RED vs. 35mm shoot-out

[Thomas James]

It's is not that people are adverse to grain but rather fineness of the grain is what is considered. For example 16mm film has plenty of resolution and color fidelity to be considered equivalent to a 1080p or 2k high definition format however the course grain structure of 16mm film disqualifies it from BBC recognition as a high definition format. However 35mm film is a perfectly acceptable aquisition format for high definition broadcasting according to BBC rules because it has a much finer grain structure than 16mm film.

 

In the near future when broadcasters set minimum standards for 4k broadcasting it is very likely that 35mm film cameras will receive the very same intense scrutiny that 16mm film cameras received. These broadcasters will either demand 65mm film aquisition or at the very least specify extremely fine grain 35mm film stocks. However digital cameras will not get a free ride. The broadcasters may insist on using only true 4k cameras which may mean that a 5k camera like the Epic or a three chip 4k camera or a single 4k 65mm chip.

[Scott Fritzshall]

It took like 60 years to go from NTSC to ATSC. The "near future" you talk about is probably at least 25 years away, if it ever even comes. The current transition in broadcast is painful enough, absolutely no one will want to go through it again for 4k broadcasting, especially when most people have small TV sets where the difference is going to be minimal.

[John Brawley]

I think the problem with 35mm film is that it is too grainy to be considered ultra high definition.

 

Who says ?

 

jb

[John Brawley]

For example 16mm film has plenty of resolution and color fidelity to be considered equivalent to a 1080p or 2k high definition format however the course grain structure of 16mm film disqualifies it from BBC recognition as a high definition format. However 35mm film is a perfectly acceptable aquisition format for high definition broadcasting according to BBC rules because it has a much finer grain structure than 16mm film.

 

 

Which BBC rules are you talking about ?

 

I've just finished watching the second series of the excellent and emmy award winning, Life On Mars. In the EPK, there's plenty of SR3's being used to photograph this show.

 

Where is it actually stated that the BBC don't take Super 16 ? Which BBC "rules" ?

 

http://www.bbc.co.uk/commissioning/delivery/

 

There are some departments within the BEEB that don't like 16mm because of the way the compress the crap out of their broadcast signals, and in particular to do with their Discovery co-pros.

 

By this logic, the BBC would rather air HDV material over Super 16.

 

jb

[DJ Joofa]

Because that is what many of the RED fans kept proclaiming. I remember one going "It can do 10,000 ISO" over and over again.

 

I think Red's marketing strategy is hurting them also as it is not taking advantage of Red's inherent potential to pose itself as a company that seriously understands issues in developing a complex system such as a video camera. As somebody, who has simultaneous academic and industrial experience, where we have developed systems that includes a lot of Red's functionality, and even go much farther than it, I guess I am well placed to make an evaluation. I am impressed at the achievement they have accomplished in a short span of time. However, I think the camera is still not there yet, but with the right effort it can be pulled off. I do however feel that Red should reduce the level of hype/mystery surrounding its products, as IMHO it is not in their best interest in the long run.

[John Sprung]

It took like 60 years to go from NTSC to ATSC. The "near future" you talk about is probably at least 25 years away,

The life span of NTSC, if all goes according to plan, will be 68+ years from the adoption of NTSC I in June of 1940 to NTSC sunset on Feb 17, 2009. For NTSC II color, it'll be 55 years from adoption in December of 1953.

 

But the transition didn't take that long. The first work on HD was done in Japan in 1971, and the ATSC standard was adopted in December of 1996. So, the actual transition will be just over 12 years.

 

Back when terrestrial broadcasting was the only way to get shows into the home, that kind of monolithic standard was a necessity. Now that we have packaged media, internet, satellite, cable, etc. it's possible for different systems and resolution levels to coexist. It's only because of the scarcity of spectrum that we really need to shut down NTSC to make room for ATSC and other new uses.

 

We may see high end 4K home theaters with ATSC upconverted, and native 4K from disc rentals, etc.

 

 

 

-- J.S.

[Jim Exton]

I really don't understand why people are so averse to grain. I for one love grain, it gives the images a lovely texture that digital lacks. If you want to avoid grain, just shoot anamorphic. Really, modern filmstocks are so low in grain anyway that it is not really an issue even if you shoot spherical.

 

I agree. Grain is not a four letter word.

[Cesar Rubio]

It took like 60 years to go from NTSC to ATSC. The "near future" you talk about is probably at least 25 years away, if it ever even comes. The current transition in broadcast is painful enough, absolutely no one will want to go through it again for 4k broadcasting, especially when most people have small TV sets where the difference is going to be minimal.

 

I totally agree with this statement. I think that instead of more resolution, TV as well as Cinema will go for 3-D in the future.

 

And I also think that 720p will be the norm for 3D-TV broadcasts. Sending 2 streams of 720p is comparable in size to 1 stream of 1080p.

 

I don't really think that anaglyph 3-D transmissions in TV will be the future. Probably it will be something like this system:

 

 

http://edimensional.com/product_info.php?c...products_id=127

 

 

 

Or a glassless 3-D monitor system viewing, which in any case requires double the bandwidth of a single stream.

 

I receive my TV signal over "Dish" Network, and I hate the poor quality so much because of the already heavily compression...and its SD resolution!

 

And like Scott says, for "small" TV viewing who needs 4K?

 

Even for 3-D Cinema, 1080p or 2K is enough.

 

A 35mm Cinema film frame has a comparable resolution of a 1080p or 2K digital camera. 4K is more like 65mm like some have already suggested.

 

Thanks,

Cesar Rubio.

[John Sprung]

...., which in any case requires double the bandwidth of a single stream.

Not necessarily. The right and left images are very similar, often more so than consecutive flat frames. With digital compression, all you'd really need to do is send one stream plus a much smaller channel containing the differences. The main stream could be viewed as a flat picture. 1080p 3D should be very doable that way. Likewise, 4K 3D wouldn't take a lot more bandwidth than flat 4K.

 

What is most important to grasp is that in the digital world, all this stuff can coexist. We just have to get the standards right, so every data stream can tell all the boxes what's what.

 

 

 

-- J.S.

[Cesar Rubio]

Not necessarily. The right and left images are very similar, often more so than consecutive flat frames. With digital compression, all you'd really need to do is send one stream plus a much smaller channel containing the differences. The main stream could be viewed as a flat picture. 1080p 3D should be very doable that way. Likewise, 4K 3D wouldn't take a lot more bandwidth than flat 4K.

 

What is most important to grasp is that in the digital world, all this stuff can coexist. We just have to get the standards right, so every data stream can tell all the boxes what's what.

-- J.S.

 

John:

 

This is the first time that I heard such a thing, in theory sounds great, but in real life and real 3-D, you need 2 totally different streams (Left & Right) to get a good 3-D effect.

 

I think that next 3-D TV generation will be 720p. 3-D digital Cinema will be dominated in most of the world by 1080p and 2K.

 

4K is more like IMAX venues...very limited...unless you down-convert to 2K or 1080p. I prefer to shoot in the latest in the first place and save me the headaches associated with the 4K workflow.

 

Besides that, 4K cameras requires at least 35mm size sensors, and that size is not well suited to get a good DOF. In 3-D we want everything in focus...the opposite of the current sought out of focus backgrounds in flat photography.

 

CR.

[A. Whitehouse]

Ultra high definition and 3d broadcast? Is this Back to the Future 2? Film will definetly be to grainy for my romps on the Holodeck with Geordie and Data...

 

Skin tones looked very odd, even after correction, to me. If you were to run the test again I would be interested to see more results shot under daylight. I imagine a better Telecine would eke far more out of the neg than is there but thanks for sharing, I found it interesting.

[A. Whitehouse]

Oh and as far as s16mm being to grainy to be considered 2K, while a 500 iso stock might be slower stocks are certainly not to grainy. In fact 7217 can be remarkably clean not to mention '01. Even '18 (I haven't used '19 yet) can look virtually grain less in HD. Its just that S16mm generally gets chosen for its flexibility and texture and its not often that an attempt is made to produce as clean an image as possible because the film makers would probably choose a different format if thats what they're looking for.

The BBCs decision has been discussed here before and it seems pretty short sighted to me.

[Brian Drysdale]

I know of one BBC Sci Fi series, where given the choice between Super 16 or Varicam for slow motion the effects house said Super 16, saying it's got a larger colour space and if required they can always de-grain and register it.

[Cesar Rubio]

This article describes the differences of 720p vs 1080i in TV transmissions clearly:

 

http://alvyray.com/DigitalTV/default.htm

 

 

Charles:

 

These tests show more than a thousand words.

 

Have you make tests of film vs CCD sensor cameras?

 

CCD have better saturation off the bat than CMOS, please see these 2 extracted frames from a 1" CCD Kodak sensor video recordings at 24 fps: (800x1920 pxs)

 

 

http://dr-3dcameraco.com/images/David-Red_Hat.tif

 

 

http://dr-3dcameraco.com/Tiffs/Corona_High_f-1.2.tif

 

 

I did not applied any post processing at all. The video images came straight out from the camera like that. I just extracted the frames to Tiff's.

 

The videos were recorded with a Pike F-210c camera using the CineForm Raw codec.

 

There are still some "bugs" that need to be worked out to get the best quality from the camera using CineForm Raw.

 

Probably in a month or less they will be finished.

 

 

It would be interesting to make a 35mm film vc CCD comparison don't you think?

 

We could do also a totally Raw uncompressed test with the camera. So we could have 35mm film (also 16mm film if somebody provides a camera), CCD uncompressed Raw and CCD CineForm Raw of the same images and settings.

 

I could provide one of my Pike F-210c cameras for the testings, and even probably go there to LA do it my-self...or you could come to WI. We can arrange something.

 

I offered a similar challenge to a one RED camera owner here in Milwaukee, but he refused (or ignored my offer, which is the same).

 

I already know that film will beat the CCD in dynamic range and saturation....but probably not as much as in the CMOS sensor test (we have to see this first to be sure, hence the testing).

 

But then we could do a short 3-D movie test with 2 film cameras and my 3-D MVC (Machine Vision Cameras) system.

 

I can assure you that since my MVC 3-D system is so small and the sync of both cameras is perfect all the time, you might consider "giving up" film in favor of digital acquisition..at least for 3-D.

 

What do you think?

 

Thanks,

Cesar Rubio.

Edited May 7, 2008 by Cesar Rubio

[Cesar Rubio]

I forgot to mention that a 720/60p bandwidth (like Fox has chosen for digital broadcast), is perfect to send two 720/30 or 720/24 streams for 3-D viewing.

 

CR.

[John Sprung]

This is the first time that I heard such a thing, in theory sounds great, but in real life and real 3-D, you need 2 totally different streams (Left & Right) to get a good 3-D effect.

Well, not so very different. What you need are two images taken at the same time from a slight distance apart, like the view from our two eyes. Do you have a 3D pair of still frames handy that we could experiment with? If we were to put both of them into .BMP files, and just go through pixel by pixel simply subtracting the right from the left, we'd find that we get a whole lot of zeroes, and where there are gradients in the actual images, the difference file would have small numbers that are much nearer to being constants. That kind of difference file compresses very well and very small. If we were to add a constant to all the numbers in the difference file, to get them non-negative and near the middle of the range, we could look at that as an image. It should be a sort of bland gray bas relief.

 

Consider a thought experiment: Instead of 3D, suppose we do a dolly shot across a static scene, moving at a speed that puts each frame one interocular distance from the ones before and after it. Put that shot through MPEG compression using I, P, and B frames. MPEG is using the strong similarities between frames shot a short interval of time apart. We could just as easily use the strong similarities between frames shot at the same time, but a short horizontal distance apart.

 

Bottom line, 3D should be very compression friendly. All we need is for the guys who really create the compression algorithms to work on it.

 

 

 

 

-- J.S.

[Michael Lehnert]

I think the problem with 35mm film is that it is too grainy to be considered ultra high definition. I saw the HD-DVD version of the 1966 movie Grand Prix which was originally shot on 65mm film at it looked great because it was sharp clear and clean. Of course Red does not have the resolution of 65mm film but it matches the cleaness of the 65mm film format. With the introduction of the Epic it will be interesting to see if digital will be a real alternative to 65mm film production.

 

Goodness Gracious Me!?!

 

I obviously didn't prepare myself mentally enough for my "catching-up-with-threads-I-missed-while-being-away-for-to-long" session, as after my reading through this thread here, I feel like Han Solo getting out of carbonite: "I'm out of it for a little while and everbody gets delusions of grandeur".

 

Is that statement for serious or just a troll's joke?!? I mean... seriously now, folks...

[Cesar Rubio]

Well, not so very different. What you need are two images taken at the same time from a slight distance apart, like the view from our two eyes. Do you have a 3D pair of still frames handy that we could experiment with? If we were to put both of them into .BMP files, and just go through pixel by pixel simply subtracting the right from the left, we'd find that we get a whole lot of zeroes, and where there are gradients in the actual images, the difference file would have small numbers that are much nearer to being constants. That kind of difference file compresses very well and very small. If we were to add a constant to all the numbers in the difference file, to get them non-negative and near the middle of the range, we could look at that as an image. It should be a sort of bland gray bas relief.

 

Consider a thought experiment: Instead of 3D, suppose we do a dolly shot across a static scene, moving at a speed that puts each frame one interocular distance from the ones before and after it. Put that shot through MPEG compression using I, P, and B frames. MPEG is using the strong similarities between frames shot a short interval of time apart. We could just as easily use the strong similarities between frames shot at the same time, but a short horizontal distance apart.

 

Bottom line, 3D should be very compression friendly. All we need is for the guys who really create the compression algorithms to work on it.

 

-- J.S.

 

John:

 

I understand what you are saying, the problem is that every of our eyes needs to see a full image at a time.

 

When the right image is displayed our left eye doest' see a thing, and vice-versa, when the left image is displayed the right eye is "blind". All of this happens so fast that we are not aware of this fact. The brain "fuses" the 2 images to make "one" out of it.

 

With the compression system that you are describing (If I am understanding it correctly), one eye would only see part of one image (the one that is different from the other correct?).

 

CR.

[Michael Lehnert]

Actually, I am not even going to wait for an explanatory reply here from that member as the convoluted melée of post #51 on what went between the Beep and the BSC disqualifies anything that could possible be brought up in his defence.

 

Just to make a point about the ludicrousness of some of the content that is shifted here:

 

If you, Thomas, were to speak about music rather than cinematography in your post quoted by me just above, your statement would read like this:

 

"I think the problem with contemporary-instrument orchestras is that they sound too abrassive to be considered of symphonic hall quality. I listened to Bach's St Matthew Passion recorded by Nikolaus Harnoncourt with baroque instruments on SACD and it sounded so great because it had volume and loudness. Of course the Kurzweil K250 does not have the scope of a full orchestra but it matches the synchronous polyphony of a symphony orchestra. With the introduction of MIDI, it will be interesting to hear if digital will be a real alternative to symphony orchestra performances."

 

I think you got the point. Good night, and good luck!

[DJ Joofa]

Ultra-high-definition? There is some fixation with increasing the resolution size without considering its full implication on color fidelity.

 

As colored objects are decreased in size, four things are found to happen in succession. First, blues become indistinguishable from grays of equivalent brightness and, second, yellows become indistinguishable from grays. In the size range where this happens, browns are confused (in hue but not in brightness) with crimsons, and blues with greens, but reds remain clearly distinct from blue-greens. On the whole, colors with pronounced blue lose blueness, while colors lacking in blue gain blueness; all become less saturated. Third, with still further decrease in size, reds merge with grays of equivalent brightness and, finally, blue-greens also become indistinguishable from gray. For exceedingly small objects, normal visual sensations are devoid of all color connotation, and only perception of brightness remains.

 

Small patches cut from large colored sheets are not as well matched visually by the original sheets as they are by sheets of somewhat differently colored material. Additionally, the indication is that any color, in a small enough patch well centered in the field of vision, can be matched by mixing only two, and not three, "primary" colored lights. I.e., the tendency of the chromaticity diagram to degenerate toward a single line for these small patches and indicates that the two primaries mixed to match the color of a tiny object may be chosen as a barely orange-red and a greenish-blue.

 

Refs.:

 

George H. Brown, NTSC Report.

 

W. E. K. Middleton and M. C. Holmes, "The apparent colors of surfaces of small subtense-a preliminary report," Journal of Optical Society of America., vol. 39, 582-592; July, 1949.

[Thomas James]

I think that if anyone has ever gotten a chance to see a good clean transfer from 65mm film to 1080p Blu-Ray disc I think that the only logical conclusion is that all this talk about the virtues of film grain is simply to ignore the fact that film grain is simply an undesirable artifact that damages picture quality. The whole purpose of high definition is to render fine detail and too much film grain simply masks and makes this fine detail that much harder to see. It may very well be that 1080p with 65mm film origination could actually outperform 35mm film origination "uprezzed" or even natively scanned at 4k and so to really take advantage of the potential of the 4k format broadcasters may indeed insist on 65mm film origination or its equivalent.

 

The problem is that there are a lot of people that either do not remember or have been deprived in this era of the multiplex theatre of actually seeing a real 70mm film production which was very common during its heyday in the 1960's I remember my parents taking me too see 70mm film productions when they first came out like the Sound of Music, Grand Prix and 2001 a Space Oddysee. So while 70mm is not the norm today it was very much the norm back in my generation and today with all this emphasis on high definition is it not unreasonable to expect large film formats to make a comeback? And in a sense these large 70mm film formats are already making a comeback for the home theatre as they are being released in the 1080p Blu-Ray disc format which allows todays consumers to get a glimpse of the superior quality of 65mm film origination.

[Freya Black]

John:

 

I understand what you are saying, the problem is that every of our eyes needs to see a full image at a time.

 

When the right image is displayed our left eye doest' see a thing, and vice-versa, when the left image is displayed the right eye is "blind". All of this happens so fast that we are not aware of this fact. The brain "fuses" the 2 images to make "one" out of it.

 

With the compression system that you are describing (If I am understanding it correctly), one eye would only see part of one image (the one that is different from the other correct?).

 

CR.

 

No, what John means is that they would be stored digitally that way. When the video player plays them back it would restore them to 2 full frames.

 

It's like the way that DVD's work. They show each of the 25 frames per second in full, but on the disk they are not stored as full frames. If bits in one frame are the same as bits in the next frame it only stores the different bits of the frame. Then when it comes to restore the image it says use the bits from the last frame and use the new bits and create a new frame that just looks like a normal frame. It's more complicated than that but hopefully you get the idea.

 

You could do the same with 3d only more so because the two images are likely to be very similar all the time, as each eye is going to be seeing the same thing only from slightly different positions. 3d would thus be VERY compressible.

 

love

 

Freya

[John Sprung]

With the compression system that you are describing (If I am understanding it correctly), one eye would only see part of one image (the one that is different from the other correct?).

Freya has it right. The idea is to use this kind of compression for storage and transmission only, and re-create both left and right full images for display.

 

Starting with a full Left and Right stream, we could make a Left minus Right stream, and compress two channels: R, which would compress just like an ordinary flat show, and (L-R), which would compress very much better because it's mostly zeros most of the time.

 

On the other end, decompress R and send it to the right eye. Take another copy of R, add the decompressed L-R to it, and send that to the left eye. The math is easy:

 

R + (L-R) = L

 

That's quite straightforward, and could be done with any off the shelf codec. It has the advantage that non-3D equipment could show you the R only as a flat picture. Get some smart compression folks working on it, and 3D could probably be compressed even better than that.

 

 

 

-- J.S.

[Michael Lehnert]

I think that if anyone has ever gotten a chance to see a good clean transfer from 65mm film to 1080p Blu-Ray disc I think that the only logical conclusion is that all this talk about the virtues of film grain is simply to ignore the fact that film grain is simply an undesirable artifact that damages picture quality.

[...]

And in a sense these large 70mm film formats are already making a comeback for the home theatre as they are being released in the 1080p Blu-Ray disc format which allows todays consumers to get a glimpse of the superior quality of 65mm film origination.

 

:blink:

 

Gosh, I don't know where to start ... I think I kinda see what your angle is, but it's so convoluted and drizzled with impossible correlations between consumer media products and professional acquisition tools, I really think I can't invest time to slash that Gordian knot. Sorry for the inconvenience. I will be off-forum 'til Monday and don't want unpleasant arguments looming over my week-end. So just enjoy Blu-ray, the killer of film grain and real reason why 65mm will become TV's most preferred acquisition medium.

 

Cheers, -Michael

 

 

 

P.S.: Sorry, John, Freya, for interjecting into your interesting debate with Cesar. Won't disturbe any further :D ! Have a nice week-end!

[Cesar Rubio]

Freya has it right. The idea is to use this kind of compression for storage and transmission only, and re-create both left and right full images for display.

 

Starting with a full Left and Right stream, we could make a Left minus Right stream, and compress two channels: R, which would compress just like an ordinary flat show, and (L-R), which would compress very much better because it's mostly zeros most of the time.

 

On the other end, decompress R and send it to the right eye. Take another copy of R, add the decompressed L-R to it, and send that to the left eye. The math is easy:

 

R + (L-R) = L

 

That's quite straightforward, and could be done with any off the shelf codec. It has the advantage that non-3D equipment could show you the R only as a flat picture. Get some smart compression folks working on it, and 3D could probably be compressed even better than that.

 

-- J.S.

 

John and Freya:

 

Hopefully you are right and that can be achievable (I need to see it first to believe that it can be done).

 

In any way, with 720/60p, 3-D TV broadcasting is doable and ready for main stream anyhow.

 

 

Are you still around Charles?

 

Or it is "strike two" for my offering of a test shoot-out with my Pike F-210c cameras?

 

Silence is the same as a negative....

 

Thanks,

Cesar Rubio.