The Film Out Process

[Cody Muller]

Hello!

 

I'm looking for information about the procedure of transferring digital files onto 35mm (and telecine it back onto digital). We are thinking of doing this with 1080p footage we shot. The footage will be transferred onto a black and white 35mm negative and telecine to 2k digital back.

 

I understand films like Nebraska and Dancer In The Dark were done using this technique. However, I have been unable to find much information on the process.

 

I understand it's expensive. However, it is hard to find any tests online, showing the before and after.

Showing a difference in grain, dynamic range, etc.

 

Do any of you have experience in this field?

 

Thanks!

 

 

[David Mullen ASC]

I don't think "Nebraska" recorded out to film and then scanned back to digital just to make the images look like they were shot on film -- they just applied film grain to their digital images. That film grain software was based on a scan of film though.

 

You could certainly do a laser recording out to film and then do a 2K scan of that recorded film, it just seems a bit silly. That approach was more common ten years ago (particularly for standard def video projects) before there was 24P and so many software tricks to add film artifacts to a digital image.

[John Rizzo]

Hello Cody

 

back in 2011 we did just that for a short film called "Curfew"

 

[John Rizzo]

Hello Cody

 

Back in 2011 we did just the work flow you are describing for a short film called 'Curfew' It was shot on Red and we recorder out a 35mm negative to camera film stock 5213.

 

Here is a Webinar that explains how it was done.

 

http://vimeo.com/35973323

Edited October 28, 2014 by John Rizzo

[Carl Looper]

There's nothing much to be gained by taking digital to film and back again other than to see that there's nothing to be gained from such and burning a bigger hole in your pocket than you otherwise need.

 

There are far cheaper ways to get exactly the same result, and by exactly I mean exactly. Not just theoretically.

 

The grain of film is not that which makes film have it's film look. You can add grain to a digital image (either by going to film and back again, or just adding grain to the digital - a lot cheaper) but you won't be any closer to a film look by either method.

 

The grain in film is just a side effect of a more important process taking place in film. Applying this side effect (ie. the grain) to a digital image does not get you any closer to obtaining what film obtains. It simply acquires the side effect ie. the downside, and none of the upside.

 

It's completely silly. But silliness is by no means a criticism. Just an observation. Its an observation that one can use to motivate the use of other means for performing such silliness.

 

C

[John Rizzo]

Hello Carl

 

Your comment is a matter of your opinion and that is fine, the people that did Curfew were more than aware of all of the options of adding grain digitally (which at this point there are many) and none of them had the same effect of this film out work flow.

We are about to do a film out test on a very big budget movie the dp and the director have tried adding grain via their post house and are till not satisfied with the final look. This DP did a Coke campaign were they did this work flow and it could be duplicated via digital only.

Is this silly as you say, maybe,I think it is more about creating different artistic looks to a final piece.

[David Mullen ASC]

I used the word "silly" too, but perhaps the word should have been "indulgent" since the technique works, it's just a rather elaborate and expensive way to achieve a result, especially on a long-form project. In this day and age where a DP has to argue with a producer to carry an extra zoom lens, where shooting days are being cut to ridiculously short schedules, to spends tens of thousands of dollars extra to take a feature length film and record it to 35mm film and then scan it back to digital just to add some grain and instability and to soften the image is hard to justify.

[Carl Looper]

Hello Carl

 

Your comment is a matter of your opinion and that is fine, the people that did Curfew were more than aware of all of the options of adding grain digitally (which at this point there are many) and none of them had the same effect of this film out work flow.

We are about to do a film out test on a very big budget movie the dp and the director have tried adding grain via their post house and are till not satisfied with the final look. This DP did a Coke campaign were they did this work flow and it could be duplicated via digital only.

Is this silly as you say, maybe,I think it is more about creating different artistic looks to a final piece.

 

Yes - sorry John - I shouldn't have used the word "silly".

 

Indeed I backtracked on that by saying "it's not a criticism" but an "observation". Which I meant quite sincerely.

 

What is the CD hoping to achieve?

 

If you are failing to get the look you are after in digital post, the fault won't be due to any fundamental limitation in digital post methods. It will just be due to failing to find the correct digital post method. There is a correct method. And it is in relation to this method that no amount of going via film will do it any better.

 

In other words the point I'm trying to make is that there's no fundamental reason to go via film. But of course if you can't solve it digitally then yes, one can certainly go via film and spend the appropriate amount of dollars.

 

But this does beg the question as to what the CD has in mind when they say the digital tests were not as good as they were hoping. What are they hoping? It could very well be they are entertaining some myth about film and grain, and this could be putting artifical limits on how they are managing the digital pipeline.

 

There could very well be a false hope in terms of what the film path might achieve.

 

Carl

[Carl Looper]

As previously mentioned, the grain of film is a side effect of a more important signal occuring between film and light (which we'll call 'analog' for want of a better term) - but this analog signal (so called) is not there in a digital signal. It is absent. More importantly you can't recreate it by just adding it's side effect (grain) to a digital signal. The analog signal will remain absent. All that will be added is the side effect. The grain. And this side effect can be done digitally. Or it can be done by printing to film and back again. The digital method is just more desireable because it should be cheaper. Of course, it depends entirely on the skills and tools and cost of those managing the digital path versus those managing the film path.

 

But thats not to say there are no uses for the addition of grain or noise, eg. one might use such methods to match grain across different shots, or to prepare material for downstream processes that might otherwise exhibit greater quantisation were grain (or noise) not added prior to such. In other words grain or noise has some technically useful properties - not just for failed attempts at a "film look". But each of these are not in any way better facilitated by a film path. Indeed you're more than likely to lose control of such in a film path. If the economic argument didn't already matter.

 

C

[Carl Looper]

The relationship between a continuous signal (an image or wave function) mediated by light, to which photochemical film is exposed, is materially encoded on film in terms of random interactions between the random arrangement of silver in the film and the equally random arrangement of photon detections. The random aspect of both ensures a statistically neutral stance in terms of the otherwise continuous signal (wave function) being encoded or mediated. A side effect of this is localised grain or noise (god playing with dice) but which has no affect on the continuous signal (the image or the wave function). Statistically (ie. globally) the noise/grain cancels out. The image/signal/wave function occupies this same domain (ie. where the noise cancels out). But the materials (photons, silver, etc) don't. They occupy a local domain, no less visible, but producing that effect we call grain or noise.

 

In film there is the opportunity for very fine details being mediated because the randomness is pure randomness, ie. statistically neutral and at all scales (down to subatomic levels). But the cost of such is noise or grain. There's a particular cost/benefit tradeoff going on in this process.

 

Now if you start with a digital signal, the signal is discontinuous or quantised. There's a different sort of tradeoff going on. The arrangement of materials is such that it sacrifices neutrality at lower scales in exchanage for less localised noise and a more computationally manageable information system. Now a digital signal (image) occupies the same domain as a continuous signal/image, but accordingly it is likewise unaffected by any localised noise. Adding noise (be it synthesised or through prior capture of neutrally exposed film, etc) does not affect the digital signal. Just as it does not affect the continuous signal. The noise is just the visibility of the localised variations. These localised variations are easy to interchange between digital and film methods. It's the signal (the image) which fails to be affected by such interchanges.

 

In other words, if you look "through the grain" of a digital signal (to which grain was added by whatever means) you'll just see the original digital signal. No different from what it was in the first place other than it might be harder to see. And in film, if you look "through the grain" then you'll see the underlying non-digital signal: the 'analog' (or continuous signal). If likewise harder to see.

 

The grain or noise is just a localised phenomena. The image proper, digital or otherwise, remains lurking there on the other side of the haze, so to speak. No different from how it began.

 

Carl