ARRI: 16mm film is 3K, 35mm film is 6K...

[Dominique Ventzke]

I can guarantee you one thing: cost was not the reason for choosing a 4K scan over a 6K for The Dark Knight.

Now, I have no idea if there might be other, DMR-related reasons for scanning 35mm at 4K rather than 6K. Perhaps it has something to do with the de-graining process? I don't know.

In that American Cinematographer article you linked David Keighley said:

If we’d had time to scan the original negative at 6K, we could have produced even higher quality. The information is on the negative — 35mm film captures the equivalent of 6K and a color bit depth of 14 bits plus.

 

[Roberto Lequeux]

Thank you Peter... you totally made my day.

What is MTF? Seriously, please teach me cause I don't have a clue. I am in over my head on this thread. My head is spinning with all this awesome heavy technical knowledge. But I humbly learn more every day.

I doubt it was directed at me, I am not sure you tried to say that either, but just for the sake of being clear I didn't try to bash Arri in any way... I mean, they are Arri you know. But I just can't help but wonder if, at least partially, the white paper was released to try to help Arri in a way till they can catch up their sensor program to Red's.


Also... isn't Digital on the verge or already matching the DNR of film? If not yet it will certainty reach that point and continue onwards. I mean... technology won't stop... so unless there is more to be squeezed out of film... someone please tell me more about this. Thank you.

 

[Dan Hudgins]

More dynamic range...

More dynamic range...

Also... isn't Digital on the verge or already matching the DNR of film? If not yet it will certainty reach that point and continue onwards. I mean... technology won't stop... so unless there is more to be squeezed out of film... someone please tell me more about this. Thank you.

You can increase both the resolution and dynamic range of film by pull porcessing and diluting the developer (fresh developer). Both of these give more highlight range.

You can increase the dynamic range of color negative film by using a re-processing method that uses two developers, the first developer uses Phenidone A, the second uses CD-3, unlike reversal process you use these steps,

0) Jet backing pre-dip and wash.
1) First Dev, Phenidone A (as the only developing agent)
2) 1st Stop
3) Non acid Fix
4) wash
5) Bleach to convert silver back to silver bromide
6) wash
7) Second Dev, CD-3
8) 2nd Stop
9) 1st Fix
10) wash
11) Bleach
12) wash
13) 2nd Fix
14) final wash

Some color shift and tracking errors will be introduced since all three layers are not tuned for the first developer, but those can be adjusted for with the HDR 16bit TIF from the scanner (use 16bit TIF rather than DPX since DPX do not have enough bits to allow for heavy curve adjustments in post.)

Phenidone A can be used with high con Sound Recording film to get a full scale with more resolution than any other system, with Pan sound recording film (without the latent printing) you could use three strip to get more resoluton that dye coupler film.

You can use Phenidone A twice to get more range, in that case you first develop and bleach fix then re-develop in Phenidone A and bleach fix before you re-develop in CD-3. Or for black and white film you develop fix bleach and re-develop fix.

Another way to get more dynamic range in black and white film is to use Elon for the first developer in reversal process of negative stocks.

To see the resolution improvments well you would need to make 70mm prints, but for the dynamic range improvements a good 35mm made optically should be enough, but a 70mm print can hold more image tones than a 35mm print do to the larger film area for the same screen size. If you make a DI your scanner will need to make multi-HDR exposures to get off the film the range of tones, and your film recorder would need to make multi-HDR exposures to put all the tones back on (to overcome some issues with the D2A and transfer curve of the CRT or modulator).

It will be hard for a single sensor to match the dynamic range of film since film uses local selif limitation, with the lens being the limit on dynamic range. The larger the format the greater the dynamic range in film (when processing is adjusted), but the lens limits the piratical limit, and also if you make the images too flat when projected they will look too low contrast to todays audience since people are used to seeing contrast pushed to the limit and video highlight blowout.

 

[Mat@imageWork]

in fact Roberto, there is many technical aspect in digital cinematography to have the big picture and understand what is really under the hood. I recommend you to listen at the conference from panavision.

MTF is a technical way to describe what sharpness is. Not only resolution of CCD describe that phenomen because depending the lenses you use, the picture could less or more sharp. Each "system" (camera, lenses, projector, etc...) have it's own MTF. A film stock also...Resolution of CCD is also connected to the compression algorythm of the signal in "RGB world", or "YUV world". A Sony F23 have 3CCD on 1920x1080 for each channel with a loss less compression. The quantification of the signal is 14bit. So it's a true HD cam that could get 12 stop of latitude range, witch is better than the red cam. But to get the full quality, you need true HD lenses that can get the frequency at a much higher level than 35mm (true HD lenses have been much more complex to design) Red cam have one CMOS sensor with a lossless compression in Red, but blue and green channel have more compression. If you don't know that but a genesis panavision have a bigger sensor Cmos than even the red, but engineer decided to "record" only a 1920x1080 signal because the overall compression algorythm make only a true uncompressed HD signal to be recorded...

Sorry for this cahotic explication, but I try to stay far from strong technical aspect of ligh&frequency, compression algorythm, etc...Learning everything is a long process, but I hope this thread have given you a small kick off to learn more. The beauty in this business is that we always gonna be student...

 

[Christopher Grant Harvey]

Nice Article on Blade Runners restoration process. Mentions scanning the 65mm neg's at 8K and so forth.

http://paulinerogers.com/files/progers.sb2.authorsguild.net/media/BLADE_RUNNER.pdf

 

[Peter Majtan]

A Sony F23 have 3CCD on 1920x1080 for each channel with a loss less compression. The quantification of the signal is 14bit. So it's a true HD cam that could get 12 stop of latitude range, witch is better than the red cam.

Come again?

 

[Roberto Lequeux]

in fact Roberto, there is many technical aspect in digital cinematography to have the big picture and understand what is really under the hood. I recommend you to listen at the conference from panavision.

MTF is a technical way to describe what sharpness is. Not only resolution of CCD describe that phenomen because depending the lenses you use, the picture could less or more sharp. Each "system" (camera, lenses, projector, etc...) have it's own MTF. A film stock also...Resolution of CCD is also connected to the compression algorythm of the signal in "RGB world", or "YUV world". A Sony F23 have 3CCD on 1920x1080 for each channel with a loss less compression. The quantification of the signal is 14bit. So it's a true HD cam that could get 12 stop of latitude range, witch is better than the red cam. But to get the full quality, you need true HD lenses that can get the frequency at a much higher level than 35mm (true HD lenses have been much more complex to design) Red cam have one CMOS sensor with a lossless compression in Red, but blue and green channel have more compression. If you don't know that but a genesis panavision have a bigger sensor Cmos than even the red, but engineer decided to "record" only a 1920x1080 signal because the overall compression algorythm make only a true uncompressed HD signal to be recorded...

Sorry for this cahotic explication, but I try to stay far from strong technical aspect of ligh&frequency, compression algorythm, etc...Learning everything is a long process, but I hope this thread have given you a small kick off to learn more. The beauty in this business is that we always gonna be student...

It sounds like you are going to have try to defend that "better than Red" part and I am not so sure it will happen. I'll watch and learn from the sidelines.

But thank you very much for the post! It did put a lot of things I know and understand in perspective. MTF is certainly a good term to know! What does it stand for exactly?

 

[Dan Hudgins]

MTF, drawing to show Digital Cinema vs. Film vs. Video

MTF, drawing to show Digital Cinema vs. Film vs. Video

MTF is certainly a good term to know! What does it stand for exactly?

I made a drawing posted here that shows how MTF is measured and the difference between video, Digital Cinema, and Film.

The RED line in the MTF graph is film, it falls off for the most part as the details in the test chart get smaller, you can adjust the developer to make it peak a little, but it has lower contrast in the mid-high frequencies than video. Film has more contrast than video in the very high frequencies. It may have more contrast at the tip top frequencies than Digital Cinema depending on the film type used and how good the lens is and if it is the camera negative or a dupe or print.

The BLUE line is video, you see video has no signal in the very high frequencies but being over peaked has too much signal in the mid-high frequencies.

The GREEN line is Digital Cinema, it has some peaking of the mid-high frequencies and more detail in the high frequencies than video but less at the very top end than film.

Digital Cinema can be made to look like video if it is over peaked more, and to look like soft film if the peaking is reduced. On a higher resolution Digital Cinema camera the curve can be made closer to film than on a lower resolution Digital Cinema camera.

The graph is the absolute value of the moving probe or video signal of the test pattern that is shot in focus. The gray area of the test chart is where the lines get too close together for your monitor to reporduce so they average to gray.

MTF is Modulation Transfer Function, or the AM signal from the test probe modulated by the image of the test chart,

Here is a link about MTF,

http://www.kenrockwell.com/tech/mtf.htm

Since you need to get an image onto the film, the quality of the lens used affects the modulation amplitude of the amplitude modulation (AM) signal output from the probe or video camera (or pixel brightness value in a digital image).

I did not show aliasing bumps for the Video and Digital since the OLPF should be heavy enough to dampen those in a well made camera. But if aliasing bumps are there they can give a false signal making the Video or Digital camera seem sharper than it is.

Which looks better, Film vs. Digital Cinema vs. Video depends on how large the image is in your field of vision, at a distance a Video image might look sharper than Film, but on a big screen viewed up close the Film will show more fine detail when in focus, Digital Cinema may look sharper than Film on a medium size screen or when the Film projector is a little out of focus, since you can only see the extra detail in Film when the projector is in true focus and a very good lens is used (non-scope?), and you are close enough to the screen to see fine details, if you are in the last row of a theater with a small screen the Video might seem to have more visible detail because the peaking compensates for deficiencies in your eyesight, hence the debate about what is "sharper", the MTF graph tells the story, but where you sit and if the projector is in focus will affect your opinion...

 

[Mat@imageWork]

Hi guys,

I certainly don't want to say that F23 is better than red. Both have their strenth and their defaults...I just try to make thing in the perspective of resolution, compression algorythm, MTF. Everything is linked together is a certain but at the end of the day, this is your choice. I try to give a other point of view of different that could help people to choice the right format for their project. I have shot on red and like it. I have shot on 35mm too and like it also. Idem for F23 and D21. Dancad you make a very clear technical explication...MTF descibe the way we can see different frequency before they got unsharp. That's what assistant cam do when they shoot a chart. More you have line, higher is the frequency. A sharp image will "see" the difference between them, before they come "blurry". That what dancad describe bellow in a very precise way. Thank you man, that's how we learn form each other, like a small wiki-thread

see you, but don't ever thing I don't like the RED cam...

 

[D Namir]

The hardest thing with film is determining what is real detail and what is grain.

Any attempt to test Red and film is a head failure. Film is made of chemicals, grain and light process while a digital image is made of picture elements.
The only way to compare is very subjective by one asking himself if the image he is seeing is good enough as the other, if he likes it, or doesn't, as the other image, if he accepts it as a standard image for movie making as for many years he accepted the other, film, since there was no alternative.

 

[Peter Majtan]

Any attempt to test Red and film is a head failure. Film is made of chemicals, grain and light process while a digital image is made of picture elements.
The only way to compare is very subjective by one asking himself if the image he is seeing is good enough as the other, if he likes it, or doesn't, as the other image, if he accepts it as a standard image for movie making as for many years he accepted the other, film, since there was no alternative.

Not necessarily for DI. Even the film has to be scanned and it becomes a digital image. It is at this stage You can start comparing them... But I do understand what You are trying to say...

 

[David Mullen ASC]

Sharpness is complicated because you may be creating more detail at the expense of aliasing. The current HDCAM-SR cameras are amazingly sharp, but fine edge details and edge contrast seem more "zingy" to me (even with no edge enhancement engaged) compared to the RED, which seems pleasantly detailed but not hyper-edgy-sharp.

As for film, I believe that since detail is stored on grains which lie in different positions on each frame, the impression of detail & sharpness may be influenced more by running the footage versus looking at a single frame, which may suggest a lack of sharp detail sometimes, yet when projected as a series of frames, it seems very nice in terms of detail. So comparing single frames between film and digital may be a bit misleading as to what is going on in terms of the impression of detail. And as I said, it is possible to increase the perception of sharpness and detail in a digital camera but at the expense of aliasing artifacts.

 

[Charles Angus]

This fact of the moving grain affecting sharpness is why ideally 35mm film DI would be carried out at 6k or 8k - to preserve the entirety of the grain structure. And to prevent aliasing, of course.

 

[Dan Hudgins]

Data in the noise...

Data in the noise...

This fact of the moving grain affecting sharpness is why ideally 35mm film DI would be carried out at 6k or 8k - to preserve the entirety of the grain structure. And to prevent aliasing, of course.

You bring up a good point, if you compress the film scans the "sub-detail" that is "below" the noise level gets filtered out.

If you keep the scans uncompressed all the way to the filmrecorder, then your brain can pick out signal from the grain when the film is moving through the projector.

It is better to scan at 6K and output at 4K even if they do not "resolve" strongly detail sub 2K size, since to see the "sub-detail" you cannot move it to the nearest 2K pixel address, you must keep the data on the right angle to your eyes view, if you reduce the resolution to 2K then fine details that cannot quite be made out in a 4K scan freeze frame will be made fatter and lower contrast when the are interpolated over four 2K pixels, in other words the fine details hold higher contrast in a 4K or 6K image because they are smaller than a 2K pixel, if you blur them onto a 2K pixel their contrast may fall below the level of perception even when the film moves through the projector at speed.

Speaking of Data in the Noise, I processed one of the NoNoBe frames using adapive bandpass of different values for the Red, Green, and Blue. The Blue and Green both had very high noise and the Red and Blue histogram did not fill be bit range.

Something odd happens when you scan granny film, the grains that stand alone seem to be emphasized and look more prominent than they would in film to film optical printing. We have that problem using a Canon XTi in our 35mm film scanner made from Oxberry optical printer parts, the grain becomes too sharp. I have taken to stopping the printer lens down too far to blur the grains a little from diffraction. In film to film printing the grains bleed together, but in the digital scans they seem to stand out more, so I wrote a filter that lets be soften the blue image more than the other two, and to not blur the edges and such. I did not find an easy adjustment that worked well since the image seems both soft and high in grain, if you look the focus seems at the back of the set on the leaf pattern, and the guy sitting at the back seems a little sharper than the ones closer, maybe. Its hard to sharpen a soft image that also has high grain, unfortunately that is most of the footage I have to work with also...

If they were using a RED ONE they would maybe have gotten better results?

(There is a little shadow detail, but you may need to adjust your monitor brightness, I have a CRT monitor that may show more than a LCD monitor at both ends. The image is for viewing at gamma 2.2 to 2.4 or so... BTW: this edit of the sample NoNoBe image should not be downloaded or linked to, it is just posted here for technical review.)

 

[Felipe Jaco]

I saw a sufficient amount of movies in theaters that were shot on Genesis camera and I have to say that to me it looks worse than movies shoot on 35mm film. Somehow it feels flatter.
I am waiting these new movies shot on RED ONE to come to theaters to see it to see where is it in duel with Genesis.

 

[Peter Majtan]

Dan - could the problem with NoNoBe be the T-Grain stock? It suppose to have larger silver crystals that are also a lot flatter. Maybe they reflect too much light and this is why the scanned grain is too RGB-ish... To ARRI's credit - the scanner was not set up for this stock, even the folks at George Eastman House were surprised to see this stock, even more so since it was used by Kurosawa... The visit to GEH and Kodak's HQ's was an amazing experience. We got a private tour to most of the "off limits" areas at GEH and we met with some of the top people at Kodak. I have seen some things that blow my mind even today. Film is far from dead...

 

[Dan Hudgins]

"speckle" in scans...

"speckle" in scans...

Dan - could the problem with NoNoBe be the T-Grain stock? It suppose to have larger silver crystals that are also a lot flatter. Maybe they reflect too much light and this is why the scanned grain is too RGB-ish... To ARRI's credit - the scanner was not set up for this stock, even the folks at George Eastman House were surprised to see this stock, even more so since it was used by Kurosawa... The visit to GEH and Kodak's HQ's was an amazing experience. We got a private tour to most of the "off limits" areas at GEH and we met with some of the top people at Kodak. I have seen some things that blow my mind even today. Film is far from dead...

As far as I can tell, and the NoNoBe looks much like some pushed T200 we shot, when you print film to film in an optical printer single isolated grains get their contrast reduced and so do not stand out so much.

In the film scanner the sensor pre-amp before the A2D may have some peaking or at least has more flatish bandwidth than what film does, it other words the sensor pre-amp has a flatter MTF curve than film, so the scanner ends up giving small grains in isolation greater contrast than they would in film to film printing.

On film you get interactions between near grains, so if you have a line in the image it will have greater contrast than a single grain, but if you use a scanner the single grains have about the same contrast as lines and larger features.

This scanner problem of making small isolated grains have too much contrast becomes an issue, because you want detail to hold up in larger features made of several grains, and if you low pass filter to reduce the "speckle" you also soften the larger features, where as in film to film printing the film receiving the image image low pass filters single grain "speckle" and boosts the contrast on lines of grains and other larger image features.

Another issue of the scanner is that the pre-amp gets a signal from only one image axis during readout, that is it reads out like a TV camera one line at a time, so the peaking goes only on one axis(1D), whereas in film the "edge enhancement" effect of bromide goes in 2D or really 3D space in the film when processed. I don't know how you could make a 3D readout sensor that would simulate what goes on in film.

With a clean fine grain film to scan these differences between the sensor and going to film are less noticable. But with heavy grain you get a sort of aliasing of the isolated grains with the sensor pixels. If the Sensor undergoes de-Bayer like the Canon XTi (tm) we are using for tests, then there is some sharpening in the de-Bayer that makes the extra contrast in isolated grains even more of a problem.

I wrote my adaptive bandpass filter to have a pixel radius large enough to find edges without locking onto isolated grains too much, it seems to help, but still does not look a smooth as film to film printing would from the same negative.

As to the T-grain stock having more "speckle" than other types of film stocks, well what other film stocks are there to shoot anymore? Maybe since they have thiner layers to reduce the amount of silver that makes for more speckle since in the older stocks you had layers of grains in depth and now they are side by side so they do not blend into a jumble and you see them as isolated "speckle" in the shadow areas. My Brother is always pointing to the black and white movies of the 1930's and saying how did they get that range, well film back then might have had 2oz. of silver per 1000' roll and today it might be 1/2oz. per 1000' even less for print stocks. The thiner layers make the image more "brittle", sharper on a test chart, but not always better on a lighting high ratio shot.

Yes it seems Kodak is "holding back" which I think if foolish since once people sell their movie equipment they will not be going back to film, they should come out with every trick they have, now.

I know they can make sharper and longer scale films than they are, we have shot sound recording film long scale and it probably beats IMAX, the problem is you cannot project it since the projector cannot hold that focus. That is the big downfall of film, the projector design, the negative is great but you cannot see the detail on the screen all the time now that there is not a projectionist to "ride" the projection lens focus while looking through his spotting scope at the screen during the show. I have thought that they should have a remote focus control on one of the seats near the screen so a member of the audience could volunteer to keep the image in focus during the show.

Maybe if Kodak comes out with better films, and the "speckle" issue can be solved for movie film scanners by using 12K+ sensors, HDR, or something, then digital uncompressed 4K projection will give movie cameras some future?

 

[Petr Dvorak]

Gee, this is complicated stuff. :wacko:

he he

 

[bryanpapierski]

bottom line

bottom line

and bottom line is, if you dont know how to light/shoot it dosent matter!!! 2k,3k,4k,10k,300k? the camera dosent make you a better DP, or the story any better!

 

[Alberto Caprioglio]

16mm film is 3K

HAHAHA

35mm film is 6K

HAHAHAHAHAHA


Show me. Scan a slice of film with the best film scanner in the world and show me. Let me see how to compare resolutions between a 24mm wide frame originated by a film running through mechanical hell... and a clean, native 6k camera with the same proportions. I really want to see this miracle. Some one please publish it full size on homepage.
6K... HA! We are talking about over 20 megapixels. We shall not waste time with all the ridiculous claims that come from "industry" and any dog who feels like barking.

...medium with numerous proven workflows.

Workflow??? film is a pain in the ass compared to anything digital.
I'd choose Magix amateur software anytime over dealing with film.

I'd say:
35mm is good average quality for movies -that's why s16 is only if you are on tight budget.
s16 is not beautiful at all to show. Have you seen The Motorcycle Diaries? In the scenes with fine details like landscapes and architectures, it was so soft that it caused me continuos tears. 3K?! Mega bullshit.

S35mm film = around 3K and good for movies
S16mm film = about 0.5 - 1K and not very good for movies.

S16 is comparable to and all right for PAL TV. And it's not bad at all for a silly thumbnail.
We can't honestly ask more to it.

film is very sharp (250 to 1000lp/mm)

It's not very easy to get more than 200lines/mm from lenses, so even if film had infinite resolution, a small area would remain very small; = poor details.
The biggest news is that I'm tired to hear this.

If only 8mm film were 1/2 standard definition (try to watch 8mm originated material on TV, if you can stand it), then 16mm would be to be somewhere between 720p and 1080p.
But it's not. Otherwise 16mm would have rich details, like clean 720p and 1080p materials have.
Also, scanning film is yet another problem.

The lens on a 4K digital cinema camera does not resolve anything close to 4K at 80%MTF unless you sharpen the image

So just do it, what's the problem?
And... the same lens can resolve 6k on a film? is it magic?

But these differences are small in the final film print projected in a movie house with the projector out of focus... It is hard to keep film flat enough in the gate to get the focus that fine at large stops

And we must take this into account of course. Another goal against film. Possibly the definitive one. What a waste. Oh, such godlike quality out of the window. Very clever.

so for real shooting would they (film vs. digital) may look about the same

oh-oow...
Then, isn't shoting film an unbearable waste?