Why 4K matters - and why it's not ready for prime time

[feppe]

This could mean that the up-conversion from 24p (roughly by splitting each of the 24p frames into two half frames) retains timing aspects of 24fps, like a 180º shutter angle meaning blurring motion over 1/48s and the position of each part of a moving objects still only updated at intervals of 1/24th of a second with a gap of 1/48s between the end of recording on one image and the start of recording of the next, whereas an image recorded at 50i has less temporal blurring and more rapid position updates.

That is my understanding as well.

Movies are shot at 24p, and converting to 50i in post will not affect motion blur. Motion blur can't be adjusted without heavy lifting in post-process - it's essentially the same as getting rid of or adding motion blur in Photoshop to each frame, amount depending on subject matter and camera movement. Thus increasing framerate in post will have some detrimental impact on the cinematic look, but the motion blur will remain.

That has my curiosity piqued: I should go to a TV shop to see how interpolating TVs look with feature films.

Here's a list of theaters showing The Hobbit in 48fps. Since it's been almost a year, it's not showing anymore at most theaters. The upcoming sequels will also be in 48fps, and I believe Avatar sequels will also.

How does "shutter angle" work with the continuous sensor output of interlaced readout on a video camera? Is it like shutter angle of 360º, with no "gaps"?

I don't shoot motion, I'm just a film geek, so I'll let those who actually know what they're talking about answer that I'm curious myself.

Anyway, if 4k becomes the rage or a magic number that people want, the camera, software and computer companies are more than happy to oblige, because they'll just sell more stuff and more stuff makes the world spin.

Not only about tech, but technicians and pros (jobs). All visual effects need higher fidelity, make-up and prosthetics needs to be more detailed and higher quality, lenses need to resolve more, etc. Focus pulling will become even more critical with 4k than it already is with digital and the shallow-DOF craze. You can see out-of-focus face shots in many (HD)TV shows and even some feature films. These were much rarer in film days - but admittedly viewing DVD movies on SDTV hides a lot of imperfections which are visible on BD and HDTV.

I share your enthusiasm for RED at the viewing end. I saw Elysium in IMAX, which was shot with 4K RED EPIC, and it was stunning.

I thought the goal of 48fps was so that 3D could be presented in 24fps "per-eye"?

Yes, but that's just one goal. 2D 48fps allows for smoother (and faster) pans, for example.

[Rhossydd]

I can only assume that your jibe was directed at me. If so, so be it, though it would have been braver to be more direct.

Not especially at you, there  are plenty of others that seem to think posting a few home video on Vimeo gives them the authority to be an expert. I've said before, you'd retain more respect by sticking to subjects you really excel at.

Credible experience. Humm. Worked in the motion picture industry since 1966. Past member of IATA, National video product manager for JCV, Panasonic, Colortran Studio Lighting, on staff at CBC television for nine years, and have worked with widely respected industry veterin and cine award winner Chris Sanderson producing dozens of productions for the past eleven years. Guess there's a bit of cinema and TV experience here.

If you'd given some actual credits of productions you've worked on made for cinema, maybe membership of BSC/ASC or similar, then you might have some credibility as a cinematographer.
Just selling stuff and rubbing shoulders with film makers doesn't really count.

As for you comments on 24fps, sorry, you're now at odds with some of the great cinematographers of the past century. Do some reading and online research. It won't take you long to realize that rejecting 24fps is a facile position.

Missed point, read what I said, 24fps isn't a necessity for cinema any more, ask Peter Jackson.

Similarly throwing up straw man arguments against shallow DOF and other cinematic conventions is childish. Examples of anything can be found, but these exceptions prove nothing other than the shallowness of your argument.

Again you're not getting it. I'm not saying shallow DoF is wrong, just that it isn't the defining property of a cinematic look.

[hjulenissen]

I don't shoot motion, I'm just a film geek, so I'll let those who actually know what they're talking about answer that I'm curious myself.

I dont shoot motion either (neither am I an expert), but is it my impression that there are several different ways that a video camera may operate. Sometimes user-selectable.

The "best" would (from many persepectives) be a true progressive signal feeding a true progressive stream at whatever desired framerate.

I believe that some "interlaced" video cameras works in the intuitive way: reading odd lines at time 0, storing them as an interlaced field, then reading even lines at time 1, storing them as an interlaced field. This would give you an e.g. 50i of 50 discrete fields /second where each field is subject to whatever integration time/mechanical shutter is used.

I believe that other "interlaced" cameras does spatio-temporal filtering: Each line of each field may be a weighted sum of neighboring (time/space) fields. This might improve noise, or it might reduce aliasing (although aliasing seems to be a desirable property of interlaced systems).

I believe that other "interlaced" cameras will simply grab a progressive frame 25/30 times/second, and split the lines into a 50/60i signal.

-h

[jjj]

I have always found the argument over 24p Vs 30P a bit odd, as the difference in look between the two is due is almost certainly down to other factors like capture medium or budget and not FPS.
Images viewed at 24fps tend to be seen in the cinema and were usually shot on film, usually S35 [i.e. a large sensor in video terms] and with large budgets, careful lighting and so on.
Stuff shot at 30fps [until quite recently with the advent of DSLR video capture] tended to have been shot on smaller chip video cameras which gives a very, very different look to film, both through sensor size and image quality of video vs film, not to mention much lower budgets too. A lot of TV has fast turnarounds compared to film and because lighting for say soaps or sitcoms tends to be set up for multi-camera shooting, you get a less interesting and a much flatter look. Bearing in mind that lighting can also cause separation, albeit in a different way to a fast lens on a big sensor/film. And if you were to shoot these things in 24p it would make bugger all difference to the overall look in my view. Which is probably why BC can't see a difference between frame rates as it isn't the defining thing that changes the look.

So do films when shown on TV at 30 fps suddenly not look cinematic? Can't comment myself as I do not tend to watch TV whilst I'm in the US, TV is 25fps here and although the look of films does not seem to change on UK TV, the impact of seeing things on a much smaller screen does.

I never saw The Hobbit to see how that looked in 48fps, because I wasn't exactly a fan of the tedious LOTR films and even as a 10 year old I thought the Hobbit was a childish book. Interestingly Douglas Trumball, who did SFX on 2001, Blade Runner, Tree of Life etc did tests with 35mm film shot+projected at different rates and concluded faster was better for emotional involvement of audience, which doesn't surprise me actually as higher sampling rates in music, which cannot be directly heard makes audio sound better in my experience. He then developed Showscan [65mm @ 60fps] to utilise his findings which was too ahead of its time to be commercially viable.
Just saw Kick-Ass 2 which was rubbish in many ways, not least of all because it looked like a cheap soap with the nasty look people ascribe to 30fps, despite being shot at 24fps. One of the ugliest looking films I've seen in a long time.

[hjulenissen]

...which doesn't surprise me actually as higher sampling rates in music, which cannot be directly heard makes audio sound better in my experience.

Despite a lot of effort, no-one has to my knowledge been able to credibly prove (i.e. repeatable, peer-reviewed science) that >CD spec digital audio leads to audible improvements.

Moving images are very different, as todays technology can be proven to be worse than the capabilities of our vision on a number of parameters.

-h

[Manoli]

Despite a lot of effort, no-one has to my knowledge been able to credibly prove (i.e. repeatable, peer-reviewed science) that >CD spec digital audio leads to audible improvements.

Huh ?
No audible improvements ? And how exactly do you define audible improvements ?

CD spec is 44.1/16bit. HD is usually 24bit and > 96k.
A bit like saying, there's no visual improvement between an 8-bit jpeg and a 16-bit RAW, or digital is no better than film.

Perhaps you should refer to Daniel Weiss.

[hjulenissen]

Huh ?
No audible improvements ? And how exactly do you define audible improvements ?

>.95 confidence in a blind test suitable for such test (e.g. ABX)

CD spec is 44.1/16bit. HD is usually 24bit and > 96k.

Yes?

A bit like saying, there's no visual improvement between an 8-bit jpeg and a 16-bit RAW, or digital is no better than film.

Nowhere like saying anything like that.

Perhaps you should refer to Daniel Weiss.

I don't know him. I do subscribe to and read scientific journals about sound. Such as the article below:

http://www.drewdaniels.com/audible.pdf

[Manoli]

-h
I don't want to hijack this thread BUT, briefly
Your article contains so many caveats it's almost self defeating

“From the many different recordings we used it emerged that almost no music or voice program, recording venue, instrument, or performer exceeds the capabilities of a well implemented CD-quality record/playback loop. The CD has adequate bandwidth and dynamic range for any home reproduction task,”

well implemented ? home reproduction ?

to begin to understand the many variables that go into a 'recording' try
http://www.digido.com/articles-and-demos12/13-bob-katz/27-back-to-analog.html

[jjj]

Huh ?
No audible improvements ? And how exactly do you define audible improvements ?

it sounds better. 

When CDs first appeared, I recall testing them against vinyl and the thing that struck me was that I tapped my foot to the vinyl track and not the CD. I only realised this after the fact and being quite surprised by it. Later when players increased their sampling rate, they sounded better and only then did I buy one. S

CD spec is 44.1/16bit. HD is usually 24bit and > 96k.
A bit like saying, there's no visual improvement between an 8-bit jpeg and a 16-bit RAW, or digital is no better than film.

No it's like saying a 24MB image looks better than a 8MB image, or a MF print looks better than a 35mm print. You don't necessarily see the underlying structure of the image under normal viewing. But the higher resolution/quality image just looks better, all other things being equal.


P.S.  Folks -  please, please don't go down the pointless 'I can't hear any difference, so you crazy audiophiles must be imagining things' route.

[BJL]

No audible improvements ? And how exactly do you define audible improvements ?

I propose "Detectable at a level well above chance in ABX testing".
And certainly not a list of differences in the procedure used to produce the recording, or "some of the spec. sheet numbers are bigger, so it must be better", which is what you seem to propose next:

CD spec is 44.1/16bit. HD is usually 24bit and > 96k. ...

Can you cite any positive evidence of detectable audible differences, detected in ABX comparisons or the like? I have read examples of "golder ears" reviewers describing in elaborate detail the differences that they perceive between two high-end alternatives, and then failing to distinguish the two at all in an ABX comparison.

[jjj]

Interestingly Douglas Trumball, who did SFX on 2001, Blade Runner, Tree of Life etc did tests with 35mm film shot+projected at different rates and concluded faster was better for emotional involvement of audience, which doesn't surprise me actually as higher sampling rates in music, which cannot be directly heard makes audio sound better in my experience.

Despite a lot of effort, no-one has to my knowledge been able to credibly prove (i.e. repeatable, peer-reviewed science) that >CD spec digital audio leads to audible improvements.

Moving images are very different, as todays technology can be proven to be worse than the capabilities of our vision on a number of parameters.

Please try reading posts before replying, there's a good chap. Did I mention CDs? Nope, I said music.  Would you say a 64kbs mp3 sounds as good as a 320kbs or a lossless format or compared to the original.

[hjulenissen]

... Would you say a 64kbs mp3 sounds as good as a 320kbs or a lossless format or compared to the original.

No, because listening tests shows us that there (can be) an audible difference. Are you switching topic from sample rates to bit rates?

...which doesn't surprise me actually as higher sampling rates in music, which cannot be directly heard makes audio sound better in my experience.

Can we please talk about a topic and not about those that discuss the topic?

Please try reading posts before replying, there's a good chap.

-h

[hjulenissen]

Your article contains so many caveats it's almost self defeating

The article is peer reviewed and published in the most relevant scientific journal. You are, of course, free to attempt to publish an article of your own, or to publish comments to this article in the JAES.

“From the many different recordings we used it emerged that almost no music or voice program, recording venue, instrument, or performer exceeds the capabilities of a well implemented CD-quality record/playback loop. The CD has adequate bandwidth and dynamic range for any home reproduction task,”

well implemented ? home reproduction ?

I.e. it may be _possible_ to design a playback system (or playback situation) where errors that are inaudible in normal situations, suddenly become audible. Try adding 40dB of gain at at narrow band around 1kHz, or applying 70dB of dynamic compression. Or equipping a CD record/playback loop that uses 8-bit ADC and DAC. This article tries to tell us what is needed from a system designed by competent people, not what is needed from a system designed by monkeys.

There are reasons to use >CD specs when recording music in your studio, just as there are reasons to use 14-bit raw files in photography while 8-bit JPEG may suffice for distribution.

to begin to understand the many variables that go into a 'recording' try
http://www.digido.com/articles-and-demos12/13-bob-katz/27-back-to-analog.html

Many noted recording engineers don't really have a clue about the issue we discuss. Really.

-h

[Manoli]

Later when players increased their sampling rate, they sounded better and only then did I buy one. S

I can't remember what the original CD sampling rate was but 44.1 has been the standard for a long time. The major improvements, if I'm not mistaken, came when the A/D converters improved, and became 'seperates', moving them out of the CD player.

No it's like saying a 24MB image looks better than a 8MB image, or a MF print looks better than a 35mm print. You don't necessarily see the underlying structure of the image under normal viewing. But the higher resolution/quality image just looks better, all other things being equal.

Agreed. You put it far more succinctly. The important part here is the "all things being equal" - which in the audio world, they rarely are.

Another example, as Jeff Schewe illustrates in his book, the undetectable difference between a Canon S100 shot and the P65+ identical shot as printed. The same two shots enlarged to A0 size - a different story.

"some of the spec. sheet numbers are bigger, so it must be better", which is what you seem to propose next:

No. I was referring to both the bit rate and the sampling rate.
http://en.wikipedia.org/wiki/Sampling_rate

"The Audio Engineering Society recommends 48 kHz sample rate for most applications but gives recognition to 44.1 kHz for Compact Disc and other consumer uses, 32 kHz for transmission-related application and 96 kHz for higher bandwidth or relaxed anti-aliasing filtering."   (Can't remember where I've seen that term used before!)

I have read examples of "golder ears" reviewers describing in elaborate detail the differences that they perceive between two high-end alternatives, and then failing to distinguish the two at all in an ABX comparison.

Yes, but they're discussing differences in hardware as opposed to whether or not different sampling/bit rates make any noticeable difference.

FWIW, my own empirical experience is this:
Yes, sampling rates can and do, sometimes, make a substantial difference.
The quality of the recording is more important than the output medium. A poor HD recording pales next to a good CD recording.

I own both a Musical Fidelity DAC (about $200) - truly excellent and excellent value for money and a WEISS DAC202 (about $ - I'd rather not say) an exceptional DAC. A difference between them exists, but has no relevance to the price differential. Nevertheless, good as they both are the WEISS makes a difference not only on HD recordings but also on iTunes 192K downloads.

It is quite true that some esoteric high-end audio engineers/manufacturers have debated the value of exceeding the 96k/24 bit level - but at this point I check out - this is the domain of audio engineers, which I am not.

[Manoli]

Many noted recording engineers don't really have a clue about the issue we discuss. Really.

I'll take your word on that. Really.

[Manoli]

... a system designed by competent people, not what is needed from a system designed by monkeys.

Now, aren't we getting a little bit over-excited ?

[jjj]

Can we please talk about a topic and not about those that discuss the topic?

Hard to talk about a topic if you don't bother to read posts correctly and then wander off on a different subject.

Plus the thread is about moving imagery, frame rate and resolution and so forth.

[Peter McLennan]

So the relevance of all this audio discussion to 4K is what?   

Are we saying this?: "4K doesn't reveal enough improvement in picture quality for the average viewer at home, and just like HD sound isn't worth it, neither is 4K picture"

If so, I agree.

(note: I said "for the average viewer at home")

[jjj]

So the relevance of all this audio discussion to 4K is what?

No relevance, it was to with frame rate.

[bcooter]

Michael's right if he likes 4k and 24fps.  

I can see his point because he instructs and shows still landscape imagery and at 4k he can show the detail he likes to present.  Makes sense to me, but his delivery method of sd cards sounds a little problematic, but I'm sure he'll work it out.

Then again so are the guys that shoot video for a show like Top Gear are also right,  because they shoot a very good video production and I don't think they are worrying about frame rates or 4k.

JJJ is right because the "cinematic look" is more than cameras, lenses, focus throw and fps.  It's the total look and films, even cheap films take a long time to get right.

It just depends on what you do.

What is interesting though that most people who have moved from stills to motion, are not going to be directing and shooting full length features, at least none that will go into a cinema.

After all there are great DP's and Directors in Hollywood that can smoke us all that aren't overly booked.  Why do you think so many of them are giving classes, testing cameras for manufacturers and doing how to videos?

I've worked on gigs where an academy award winning director, (one who has a current movie hitting the screens as I write this) worked as DP where the director knew less about his craft than most amatuer film makers but the award winning director did it because . . . a gigs a gig.

Still, comparing 4k to 2k on a movie screen is almost a mute point for this forum because that's not what most of us are hired to do.

Me, I think if I had to describe what I did now I'd say multi media.  Shooting MOS lifestyle on the beach with a moving camera is not a lot different than shooting the stills, other than you gotta stop shaking around and I know that when a client reviews dailies the last thing they care about is camera make, 4k or 24fps.  

They care about the shot that resonates with them, their product, their marketing plan, or just personally turns them on.

Lately we've booked a lot of dialog work and dialog vs. mos is night and day.  I love dialog directing, but find the pace slow and I really have to forget most of what I know from the still world, to do dialog correctly.

The point I was trying to make in my earlier too long responses was learn the basics (the same points I think Michael was trying to make) and then have a reason if break the rules, because really . . . there are no rules.

I can pull up google info on the last 4 movies I've seen, last 10 tv shows and I can promise you not one will have the same workflow, camera lens combo, editorial suite, coloring and finish apps as the other.

It's all a roll your own business in digital motion, much like what digital stills has and is still going through.

In fact two of my favorite tv shows moved from RED's to Sony's because they were Sony properties and I never could tell the difference when I watched them and I really explored them.

Now.  The most interesting aspect of motion imagery to me is workflow.  I think the workflow of motion is the black hole of time.  Just getting one light dailies out on a large scale is maddening and if you shoot multiple cameras with multiple file types it's pure hell.

We're finishing up a gig we shot in 5 parts, three countries, 6 cities.  The first part I shot like a mad man, because the creative brief was huge, the time allowed not so huge.  I ran gh3's, Sony FS100's, A go pro, A g16 for underwater, Two RED One's and a Scarlet.  Had a reason for each camera, mostly how fast and good I could get it in the can (or is that in the drive?).

I had a billion hours of footage and it took a lot of long, long days and nights to get it out to the client for review.

The next productions I slowed it down and shot 99% of the footage on the RED's, only used the gh3's for tight spots like cars or fluid fast movement like on the beach.

The reason had nothing to do with 4k, 2k, or fps.  I stuck with the RED's because I thought they produced the prettiest file, gave me the most head room for processing, but the main reason? . . . was Cinex.

With the RED's I could sit down for a few days, correct hundreds of files and then let the RED rocket take over for the evenings, processing out prorezz with an image that was based out and professional enough to show.

The time saved was weeks, not days so as much as I respect these new little cameras, I'm sticking with the RED's as much as possible, but once again, it's workflow.

IMO

BC