last updated 01 Nov 2009
NEIL SADWELKAR

Film DI Basics

Film Digital Intermediate - Basics

go to ...

QUICK INDEX

What is DI? Or Digital Intermediate?
A rather simple explanation of this new process for finishing films for release.


DI is 'in'

Anyone who's in the film business sooner or later has to deal with the magic two letter word - DI. Almost everyone seems to be doing DI for the past 6-7 years now.

So what exactly is DI? 

To get you up to speed on this process that's supposed to change the way movies get made, I've tried to put together a coherent explanation. From my experiences, from observations, and reading.

A standard definition

The basic definition first. DI stands for Digital Intermediate. It is a process, not an art or a science. Its that part in a film's production that needs to be done after editing and before printing. 

To explain this slightly better, DI is the process of taking the EDL or a cut list of an edited film, scanning only the used shots from the original film at a very high resolution, corlour correcting the shots one at a time, to make the film look distinctive, maybe doing effects and blending CG with "normal" footage, and finally making out a negative from which the producer can make multiple prints to show in theatres. 

This entire process is Digital - done on computers and hard disks. And it is an intermediate stage, between editing and release. So, Digital and Intermediate - hence, 'Digital Intermediate'. Or, in short, D I,

There are various stages in this process and you can read more on scanning, conforming, grading and recording on other such stages at various pages on this site. 

 

I'm not a "DI expert" with a decade of experience or anything like that. In fact no human has over 7-8 years experience in DI. Because that's how young this business is.

Actually DI replaces a traditional process followed for almost 6 decades.

A process that was done optically on a system called a colour analyzer.

With DI, ts possible to do the same thing digitally.


   

Isn't this the same as "reverse telecine"?

Yes and no.

A lot of people confuse these two processes. Here in India many ad film production persons are used to taking TV commercials on a DigiBeta tape and doing a 'reverse telecine' to film for theatrical release. So much so that I've actually met people who believe all films are made this way!! So if that is reverse telecine then DI is not reverse telecine.

OK let's go over this a bit slowly. Telecine is the process of converting cinema (film) to television (video). By video one means video at PAL resolution. By that definition reverse would mean converting television (video) to cinema (film). Again at PAL resolution. That's not DI.

Digital Intermediate also does both these processes, but works at a much higher resolution

In a conventional telecine process, frames from 35 mm film are converted to 720x576 pixels and put on to a tape. If, you reverse telecine from tape to film you have a situation where a small 720x576 picture is blown up to 2048x1556 (eight times the size).

When doing DI, the film is scanned at 2048x1556 pixels (also called '2k' resolution) into digital files. One file for each frame. After that point all the systems in the DI chain work on these files but they leave them at 2048x1556. Finally after all the action is over the recorder also takes these 2048x1556 files and records them out to film. So the original resolution is preserved. 

So, the last stage of the DI process where digital images are recorded back to film, is a sort of a reverse telecine.

 

Try this ... Take a 3 megapixel camera. Shoot a picture. Open it in PhotoShop. Scale it down to 720x576 and save it.

Now open the small 720x576 picture in Photoshop and scale it up to 2048x1556. And compare it with the original

You'll know what reverse telecine does to your picture.

 

But is film '2k' or more? This is hotly debated with some saying 2k is 'enough' while others saying '4k is necessary to capture the full resolution and some even hold that at least 8k is needed.

The practical reality, even in 2009, is that except for a very few, most films - in India and the world - are finished at 2k.

     

What happened before DI?

Just what happened before this new DI business was invented? I mean, weren't films made for over a century? Did Chaplin or Hitchcock do DI, or did K. Asif, Satyajit Ray and Shyam Benegal do DI for their films in the 70s and 80s? 

No they didn't.

Long long before the DI process was refined to where it's at today, films were shot as they have always been, edited as they have always been and finally graded optically on a 'grading table' or analyzer, shot by shot. Even now and for the forseeable future, some films will still be made this way.

Nothing wrong with the process, you can make great movies even without DI. DI just gives you more creative control over the final product expands the possibilities of story-telling.

Enuff of the theory and history. On to the process. Some of this is pretty basic. So feel free to fast-forward. Don't go very fast though. This is a short page. This explanations are rather simplistic and leaves out the nitties gritties. So if you're a techie whose been this way before, you'll find yourself saying "Yeah right, but what about ....?

 

 

 

In the times before DI, the process of grading a film didn't have a fancy name like Digital Intermediates or a short form like "D I".
But if you just want to have a name for what was before DI, then it is OI - Optical Intermediates.

     

Putting the film together

First the film is shot and processed. All five to seven hundred rolls of 35 mm negative. After these rolls are processed, they are telecined. Meaning converted to video.

In this telecine process, the telecine machine "reads" the "keycode" printed at the side of the film (between the sprocket holes) and assigns a timecode to each frame. It then makes out a table of keycodes vs. timecode. This is saved as a "log file" so the editing system can keep a track of keycodes which we're going to need later. 

These 'telecine'd' video rushes are captured into an Avid Film Composer or a Final Cut Pro system. The sound is also similarly transferred - it may be on DAT tapes or Nagra spools or hard disk recorders.

In this film edit system, picture and sound are cut together in a creative way and the film story is assembled. This is editing. You could say this is the second time the film is 'directed'.

After the edit is done, an EDL (edit decision list) is made. An EDL is a human readable lists of numbers in a tabular format. It descibes the edit numerically.

 

 

 

This is a fast forward explanation about a non-trivial occupation called feature film editing. A form of mediation that permits one to deny hunger and sleep.

 

     

Before doing DI...

After the edit, comes an intermediate stage before the final print of the film is made. Conventionally this intermediate stage was done manually and optically.

Since about 2001, computers and storage reached that point in their evolution where it became possible to do this "intermediate' process 'digitally'. Hence, 'digital intermediates' which is the point of this whole web page. And I'm really repeating this too often now.

But before you can start with the first stages of the DI process, you need to prepare the material you're going to DI. EDLs, relerence dumps, printouts etc. need to be done

More details on the preparatory work needed before DI on my DI - preparation page.

 

 

 

 

     

Scanning

The EDL made by the edit system - Avid or FCP or whatever - system contains numbers which tell the scanning system where the used shots are located. This scanning system then scans the film at '2k' and turns each film frame into a digital file - 2048x1556 pixels (12 Mb per frame, 1000 Gb/hour). Or '4k' meaning 4096x3112 (50 Mb per frame, 4000 Gb/hour).

Scanning is done either on a scanner or a datacine. Scanners work rather slowly - from 1-4 sec per frame, while datacines work fast 12-24 frames per sec even more.

Then there further complications - how do you conform the scans to the EDL? Scanning file formats, log and lin, 10 bit and 16 bit ... more on scanning on my scanning page.

 

 

 

I know, I know. An EDL contains timecodes but film has no time code, only keycode. So how?

More on this in page on scanning and conforming.

     

Conform

After all the shots are scanned, they lie on a rather large hard disk. Then a 'conform' system conforms them. This conform system is a computer that reads an EDL, and looks for the necessary shots on the hard disk, and then builds a timeline. Just like doing a normal 'online' or upgrade.

After conform, you have your whole film in the form of a timeline at full film resolution - 2k. All through the grading process, it is necessary to ensure the 'digital negative' is in sync with the original edit. The conform system can do side by side comparisons and ensure this.

But most conform systems are rather primitive editing systems as most real editors find out to their dismay when they get involved in DI.

The nitties and gritties on the conform process at a DI - conform page.

 

Most DI systems manufacturers trivialize this conform process. Because their system cannot do it cleanly and with 'zero' errors.

Make no mistake about this. No matter how well your film is graded it still has to be an exact clone of the offline. And this is something you find out when you make a married print. If the conform system is not 'zero error' the sound will not be in sync.

Then try figuring out with your producer exactly who pays for the print.

     

Grading

The conformed timeline is opened on a grading system. Sometimes it is the conform system that does the grading as well. And sometimes there's a dedicated conform system.

Either way, on the grading system it is possible to adjust colours of the film on a shot by shot basis. It is even possible to change colours of just one part of the image.

You can give any scene or shot a completely different feel than was originally shot. Or just enhance the feel that cinematographer designed through his lighting. If editing is the second time the film is directed, then grading is the second time a film is filmed.

Grading systems work similar to the grading consoles one is familiar with, in telecine suites. Primaries and secondaries, control surface, masks, and other details on my DI - grading page.

 

 

 

     

Calibration and Colour Management

While grading, one is looking at the picture on a display - a monitor or a projector. It's a digital representation of the original negative. A digital negative. What is needed is a system that ensures that the picture that one sees on the display is exactly what is printed on the negative. This is the Colour Management System or CMS.

And then there's calibration. Different displays behave differently even the same display behaves differently over time. Calibration ensures that the display shows the same picture through the course of its life. And one or more displays in the same facility show the exact same picture.

Calibration and colour management is done in more detail on my DI - CMS page

 

 

This is a really serious aspect of the DI process one that has made many a DOP and director nervous. How do you know what it's really going to look when it goes out to film?

And that's where a colour management system comes in.

     

VFX, 3D, other elements

Many films have effects of some kind. From simple dissolves and wipes to complex composites of live action and computer generated imagery - CGI. What is commonly referred to as 'opticals'

Putting together a fim negative with a computer generated image is no trivial task. Both are completely different mediums and respond differently to manipulation. Fortunately, if one finishes a film as a digital intermediate, then the entire film exists digitally. Integrating this digital negative with digital images made by a computer is relatively manageable.

 

 

 

     

Printing to film - recording

After all the grading magic and the conform magic, the film has finally to make its way back on to a 35 mm roll of plastic. So that people can buy tickets and watch it in a theatre.

Film recorders take a 'digital negative' from a hard disk and record it back to a 35 mm film. Different technologies are used for this - lasers, LCDs, CRTs. But the common thing is that film recorders are slow. Recorders are rather slow - 1-4 sec per frame. One day per 2000 ft reel.

More on recording on my DI - record page

 

 

 

 

 

 

 


 

 

go to ...

QUICK INDEX