Sunday, February 15, 2009

A Question on Traveling Mattes

Here is a question that I have wondered about a lot:
Ub Iwerks received an Academy Award for his groundbreaking work on Traveling Mattes, for combining live actors and animation in Mary Poppins. Ub had worked with this since 1923, when he figured out the technique to combine Alice with her Cartoonland.

The actual 1963 camera used a beam-splitter prism to record the live actors on black on one roll of film, and the matte on another. My question: why not have both on the same roll, and eliminate the possible difference between the two rolls, like shrinkage etc.?
Having the characters and the bright yellow sodium matte area on one reel and then separating them by filtering them in an optical printer seems to be a more controlled solution. Then why split it in camera instead? Were optical printers not precise enough?
Did the colors bleed? Was it hard to separate if all was on one roll?
Anyone with a definite answer is welcome to comment!

Below three images of the retrofitted three-strip-Technicolor camera, dubbed Traveling Matte Camera #1.
First Ub with Bob Broughton, who passed away a few weeks ago, then Ub alone with the camera, and finally a shot of the open camera, the front flipped down exposing the special prism, of which there seems to have been only one, very closely guarded for many years...
Ub Iwerks & Bob BroughtonUb with Traveling matte CameraUb with Traveling Matte Camera open
[Note: I rephrased my question in a comment...]

Labels: ,

12 Comments:

Anonymous Pseudonym says...

It's not enough to have the character on a black background. You also need to have the background with a black hole in it, so that the background doesn't bleed on to the character when the two elements are printed together.

For that, you need a negative of the character's matte. And for that, you need the character's matte to take the negative of.

You might be able to do it with one roll if you had a rostrum stand which handled aerial image projection, but I don't think that had been invented in the 20s.

Sunday, February 15, 2009 at 9:50:00 PM PST  
Anonymous Michael Sporn says...

I believe you had to have two separate rolls of film to bipack the matte and BG - so that you could come back and lay in the characters on another run through the opt printer. Producing two images from the original would probably keep it most accurate - if the gods were on their side and nothing was amiss.
At least, that's how I understand it.

Monday, February 16, 2009 at 5:14:00 AM PST  
Anonymous Hans Perk says...

Thanks guys, but I think you miss the point in my question. Let me rephrase!

If you had one reel with characters against the sodium screen background color, you could make a matte filtering that color (and after that, a counter matte) in an optical printer, using the original reel for the characters, so no generation loss occurs for the live actors.

If you have the matte on one reel and the characters on another, you are open for stretching of film, weave of the camera mechanism etc., things which in a printer would be (or should be) much more controllable. Then why go for two rolls instead of the one?

Using the second reel as a matte is basically a real-time on-the-spot version of an optical printer - it is still a projection and not a contact copy. Maybe it was deemed easier to do it this way than having to micro-adjust a printer? Instead developing a special "sodium screen color sensitive" film would have been preferable, as one then could contact print directly from a single-reel print. Was this ever tried?

Monday, February 16, 2009 at 7:09:00 AM PST  
Anonymous Pseudonym says...

If I'm understanding your question correctly, you're actually describing how the sodium matte process worked.

The idea was that you used sodium vapour lamps to illuminate the matte screen. They're essentially identical to the vapour lamps used in treating jaundice in babies, only using a different element; they are designed to produce very narrow band of wavelengths. In a good lamp, there's only one or two wavelengths present, and they're very close on the spectrum.

The useful feature of the narrow-band yellow light that these lamps produce is that it fit between the bands of colour that Technicolor or Eastman colour film recorded. So on the colour negative, the characters just showed up against black, not yellow.

The other film in the bipack was only sensitive to the sodium vapour light, so that only the matte showed up on that film. Essentially, you got red, green and blue on one film, and alpha on the other, automatically separated out.

Now it might have been possible post-1952 to produce Eastman colour-like film with four types of dye (red, green, blue and sodium vapour wavelength), assuming there was no chemical problem. However, it seems to me that while this would simplify the acquisition, you'd still need to make copies of the matte channel for optical printing, and during printing you'd have to use a lamp that contained all wavelengths except the sodium ones, and that's assuming that the film dye transmits the same wavelengths as it captures. This sounds even more error-prone than the sodium matte process.

Monday, February 16, 2009 at 10:37:00 PM PST  
Anonymous Hans Perk says...

Thanks, Pseudonym - the "yellow band in between two of the color film's sensitive areas" explanation is something I had not thought about before.

I cannot help thinking, though - could one not filter that out in camera, for any color light? And, if the light were to be captured on the film, could one not make the appropriate mask by filtering only that color to a matte negative?

Tuesday, February 17, 2009 at 5:35:00 AM PST  
Anonymous Floyd Norman says...

Wow! Great questions.

I won't pretend I totally understood the sodium matte process. I do know that our optical composites were better than anybody else's at the time.

I never met Ub, but I did have a chance to talk with Eustace Lycett, and he even offered me the opportunity to hold the special lense Walt had purchased from the Korda's in the U.K. Of course, I wouldn't dare touch the thing because it was one of a kind.

Wednesday, February 18, 2009 at 2:17:00 PM PST  
Anonymous Pseudonym says...

Floyd, do you know anything about the history of aerial image projection? Was that another Ub invention?

Wednesday, February 18, 2009 at 10:26:00 PM PST  
Anonymous spokeshave says...

The sodium process was also ideal for creating a perfect matte of semi-transparent items, like the veil on Mary Poppins hat. A glass jug of water poured into a glass would also get a matte far better than blue screen would give you or any other matte process. I thought that the sodium process used 3 strips of film so one got both male and female mattes in one go? Plus having all your rolls from the same batch helped reduced shrinkage? I think aerial image is an Oxberry invention, possible...

Thursday, February 19, 2009 at 9:19:00 AM PST  
Anonymous Hans Perk says...

I remember recently seeing a photo of one of the Disney Multiplane cameras using only one level, with a projection unit in front of it that clearly was an aerial image unit. I do not know anything of its history, though...

Friday, February 20, 2009 at 2:32:00 AM PST  
Anonymous Floyd Norman says...

Yes, I would agree. Ub had nothing to do with that.

I believe Aerial Image was an invention by Oxberry. My partners and I used this technique way back in the sixties on a film we were doing. And, of course, we were shooting our animation on an Oxberry camera.

I remember thinking we were pretty high tech even back then.

Thursday, February 26, 2009 at 6:28:00 PM PST  
Anonymous PXLpainter says...

This is a great article!

For historical reference and clarification, I recently interviewed Petro Vlahos (93) who is the inventor of the color difference traveling matte process and developed the one and only multi-coated prism (he holds the patent) that he sold to Ub Iwerks. Ub perfected the camera process as shown in this article, but he didn't "invent" anything.

FWIW - the sodium vapor light which is around 1300ºF (700ºC)color temperature is split off using the prism and thus eliminates the filters Ub was originally using which cost about 2-3 stops of light in the camera.

Also noted, Petro Vlahos developed and patented the electronic blue screen traveling matte process, for which he won a technology Oscar for after the production from MGM's Ben-Hur.

*Shameless plug: you can read more about this in my book "The Green Screen Handbook" and I will be posting more on my blog in the months to come as well as an upcoming documentary with Petro and many other pioneers of the compositing industry.

Cheers,
Jeff Foster
http://PixelPainter.com

Sunday, January 10, 2010 at 10:03:00 PM PST  
Anonymous Brad says...

The camera used only two magazines of film. One had color negative film, the other black&white negative film (to record the matte). Since the sodium was so narrow in bandwidth, and it was out almost out of the gamut of the color negative film, it didn't cause the same level of color contamination that, say, blue screen does, but it was still there. You can see the yellowish fringes. The background did not shoot as black, though. In fact, if you ever see color photos of a stage where the sodium screen was set up (there are some production photos of Mary Poppins) you'd notice the background appears almost brownish-orange probably because even photo camera film must have had difficulty picking up the intense sodium.

Many years ago, at Expo 86 in Vancouver, there was a exhibition (of holograms) which also included a room bathed solely in pure sodium light. As you entered it and passed through it, since the spectrum was so narrow, anything in the room looked like it was in black & white, including oneself. Everyone in the room looked like they were in Pleasantville.

Sunday, April 18, 2010 at 11:01:00 PM PDT  

Post a Comment

<< Home