Gear
×
×
  • Create New...