Film is film and video is video.
And the video of a film is NOT that film.
Like the parable of the cave, told by Socrates, the video of a film is a shadow cast on the back wall of a cave by the real event taking place outside.  The shadow is not the thing itself.  The video is an after-image of the film, an electronic impression.

A widescreen film is a story told in images the shape of a mailslot (almost two and a half times wider than they are tall) and about the size of a Winnebago.

The video of that film is a story told in images which are almost square in shape and about the size of a large cat (17 to 30+ inches, measured diagonally … or from nose to tip of tail).

These are fundamentally different media.
Chalk and cheese.
Square pegs and round holes.
Turning one into the other has at best been a dicey business.
This discussion concerns primarily the problem of converting between the most dissimilar of the two media … a widescreen film, which plays in a theater at a 2.4:1 aspect ratio transferred to video (squarevision).  1.85:1 films are not discussed, because (a) they generally transfer reasonably well and (b) I don’t shoot in that format anymore.

The average home video monitor displays about 300 lines of picture information.  A 35mm release print of a film has the resolution, or ability to show detail, equivalent to about 2500 lines (this number varies from film to film and from expert opinion to expert opinion).  It is safe to say that film shows us at least six to eight times the amount of detail that video does under the best of circumstances.

These points are made by way of preamble to the following statement: Letterbox is NOT NECESSARILY the answer.

I make this statement as a filmmaker who has worked in the widescreen format and who also believes himself to be of sound mind and body.  I make it knowing that many purists, aficionados and collectors will consider this position heretical.  Many of these people believe that a matted or letterbox version of a film is the only acceptable or definitive version of that film.  And their reasoning is sound … that a director and cameraman composed the shots for that shape of image.  That cropping gives you less picture … that some of the film is lost.

My answer to that is simple:
On video, no matter what the format, ALL of the film is lost.
There is no film.  There is only video.

The purpose of this discussion is to take a hard look at what is best for a movie in this smaller, lower-resolution medium … What technique creates the TRUEST version of the film on video … truest to the narrative, to the excitement, to the emotion.

Consider the following: an NTSC video image letterboxed to 2.4:1 wastes almost half its height in masking.  Thus the remaining image consists of between 150 and 200 lines of picture information.  Even on the very best pro-sumer monitors, this just isn’t enough data to give a very clear picture.  In fact, another way of looking at it is that it is about half as clear as a full-frame, unmasked video image.

If one accepts the fundamental premise that even full-frame video is a compromise in available resolution compared to film, this near-halving of image quality should be considered absolutely unacceptable.  But purists seek it out, even though it is not so much widescreen as shortscreen.  You are getting less, not more.
The image is a furry, unresolved electronic shadow of the original.

This problem is exacerbated by the fact that letterbox videos suffer a further "distancing effect" by the very fact that they are preserving the entire film frame.  Its very virtue is in some ways a deficit.  Since in letterbox twice as much of the original image in visible while the width of the picture stays the same, you don’t have to have a degree in quantum physics to conclude that everything on the screen is half the size that it is in a pan&scan transfer.
Actors’ faces often become so small on the tube that they consist of just a few pixels.
Nuances of expression, and even recognizability, are often lost.

The masking, and the drastic reduction in size of the subjects in the shot, combine to produce a distancing effect.  We are "taken out of" the film.  We are less involved.  In fact, we miss a lot that the actors are doing in the scenes.  A nuance of expression, a gesture, a texture, a twinkle in the eye … all these are lost forever.  The expression on the face of an actor standing in a full shot (seen head to toe) may be unreadable.  Powerful and moving moments in a film will seem somehow lackluster.

Not all directors punch in for a close-up every time a dramatic moment comes up.  That would be a boring, knee-jerk response to shooting a scene and many great moments are allowed to play in master.  In the theater this works perfectly.  We see all that we are meant to see.  In a pan&scan transfer of the film, the actors’ faces will be big enough to convey the same emotions as were experienced in the theater.

In a letterbox transfer, this won’t be the case.  Some of the emotionalism is lost, some of the humanity.  The story becomes somehow less compelling.  We are left with merely an intellectual appreciation of the director’s eye for composition, and sometimes an enhanced sense of grandeur because the characters become dots in the landscape.

Granted these losses must be weighed against the aesthetic losses with a pan&scan transfer (of a ‘Scope film) in which a shot of two people talking may be missing one whole actor.

I am not saying that all video is bad; I am simply pointing out that there are fundamental differences between film and video, and that video must be addressed on its own terms.  There are solutions, and I have used them in my recent films.  But the solutions involve the way the film is shot originally, as well as the way it is transferred.

There are two ways to shoot a film for release in the anamorphic format.  The first, and to date the most common, technique is to shoot with an anamorphic lens.  This lens puts a "squeezed" image on the negative.  The squeezed image is printed by direct contact onto the print.  In the theater it is unsqueezed with another anamorphic lens, creating a widescreen projected image with a ratio of 2.4 (width) to 1 (height).

There is another way to do an anamorphic release at 2.4:1, and this was used for both The Abyss and Terminator 2: Judgment Day.  It has been around for about ten years, and is becoming quite common.  Generally called Super-35, it consists of using "spherical" lenses to photograph the image, creating a non-squeezed picture on the negative.  The entire width of the negative is used, from the perforations on the left to the perfs on the right, including the area which is normally reserved for adding the soundtrack later.  In order to make a release print, the image on the negative must be optically squeezed in a process called "formatting" onto an internegative.  This internegative is then contact-printed, yielding an anamorphic release print which is indistiguishable from that of a film which was originally shot anamorphically.

The important distinction between these two formats as regards video is this: the anamorphically photographed film has NO additional picture information at top or bottom beyond what is seen in the theater.  Thus to get it onto video you have to throw away about ONE HALF of the image by cropping the sides.

At its worst this can mean that a shot which contained two characters facing each other will now contain only one.  Or maybe just the noses of both.  This is a horrible compromise.  Even with modern telecine equipment with naturalistic "non-linear" pans to simulate camera moves, transferring is basically image triage, trying to save what is most important and letting the rest die.

The situation is even worse if the director of the film is dead, unavailable or doesn’t care, and the transfer is done by some well-meaning hourly-paid colorist who must make fundamental narrative decisions about the film (this is essentially co-directing the movie without credit).

Obviously a letterboxed version is a truer representation of the film in this case.  The loss of image resolution and the "distancing effect" on performances represent the lesser of the two available evils.

But if the film is NOT shot anamorphically, but is shot in the Super-35 process, another option is available.  An option between the hard-to-see letterbox and the dreaded pan&scan with half the picture missing.

It is possible to have a dramatically exciting and visually complete PAN&SCAN transfer-to-video of a film which was exhibited in anamorphic widescreen in the theater.

This option requires that the filmmaker take responsibility BEFORE the film is shot, at the point that the photographic format is decided upon.  And it requires that the filmmaker shoulder responsibility again, AFTER the film is done, by carefully supervising the transfer.

If the film is shot in Super-35, the theatrical image uses only a band across the middle of the negative.  Think of it as letterboxing in the camera.  Just as a letterbox movie on video creates a 2.4:1 ratio by masking the top and bottom of a square frame, the Super-35 negative doesn’t use almost 50% of its height, since a full-aperture 35mm film frame is almost as square as video.  The unused part of the negative is photographed, but not used in making the theatrical release print.

This additional image area is available during the video transfer.  It may sometimes have a light in it, or a microphone or some dolly track.  But usually it’s just fine.

It can be used to recompose the shots, so that all the actors are in frame.  In principle it would be possible to transfer the 1.33:1 full-frame Super-35 negative almost directly to the 4:3 ratio video frame (they are almost identical in shape).  In practice this introduces too much dramatically extraneous stuff, and makes the shots feel too "wide."  A tight close-up becomes a medium shot and so on.

It turns out that treating the Super-35 negative as if it was originally a 1.85:1 negative works almost perfectly.  Some slight panning is required, but it is mild enough to be virtually invisible on today’s equipment, especially if it is artfully hidden within existing camera moves.  I have found that with the extra picture available, panning, tilting and even zooming can be used to really enhance the dramatic impact of the shots.  In some cases, the shots can be made even better than they were on film.

In fact, the transfer of a Super-35 film to video offers a whole palette of techniques to optimize the image.  For example: let’s say you have a wide establishing shot in a film, in which the camera slowly approaches your principal actor.  The goal of the shot is to show the setting, and the character’s relationship to the setting.  Recognizing the actor is important, and some facial reaction that he has may be critical to the moment.  This may seem to be a no-win situation in video.  You either sacrifice the grandeur (in pan&scan) or you sacrifice intimacy with the actor (letterbox).

With Super-35 you can have both, with a z-axis move (zoom) added on the telecine, hidden within the existing camera move.

You start by scanning almost the entire negative area at the beginning of the shot.  This gives you almost all the width you had in a theater, plus some stuff above and below frame that you never saw.  Then as the camera pushes in, you can use the telecine to scan closer on the z-axis (zooming in) so that your final frame is close enough to read the actor’s thoughts on his face.  The theatrical "experience" of the film has been approximated or "simulated" using subtleties of the advanced transfer technology currently available.  The zoom is absolutely undetectable within the existing camera move.

In the pan&scan transfer of The Abyss: Special Edition I had the following problem shot: Bud (Ed Harris), wearing the fluid-breathing suit, stands at the edge of the cliff overlooking the great abyss.  Then he gathers his courage and leaps off.  A very important shot dramatically.

In the letterbox version, you get the scope of the shot, and a sense of the vastness of the abyssal depths, but because Bud’s figure is so tiny, it might as well be a G.I.Joe doll.  Since the reality of the shot comes from seeing that this is a real human being standing there, it wasn’t working.  In the pan&scan transfer, I framed in closer to Bud, so that we see him clearly, gathering himself for the jump.  Then as he starts to go, I pan and tilt down to emphasize the black void beneath him.  This type of tilting move would not have been possible on an anamorphic negative.  The result is the most dramatically powerful version of the shot possible in a video format.  It is also a different shot, in a sense, than it was in the film.  In the theatrical release, there was no movement in the shot.

Some would call this blasphemous.  But, as an actor once said to me, we’re not writing pages for the Bible here.  This is movies.  There are no rules.  And only one commandment: tell the story the best damn way you know how.  In video I can’t tell it using a 50-foot-wide screen to make my point, so I’ll do it differently.

I don’t feel too guilty about using these techniques to make the video image as good as it can be.  Video is a compromise medium to begin with, so a little revisionism in the service of a better viewing experience is, I think, justifiable and desirable.

The point I’m trying to make is that it’s not as simple as it was a few years ago.  There are more than two ways to skin this cat.  I am asking discerning fans and collectors not to think in binary terms … that pan&scan is bad and letterbox is good.

If (and only if) a film is shot in Super-35, you should consider the possibility that the pan&scan transfer is a superior viewing experience in the video medium.  Clearly letterbox will still have a certain archival value, in preserving compositional elements of the film.  It is a necessary record of the way the movie looked on film.

Today movies have a dual life.  Most films are seen by more people for the first time on video than in theaters.  In fact the entire business of financing films these days is the business of selling or licensing video rights to make the damn things in the first place.  Very few films are profitable in the present marketplace solely on their theatrical returns.

So I believe it is more and more important for filmmakers to consider the needs of the video market while they are making the film, or even before they start.  It creates a much better product later.

To properly transfer a feature film to video may take 50 to 100 hours of painstaking supervision on the part of the director.  Brightness, primary and secondary color correction, frame size and position … all these factors have to be optimized for every single shot in the film (3000+ shots).  Often the color is corrected at several points within a single shot by adding undetectable color dissolves, and all the moves on the x, y, and z axes must be programmed and checked for smoothness and organic flow.

The problem with Super-35 is that all of those wonderful options in transfer are a double-edged sword.  If they are used as creative tools by the same director who shot the film in the first place, then they are valid enhancements.  If that director is not present, then these options become merely additional ways to screw up.  Shots can be too wide, too tight, or the emphasis can be in the wrong place.  So the onus is really on the filmmaker to make it all work, to deliver the highest quality product to the marketplace.  It is a new responsibility of the video age, and directors must accept it.

But let’s face it, a pan&scan transfer is a lot more work than a letterbox transfer for the director.  And even hard-working directors, who are demons of energy on the set, tend to get lazy come transfer time.  But if they do … they are ripping you off.  You deserve the very best version of that movie which can be wrested from the film emulsion and put down in pixels.

What I do is lay down the pan&scan transfer first, because it is far more labor-intensive on my part, and then let the colorist do the letterbox transfer using the same list of color corrections (recorded on a floppy disk).  Because letterbox requires no repositioning or XYZ moves it is therefore, ironically, a much easier process.  My creative/technical chief at Lightstorm, Van Ling, supervises the letterboxing and I check the results before it is laid down on D1 tape, to make sure the color has not drifted.

So in selecting a collector’s edition of a recent film, one must probe beneath the surface.
Did the filmmaker supervise the transfer?
What format was the film shot in?
The 1.85:1 format?
The Super-35 2.4:1 format (shot spherical / projected anamorphic)?
Or the various older 2.4:1 formats (shot anamorphic / projected anamorphic) such as Cinemascope, Technovision, and Panavision?

This factor alone makes a huge difference.  In the first two cases, the best extraction of the image for video may well be the pan&scan version.  In the case of the anamorphically photographed films, the best version will likely be the letterbox transfer.  It is still probably the lesser of two evils for films shot in "Scope."

At least until we get some kind of high-definition video.
Then, of course, the poor directors all have to go back and transfer their movies all over again.
Oh well.

The important thing in the meantime is to make sure that both letterbox and pan&scan versions are available, and that the collectors consider all the facts in deciding what to buy.

To sum up: I wanted to catch your attention by saying letterbox is not the answer, but clearly, it has its place.  For classic films shot in anamorphic, it is the best record we have on video.  But looking forward, to a future where we still have a choice how to shoot the films and subsequently view them in the video realm, both video consumers and filmmakers should be more flexible in their thinking.

Super-35 makes little sense other than as a way of shooting a widescreen film and having it be video-friendly later.  If this technique is not embraced, and indeed lobbied for and rewarded by those it benefits the most—the video consumer—then filmmakers will be slow to change their thinking.

Just what you need, right?  Another crusade, just when the letterbox crusade was starting to finally have some effect.
Think about it.
Look at both versions of The Abyss or T2 and judge for yourself.
(See … this was just a ploy to get you to buy both.)

Sincerely,

Jim Cameron