Is the future FF35?

[David Mullen ASC]

Yes, but to get good 1080P/2K RGB quality, you have to shoot at least 3K RAW if not 4K RAW. Remember that RGB resolution ends up being 3/4 of the original Bayer image.

And if 4K projection starts to take off, you may find yourself wanting to create a 4K project to take advantage of that, which means shooting in 5K RAW at least.

 

[Michael Hastings]

Having seen quite a number of recent feature films shot in 1080p HD (on 2/3" sensor cameras no less!) that looked visually stunning projected on regular suburban cinema screens. I'm sticking by my assertion that we've already reached a point of diminishing returns in regards to image quality. When you can have a s35mm film pristinely scanned at 2K, and then projected digitally at 2K (so it maintains that pristine appearance) and have it look absolutely gorgeous on a LARGE cinema screen - how much of difference is four times that resolution going to make? Are audiences going to care? Are they even going to notice? I'm not at all convinced they will.

At which point the notion of quadrupling the amount of data you have to deal with (and absorbing all of the costs that are associated with that) seems pretty unnecessary.

Yes, but to get good 1080P/2K RGB quality, you have to shoot at least 3K RAW if not 4K RAW. Remember that RGB resolution ends up being 3/4 of the original Bayer image.

And if 4K projection starts to take off, you may find yourself wanting to create a 4K project to take advantage of that, which means shooting in 5K RAW at least.

I agree which is why I suggest epic x really is the sweet spot - 5K acquisition to 4k display is probably the next noticeable jump.

From there I think you would need to go to the 9K 645 format with some way to get it printed at the highest effective resolution to Imax film for Imax display in their theaters. (I'd stick a zeiss 645 35mm lens to shoot my wide angle underwater and then I think you'd have something - 9K supersharp great whites coming at you from the imax screen)

I'm assuming no one will have an 8K projector anytime soon. BTW what do we figure the effective max resolution (in K) is for Imax size film?

 

[David Mullen ASC]

I'm assuming no one will have an 8K projector anytime soon. BTW what do we figure the effective max resolution (in K) is for Imax size film?

We were just talking about that:
http://www.reduser.net/forum/showthread.php?t=32409

The short answer is that an IMAX negative is almost 3X the width of a Super-35 negative, and since "K" refers to horizontal pixel resolution, just take whatever "K" you think Super-35 is and multiply it by 3X.

If you think 35mm neg is 4K, then IMAX neg is 12K, if you think 35mm neg is 3K, then IMAX neg is 9K. If you think that 35mm projection is 2K at best, then IMAX projection is 6K at best.

 

[Roberto Lequeux]

The cost difference from shooting 65mm over 35mm to shooting 6k over 4k is what is massive imo. The gap is going to be non-existent in comparison.. unless I am sorely mistaken in what is largely a guesstimation. Whatever number of films were shot in 65mm is in no way an accurate measuring stick.

What about shooting 5.3'k? That should be much more manageable for focusing. Red should have no issue with and enough interest in making that "mode" available on a FF35 sensor or bigger. I wonder why there is so little talk about that resolution, cause in my humble understanding you would be quite future proofed with "only" a good 4k projection.

 

[I Bloom]

I agree which is why I suggest epic x really is the sweet spot - 5K acquisition to 4k display is probably the next noticeable jump.

From there I think you would need to go to the 9K 645 format with some way to get it printed at the highest effective resolution to Imax film for Imax display in their theaters. (I'd stick a zeiss 645 35mm lens to shoot my wide angle underwater and then I think you'd have something - 9K supersharp great whites coming at you from the imax screen)

I'm assuming no one will have an 8K projector anytime soon. BTW what do we figure the effective max resolution (in K) is for Imax size film?

A DLP projector is just reflecting light off a special microchip that has tiny mirrors on it's surface.

So if they can make an 9K sensor for 40 grand why wouldn't an 8K projector be close behind.

IBloom

 

[David Rasberry]

A DLP projector is just reflecting light off a special microchip that has tiny mirrors on it's surface.

So if they can make an 9K sensor for 40 grand why wouldn't an 8K projector be close behind.

IBloom

You reach a point of diminishing returns where the perceived increase in image quality is either unnoticeable or so slight as to not be worth the large increase in cost. The processing cost scales with the total number of pixels or the square of the increase in H and V resolution over 4k. There are 4x as many pixels to deal with in an 8k image over a 4k image. The production/distribution system will eventually settle on a happy medium for most venues and the ultra high res stuff will be reserved for things like IMAX.

 

[Tom Lowe]

Of course, it's not a matter of if 4K projection becomes the new gold standard for cinema. We know for sure now -- with news from AMC and Cinemark -- that it will be, and soon.

This means you really need to shoot 6K for a super-pristine 4K finish. Yes, you could shoot 5K, but you're cutting it close, and leaving yourself basically zero headroom for cropping, reframing and stabilizing the footage. I know a lot of people say that RGB is 80% resolution of RAW, but I am much more comfortable with saying it's 60% or 70% at most, depending of course on the camera and software. That's my experience, anyway.

To me, there is no doubt, no second guessing, no question whatsoever -- Epic FF35 is the future.

 

[Tim Whitcomb]

Having seen quite a number of recent feature films shot in 1080p HD (on 2/3" sensor cameras no less!) that looked visually stunning projected on regular suburban cinema screens. I'm sticking by my assertion that we've already reached a point of diminishing returns in regards to image quality. When you can have a s35mm film pristinely scanned at 2K, and then projected digitally at 2K (so it maintains that pristine appearance) and have it look absolutely gorgeous on a LARGE cinema screen - how much of difference is four times that resolution going to make? Are audiences going to care? Are they even going to notice? I'm not at all convinced they will.

At which point the notion of quadrupling the amount of data you have to deal with (and absorbing all of the costs that are associated with that) seems pretty unnecessary.

I think you are forgetting about RED RAY. 4K at 10MB per second at REDUSER party. RED is clearly thinking ahead...and its my guess that RED RAY (and to some extent RR) is what is going to enable handling all of this insanely large file size coming our way.

 

[Scott C.]

I'm guessing the hope is that the larger sensors will afford enough additional dynamic range to win over the majority of film holdouts. I think for reasons of latitude the workflow will be to acquire at 6 or 9k and then down res to 2 or 4k for theatrical release.

There may be a law of diminishing returns in place, but there is also Moore's law. In 5 years, storing this stuff won't be as much of an issue. Putting that prediction aside for a moment... how far are you willing to go today to get a better look into your blacks, and protection for your highlights? What is parity with film worth, and how does the cost of storing that much media compare with typical stock/lab budgets today? Who has done the math? What is the tipping point?

One other question: If all other costs in money, lens and camera choices were equal, how many DP's here would have opted to shoot their last film (In whole or in part.) in IMAX?

 

[Daniel Browning]

Is the future FF35?

I have no idea. For cine, I'm pretty sure it will be at least 2/3".

So, will further break throughs come from better pixel technologies for existing s35 sensors

My guess is that technology will continue to march forward as it has. Theoretically, it's possible for a lot more improvement to read noise ("shadow" noise or "electronic" noise) and dynamic range. I don't see any indication that improvements in those areas will stop soon. However, photon shot noise (which affects all parts of the image, including shadows) will not improve much, if at all, because current tech is already so close to the theoretical limit.

...or will the sensor be swapped out in favour of the larger FF35 sensor?

One good reason for larger sensors is less photon shot noise, particularly now that it's not possible to improve that with existing sensor sizes.

The still camera industry with their FF35 sensors is going full throttle towards creating quality cine capable cameras, thus putting pressure on the lower end cine world.

I wouldn't put it that way. More like "still camera industry added digicam-style video with junky artifacts to their cameras and accidentally started a firestorm in the lower end cine world."

However, an obvious technology obstacle is the lack of cine-grade FF35 lenses.

Agreed.

Another obstacle has been narrow depth of field. Faster FF35 sensors enabling smaller f-stops and longer lenses means focusing with FF35 sensors will be very similar to s35 cameras.

That sounds right to me! I wont go into the big explanation why larger sensors have the same brightness and noise, even when used at narrower f-numbers than smaller sensors.

Can you quantify that [massive difference in quality]?

I've only seen one other person, James Derek Holloway, mention one of the big reasons: *contrast*. FF35 lenses, when set to the same DOF as S35 lenses and displayed at the same size, generally have more contrast.

Others already mentioned other advantages: when used in ample light, FF35 has less noise and more dynamic range for the same DOF. FF35 in low light at the *same* DOF, the noise and dynamic range is only the same. FF35 in low light with thinner DOF has less noise and more dynamic range again.

But Tom... ISO, DR and resolution.. are not directly related to FF35. Those are properties of the camera / sensor and not the format or size of the imaging area.

You make a good point. Tom is only correct when one makes the very reasonable assumption "all other things being equal" (AOTBE). When that is true, then noise, DR, and resolution are directly related to the format size. But the practical reality is that things are often not equal, and small format with good technology, compression, software, etc. can beat a large format in some or all conditions (e.g. D3X beats MFDB in low light).

But what happens is, you begin to hit a wall at the APS-C/S35 sensor size, in terms of your ability to pack in large numbers of photosites while maintaining great DR and ISO/ASA ratings. For example, Canon really hit a brick wall at 15mp with their 50D. You just can't keep cramming more and more pixels onto a relatively small sensor like S35. At some point, you run out of real estate.

For the benefit of the reader, I would point out that I am not in agreement with Tom on this one. The pixel size of the 50D has nothing to do with why it has lower quality (e.g. less DR, more noise) than the 5D2. The reason is that it has a smaller sensor. Even if it has *larger* pixels than the 5D2 (e.g. a 2 MP 50D), it will still have worse noise and worse dynamic range. I went into more detail here: Myth busted: small pixels have worse performance than large pixels.

Scott, ISO 500 is for girly men! We should be talking about ISO 1000 or 1600!

Pfft. 1600 might cut it for Californians, but up here we need ISO 6400.

With an FF35 setup you can always shoot the S35 formats cropped when you need the extra DOF or to match a production work flow.

That's true, but I would add that the optimal solution will probably be to stop down instead of crop. Stopping down for the same DOF will result in the same noise: the only thing that might be worse is compression artifacts, but I'm guessing they will not be bad enough to outweigh the increase in contrast you get from using the full frame stopped down.

At which point the notion of quadrupling the amount of data you have to deal with (and absorbing all of the costs that are associated with that) seems pretty unnecessary.

Good point. Although it will probably possible to get the same quality and file size as lower resolutions just by using higher compression, that does not result in an advantage.

I'm guessing the hope is that the larger sensors will afford enough additional dynamic range to win over the majority of film holdouts.

FF35 has add one and a third stop more than S35, AOTBE.

 

[M Most]

Of course, it's not a matter of if 4K projection becomes the new gold standard for cinema. We know for sure now -- with news from AMC and Cinemark -- that it will be, and soon.

Define "soon." I see this transition taking at least 3 or 4 years, given the economic conditions and the fact that 4K DLP won't even exist until late next year. Film projection is the dominant theatrical reality for some time to come, no matter how much some of us may want it not to be.

To me, there is no doubt, no second guessing, no question whatsoever -- Epic FF35 is the future.

There is always doubt and questions when you're trying to predict the future.

 

[Antoine Baumann]

Is the futur FF35? Well which futur? Theatre realased motion picture? Seems to me, the vast majority of those is still shooting film.
FF35 will be the futur, when digital cinema has really taken over film. Not so soon IMHO
As for quality reason, well I think vision3 has nothing to worry against s35 digital camera and I don't think ff35 will change that.


cheers,
antoine.

 

[Michael Brennan]

Here doggy doggy

Here doggy doggy

Sometimes it is the tail that wags the dog.

The evolution of the physical cinema screen sizes and aspect ratios have been based based on numerous factors, not all of them actually driven by new camera technology.


I wouldn't discount that the concept of small 200 seat cinemas, with a floor to ceiling 4x3 aspect ratio screen, to create an immersive IMAX like experience.

Perhaps the tipping point will be LCD type panels that can be seamlessly built to create say a 20ft x 15ft 6k screen. It is cheaper to build a multiplex cinema complex without projection booths!

The precedent of radically altering the layout of cinema buildings has been made when cinemas went from large single screens to multiscreens.

The future choice of format may well be influenced by factors other than the DP's taste of DoF and resolution.

Most of the difficulties of actually shooting on 65mm (noisy, heavy cameras with short endurance) are solved by shooting digitally.

Perhaps all it needs is a high res display solution that can be scaled to at least 20ft x 15ft?

Mike Brennan

 

[Michael Hastings]

Seems to me that the point people are missing here is the difference between the production format and the output format.

I think 3 factors push toward keeping the APS-C/S35 sensor size as the normal standard.

1) Focusing for MOTION pictures. I don't shoot a lot of feature type material but my understanding is that a) shallow dof for S35 already makes accurate focusing difficult enough, especially as camera or actors are moving. With Still photos a) you tend to shoot several frames and you have a good probability of nailing the focus and nobody will ever see the poorly focused shot b autofocus helps A LOT with stills, increasing the odds of getting those well focused shots and c) FLASH! - you easily have the ability to provide plenty of light since you only need a burst for that 1000th of a second or so. Movies need high levels of continuous lighting so quite often you are shooting at or near wide open - exacerbating the focusing problems.

2) As CMOS and new sensor technology improves (and billions are being spent on sensors every year and it has only been in the past 5 or 6 years that most people have embraced digital - so we are just midway in the development cycle). Even the pixels on APS-C SLRs seem huge compared to the minuscule pixels on most of the consumer cameras - that do a pretty amazing job - so we can expect to get both dramatically increased resolution and dynamic range out of APS-C/S35 sized sensors. So the technical need for a larger sensor will become less.

3) Lenses: those of us that can live with using still camera lenses on Birger smart or other mounts have no real problem whatever size sensor the manufacturers put out. But for true cinema style lenses it is doubtful whether there will be much selection for larger format lenses with large and constant apertures; geared precision focus and iris mechanisms; standard fronts, etc.

We already have T1.3 Master Primes, Ultraprimes, Cookes, Optimo zooms, etc. and RED has shown how much you can do with modern tech and somewhat higher volumes and the REDONE has already encouraged other manufacturers to do similar things. But this made sense because the RED created a backlog of PL lens need that is rapidly being filled. But all of those lenses are S35 (the Zeiss compact primes don't really fit the spec) and I doubt whether there will be sufficient demand for the larger formats to foster future development (Imagine how much larger a t1.3 masterprime FF35 would have to be - and cost) SINCE the FF35 sensors will crop to S35 with more than sufficient resolutions - and which will increase every few years anyway.

Taking my production/output point one step further: I do believe there is a significant place for ff35 or more likely 645, in Imax type shooting. There is already a pretty large network of Imax theatres in existence. In the past SHOOTING true superhigh res IMAX type motion pictures has been a huge hassle with the extremely large cameras/film needs etc. But uprezzing HD video doesn't really cut it.

Now the Epic 645 gives us an answer: it really isn't much more difficult to shoot nature, aerial, underwater, etc. with the Epic 645 than it will be with a scarlet. Take for example my underwater work - the brains are almost identical in size/shape so by using a Superwide 28 or 35mm lens (I can get a contax zeiss 35mm for about $1700) on the 645 I can get 9k underwater sharks, shipwrecks, etc. with no more difficulty than if I shot it with Scarlet (just need to carry a 2 or 3 more hard drives). Yet I get 4 times the res of the REDONE or sixteen? times res than if I shot it with a 1080P camera. Now we are talking about something really making good use of a transfer to Imax film.

 

[david farland]

FF35 has add one and a third stop more than S35, AOTBE.

Dan, Are your predicting this on the greater volume (or area) of the pixel well is ~20% higher? Red are saying it'll be 3 stops higher than R1 (2 greater than Epic s35).
Hopefully this produces a FF35 sensor that is >1 stop faster i.e 320 -> 450/500 asa.

Which also answers some of Michael H's dilemnas. First one being DOF for FF35.
Depending on your Coc for s35/FF35 lenses, say the FF35 sensor is 1 stop faster.
So the equations you're likely to have are:
S35 (Coc=.017) x 50mm x f4 x 4 metres = .88m DOF
FF35 (Coc=.025) x 75mm x f5.6 x 4 metres = .80 DOF

Next is sensor technology. A larger sensor area with bigger pixels will usually give you a better image quality than a smaller sensor with smaller pixels.
So whatever way you look at it, if we can expect to get both dramatically increased resolution (least important) and dynamic range out of an s35 sensor we can expect the same or better performance from a FF35 sensor.

Next is lenses. They a bit different. If you look at Red's, Uniq Optics or Zeiss compact primes (ZF) lenses, you'll find they are close or exceed the required 43mm optical circle (or 41mm for 16x9) for a FF35 sensor.
Where they can't cover and most need a redesign is ~35mm and under. Remember a FF35 x 35mm lenses has a similiar field of view as a S35 x 25mm lens. And a 25mm is equivalent to 17mm.
So yes there will be a lack of wider cine FF35 lenses...well except for Red's electronic FF35 lenses, still glass and Zeiss compact primes.
I'm expecting like when Red released the RedOne and the lens manufacturers responded with flooding NAB with s35 glass, they'll respond similiarly if there is a demand for FF35 cine lenses.

Dave

 

[Michael Hastings]

David:

I agree with most of what you are saying, except maybe to explain a little further what I meant with lenses. I think the influx of a bunch of new cine lenses was a one time thing where RED effectively doubled the worldwide universe of PL mount cameras and basically all of these companies could see that there all of a sudden was a market for these cine style lenses where they already had the various designs i.e. the S35 Master Primes, UltraPrimes cookes and the other lesser brands and it was just a matter of updating them and making in a sufficient quantity to bring the price down.

As I said the compact primes don't really fit the full definition of what we think of true cine style lenses, in that they don't have consistent full open Tstop, again because they are really SLR still lens optics (originally the Contax Yashica Zeiss updated to the Zeiss ZF and then modified to the compact primes)

Even with modern lens technology, making a super sharp, minimal breathing, small t-stop, multiblade iris, smooth operating cine lens THAT WOULD COVER FULL FRAME is a much more difficult challenge than bringing out some new lower cost S35 lenses. Imagine how large and heavy a 28mm (18mm equivalent) Master Prime that covers full frame will have to be to achieve a t1.3 or even a t2 if you thought you could give up a stop.

The most serious players that would have the budgetfor these lenses are very likely to be quite happy to shoot S35 even if they do it on a cropped FF35 Monstro sensor. When their isn't a clear market need for them - the level of uncertainty will be hard to justify the investment required to make a series of entirely new design lenses. In other words it was quite obvious a year ago that there was going to be a substantial market for new S35 lenses so several groups stepped up. It won't be so obvious for FF35 and with the current and near future economic climate it is unlikely we will see that kind of investment for quite a while if at all. Also, the availability of things like smart EOS mounts (which are just now really coming online) for numerous existing FF35 lenses will create even more uncertainty for a potential developer.

As far as sensitivity, I don't think the availability of really good 500ASA film has really altered the desire for DPs to use Master Prime t1.3 lenses. I think high end shoots/DPs will perceive that t1.3 to t1.8 is the sweet spot for lenses and will demand that in their optics - and use that extra sensitivity to reduce the lighting budget.

BTW This is all rambling speculation that I'm pulling out of my a.. and could be way.:biggrin:

 

[Daniel Browning]

Dan, Are your predicting this on the greater volume (or area) of the pixel well is ~20% higher?

Not at all. It is based on the fact that FF35 has more than twice as much light falling on it when used at the same T/stop, which will move every exposure zone 1.3 stops to the right of the noise floor (i.e. if you used to clip blacks at -9 EV due to noise, now you'll do it at -10.3 EV). I would make the same prediction even if the FF35 pixels were planned to have 4 times smaller area.

A larger sensor area with bigger pixels will usually give you a better image quality than a smaller sensor with smaller pixels.

Actually, that's a myth. Larger sensors are better, but it has nothing to do with pixel size. They are better even when their pixels are smaller. A digicam-sized sensor (5.6x4.15mm) sensor with super-large pixels (0.048 MP) will not have superior performance to a 56x41.5mm sensor with super-tiny pixels (48 MP). Judging sensors by per-pixel performance numbers is just as misleading as judging engine performance by per-cylinder numbers. A tiny engine with a single cylinder will not beat a large engine with small cylinders. What matters is the total horsepower of the engine and the total light gathering power of the sensor.

Technology is vital, of course, and the ingenuity of sensor designers is exactly why performance has scaled so well with pixel size all these years.

So whatever way you look at it, if we can expect to get both dramatically increased resolution (least important) and dynamic range out of an s35 sensor we can expect the same or better performance from a FF35 sensor.

Agreed.

Imagine how large and heavy a 28mm (18mm equivalent) Master Prime that covers full frame will have to be to achieve a t1.3 or even a t2 if you thought you could give up a stop.

FF35 T/1.3 MP would be a beast for sure. The weight of a lens is proportional to the cube of the focal length when f-number is constant (and AOTBE), and since FF35 needs focal lengths around 1.6X longer, that's a lot of weight. But when you give up 1.3 stops, is largly neutralizes the effect of longer focal length so that weight stays the same. (Another way to look at this is to realize that iris diameter remains the same, which is the first order effect relating to weight.) At 1.3 stops slower, the FF35 lens will It will also be easier and less expensive to make, since the spatial frequency is so much lower. As far as optics are concerned, it's easier/better at slower T/stops and larger image circles than fast T/stops and small image circles.

 

[Steve Sanacore]

You realise that one stop is double the light don't you?

A stop and a half is literally multiplying the lighting budget by three times.

Or-

A stop and a half increase in ISO instead of more lights would work too.

 

[Joseph Ward]

This is still being said, how long are we going to discuss this...:sarcasm: That being said, Yes! FF35 is the future, at least for me!:ihih:

 

[david farland]

Actually, that's a myth. Larger sensors are better, but it has nothing to do with pixel size. They are better even when their pixels are smaller. A digicam-sized sensor (5.6x4.15mm) sensor with super-large pixels (0.048 MP) will not have superior performance to a 56x41.5mm sensor with super-tiny pixels (48 MP). Judging sensors by per-pixel performance numbers is just as misleading as judging engine performance by per-cylinder numbers. A tiny engine with a single cylinder will not beat a large engine with small cylinders. What matters is the total horsepower of the engine and the total light gathering power of the sensor.

Not sure how engine cylinders come into it but the larger pixels in the Epic FF35 system should (hasn't actually been built!) have a higher dynamic range....and to me that's better.
D