V: HAL’s point of view

Filming HAL’s point of view (POV)

This article is a very deep dive into what lens was used to film HAL’s point of view (POV) in 2001: a Space Odyssey. An obscure question, to be sure.

Let's begin by taking a look at a screengrab from the movie. This shot is supposed to represent what HAL can see from his stationary glowing red eye in the operations console within Discovery's centrifuge.

A few things to note about this scene. First, the shot is crazy wide-angle, covering almost a full 180° view of the interior of the centrifuge. But the shot also has a visible image circle, with black areas around it. Therefore the lens used in this shot was a circular fisheye. 

Second, although a circular fisheye was used for the HAL POV shots, the lens wasn’t able to project a full circle onto the imaging area. Whatever took this shot is almost circular in its coverage, but the image has been cut off the top and bottom. This is particularly noticeable if you overlay a circle over the picture, as above.

Third, the image quality in these POV shots isn’t all that great. There’s a generally diffuse softness to the picture, some glow around highlights, and a lot of chromatic aberration (colour fringing around the edges). That means the shot was probably filmed with the lens wide open, which suggests it was a slower lens. Most of the lenses known to have filmed the centrifuge scenes had T-stops of about T-2.8. 

Wasn’t 2001 filmed using Panavision gear?

During the closing credits of 2001, a title card reading simply “Filmed in Super Panavision®” appears. This was a marketing name for the 70mm spherical lenses and cameras supplied by American cinematography giant Panavision.

However, the reality of Kubrick's use of Super Panavision is a bit more complicated. Since 1964 Panavision has famously only renting gear to production companies, and doesn't permit the sale of their kit. And I don't think they ever released a circular fisheye lens for their format. The widest lens they offer, so far as I know, is 28mm. But there are many scenes in the movie that were clearly shot with much shorter focal lengths than 28mm. (the shorter the focal length of a lens, the wider the field of view)

Douglas Trumbull was interviewed on this topic in 2005, and said:

“There was a commitment to Super Panavision, they called it Cinerama, but (Kubrick) was feeling kind of frustrated with the lenses available from Panavision for that medium. So we started experimenting with Nikon lenses and he found out that the Nikon lenses, which were designed for the 35mm slide format, actually had a field of image at the back of the lenses that was enough to cover the 65mm film format, which was just a little larger. And so, even though we were forbidden from publicly announcing that Nikon lenses were being used, we were using them all the time.”

Wasn’t 2001 filmed in Cinerama?

There’s a lot of confusion over the name “Cinerama”. Cinerama was initially the name of a complex three-lens system used to make immersive widescreen films; a technology generally known today as “three strip” Cinerama.

This innovation was introduced to the public in 1952, and a handful of movies were released using the process. True 1950s Cinerama was extremely cumbersome, expensive, and complex. A massive 35mm film camera, equipped with three 27mm wide-angle lenses, was used for filming. The camera contained three separate camera movements, for recording the scene seen by each lens to three separate reels of film. A single rotating shutter was used by all three lenses, guaranteeing that the footage was properly synchronized. 

Projection of the movies to an audience was difficult, since three separate 35mm projectors (each controlled realtime by three projectionists in three booths) were needed to show the image on an enormous curved screen. Two shimmering rows of sprocket holes were visible in the system where the three projections overlap, but it still provided a pretty awesome amusement-park experience; albeit perhaps not one best suited for effective narrative storytelling.

You couldn’t even change field of view (ie: focal length) with three-strip Cinerama. The camera was forever fixed with a panoramic field of view of 146°.

The inconvenience and expense of the system meant that no three-strip Cinerama film was made after 1963 (2001 came out in 1968). However, to preserve their investment in the name and the theatres built for the system, some Panavision-shot 70mm films began to be branded as “Cinerama” productions. When shown in Cinerama theatres with curved screens the films had an immersive wraparound quality similar to three-strip processes. The overall image resolution of 70mm Cinerama was lower than that of three-strip, but you didn’t have the visible sprocket hole problem, or the cost of three separate projectors and trained projectionists.

So. What film process was used for 2001?

2001 was one of those Cinerama-branded 70mm films. It thus did not use a three-strip process – it was filmed primarily using Super Panavision 70, which runs 65mm film through the camera, and employs 70mm film for projecting the final picture. (the extra 2.5mm on each side of the sprocket holes stores audio tracks) Some of the Super Panavision 70 cameras seen on 2001 sets bear the “for Cinerama” logo for this reason – it was purely a marketing exercise.

At its initial release 2001 was shown reserved-seat “roadshow” style, in both Cinerama theatres and regular 70mm-equipped theatres. And the former, projected on the enormous curved screens and accompanied by the Cinerama branding, seems to have given some viewers the impression that they were watching a three-strip movie. (a myth that endures to this day) Finally, prints were struck in 35mm for distribution to regular theatres.

But as noted above, Kubrick found the limited range of lenses available from Panavision problematic. Fortunately, Super Panavision 70, Todd-AO, and MCS-70/Superpanorama 70 were compatible 65mm spherical processes with an aspect ratio of 2.20:1. (2001 did not use Ultra Panavision 70, which was an anamorphic process with an aspect ratio of 2.76:1) This gave him some choices.

Some brief segments of 2001 were filmed using Todd-AO cameras, as can be seen clearly in behind the scenes photographs such as the one in the “Fairchild-Curtis lens” section below. I have also seen a photo showing that the psychedelic sequence was filmed by a Todd-AO camera on the "slit-scan" rig built by Douglas Trumbull. The Todd-AO cameras were apparently based around “Fearless” 65mm cameras from the 1930s, amazingly enough. They had exposed belts and pulleys for the take-up reel on the magazine and looked totally ancient.

Finally, I’ve heard the claim that certain effects sequences – specifically some of the colour-altered aerial footage in the Stargate sequence – may also have used European MCS-70/Superpanorama 70 cameras, though I've only seen references to it here. The idea would have made sense, since the MCS Field Camera was relatively light weight for a 65mm camera, and included a rotating mirror mechanism that meant framing was much easier; all ideal for aerial photography.

The film’s short handheld sequences – the TMA/monolith excavation ramp and Bowman climbing the ladder – were shot by Kubrick himself, using the Super Panavision handheld camera — 24 pounds/11 kilos of 1960s film technology!

A Mitchell BFC 65mm camera can be seen inside the alien hotel room – I don't know why they used that instead of a Panavision camera.

Interestingly, one behind the scenes photograph clearly shows a 35mm Mitchell BNC camera inside the centrifuge. This seems to have been used to film the brief interview scene where the two astronauts are sitting casually at the HAL console, with their backs to HAL. (a somewhat inexplicable scene – where was the camera supposed to be, narratively?) The BNC's footage was then reduced optically, since the simulated IBM proto-iPads on the dining table had screens lit by 16mm projectors.

The Fairchild-Curtis lens

The Fairchild-Curtis 160° f/2.0 lens, now in the collection of the Academy Museum of Motion Pictures in Los Angeles, is an unusual one no matter how you look at it. It was constructed by the Fairchild Camera and Instrument Corporation, a military contractor not really known today for making lenses. And it's an enormous beast of a lens with a massive front element the size of a small dinner plate.

It appear to have been a one-off product (serial 1001) which wasn't directly connected to the usual Cinerama-branded movie systems, despite the "Cinerama" labelling on the lens cap. It seems it may have been derived from a custom lens manufactured for “Journey to the Stars,” a 15 minute short film shown at the US Pavilion at Seattle’s World’s Fair, the Century 21 Exposition, in 1962. A 1963 American Cinematographer article notes that the lens was very slow and had to be stopped down considerably for best effect.

This lens was used on a modified Cinerama 70mm camera to film simulated planetary sequences, which were then projected to audiences on a large dome for the film To the Moon and Beyond, for the New York World's Fair of 1964. It was auctioned off in 2004, and acquired by the Academy Museum. The private collector who sold the lens billed it as having been used to film the HAL POV shots from 2001.

That would be pretty cool if true, but regrettably I have found evidence that the lens was never used for this purpose, and indeed could not have been. Instead the lens appears to have been used to film extreme wide angle shots of sets in the film.

The Fairchild-Curtis and 2001

We know that the Fairchild-Curtis lens was used to film 2001 for two key reasons, hearsay aside. First, there are continuity reports which specifically identify the lens as having been used to film certain scenes. And second, there are photos of the lens actually being used on set.

Here's one of the continuity reports, which describes the lens being used in the centrifuge set. Note that it indicates that the 160° lens was used with a Todd-AO 65mm camera, which would definitely have annoyed Panavision!

And here we see a camera with the Fairchild-Curtis lens on it. Note how the camera is on the "rubber tyred dolly" (UK spelling) mentioned above, and how the camera is getting ready to film a rotating centrifuge. Note also that the camera is marked “T-AO”, indicating it's a Todd-AO camera and not a Panavision camera. The six silver squares on the magazine/reel housing are additional tells. Finally, it isn't a blimped camera (housed in a noise-reducing enclosure), indicating that the footage they were filming would not include location sound. Any dialogue or other sounds meant to be audible to the actors would have to be recorded later.

Now remember that the HAL point of view shots were never taken from that position. If this lens had been used for a HAL POV shot, then HAL would have had to be floating along near the centrifuge floor.

However, I don't believe that the footage described above was actually used in the final film. The finished movie does have a shot of the centrifuge after HAL has killed the sleeping crew, but that view of the hibernacula is not a rotating sequence. The resulting stationary shot makes them look even more coffin-like and grim.

I don't have a list of all the scenes in the film, so I can't tell what scene 203 (see the clapperboard above) was. But the Fairchild-Curtis lens was used to film an astronaut moving around inside the ring, such as the shots of Bowman walking around with his sketchpad, or Poole jogging and shadowboxing.

It's possible, then, that this setup was used to film this scene of Poole exercising. Note how the lens is aligned to the centrifuge rim, and how it’s very low down – you can see the underside of the tables. Both camera and actor Gary Lockwood would have been at the bottom of the centrifuge at this point.

The Fairchild-Curtis lens was also definitely used for shots in the pod bay. In fact, I believe the majority of the wide angle shots of the bay could have been taken with the lens. You can definitely see how the lens used to film those shots wasn't fully rectilinear. In other words, it didn't fully correct the appearance of straight lines as they neared the edges of the frame - there's a lot of barrel curvature. But the lens is also not as distorted as the image from a full-frame fisheye.

It was also used EVA pod cockpit shots, possibly the iconic shot looking down showing the pilot sitting before the glowing banks of buttons and displays. There are photos in the Kubrick Archives of a camera being set up for the full-sized EVA pod cockpit set, and the Fairchild-Curtis lens is clearly being used.

It's worth noting that the usual ultra-wide angle lens for the Todd-AO was the 12.85mm 128° “bug-eye” lens. But you can tell that it wasn't this lens in use in these behind the scenes photos, because the Todd-AO lens has a low rim around the primary element, and no white text on the front the way the Fairchild-Curtis lens does. Also, there is a behind the scenes photo, taken by set photographer John Jay, of a 65mm Todd-AO camera (its reel housings have six characteristic silver squares around the perimeter, and it has “2 T-AO” on the front) with a large lens bearing a lenscap inscribed “160° CINERAMA F-2”.

The Fairchild-Curtis and the evidence for HAL POV

In an LA Times article by Robert W. Welkos on the same subject as this whole writeup, Kirk Wooster, the private collector who once owned the Fairchild-Curtis lens, claimed that it was used for filming the HAL’s point of view sequences.

The sole written material in favour of this argument, that I'm aware of anyway, is a one-sentence image caption on page 444 of the June 1968 issue of American Cinematographer magazine. This states that, "Subjective point-of-view from inside the computer was achieved with Fairchild “bugeye” type extreme wide-angle lens covering a field of almost 180 degrees."

Unfortunately the immediately preceding sentence reads, "A computer's-eye view of an astronaut presenting artwork for inspection", but the photograph that this caption is attached to is of Bowman walking through the centrifuge tunnel carrying his data pad, and not a fisheye lens shot at all. And of course the Fairchild-Curtis covered 160°, not almost 180°. Other photos in the article are printed upside-down. So there may be accuracy issues with this secondary evidence here.

The overall article was written by Herb A. Lightman, based on interviews with Kubrick and others, and it's definitely a valuable source of key information on the making of the film, but it's not known who wrote the caption and where that information originated.

That said, is it possible to adapt a wide-angle lens to simulate a circular fisheye? Yes, it is. You could attach additional optics to create a simulated circular fisheye effect. Perhaps that's what was done.

and the evidence against

First, according to Douglas Trumbull in the same LA Times article, the massive Fairchild-Curtis lens was not a circular lens that projected a round image onto the frame – it was a full-frame lens that covered the whole image area. He's specifically quoted as saying, "Any claim by Kirk [Wooster, the private collector] that the Fairchild-Curtis lens was used for the HAL [point-of-view] shots is just not true." If he was correct and his comments have been paraphrased accurately, then the Fairchild-Curtis HAL POV theory is wrong.

It's true that Trumbull did not work on the live-action filming of 2001, but he was a very capable and knowledgeable special effects expert. He also worked on To the Moon and Beyond, the New York World's Fair film that used the same lens. If he stated for a fact that the Fairchild-Curtis lens was not a circular fisheye, then I think that's pretty convincing.

Another data point is a comment allegedly by cinematographer Jim Dickson. He wrote to a Wordpress blog:

This Fairchild lens seen in the glass case is the lens I used to photograph the film “To the Moon and Beyond” in 1963 for Graphic Films,Cinerama and the New York Worlds Fair , it was used later at MGM in London to photograph the interior of Hal the computer in the film “2001” for Stanley Kubrick. It was not used as the subject “Hal ” seen talking in the dialog sequences with the two astronauts. that lens was a newer looking 8mm Nikkor fisheye. I was there as I photographed the Stargate Slitscan sequence and other items for Trumbull and Kubrick.

If that post was actually by Jim Dickson, and I see no reason why it wasn’t, that’s very interesting indeed. When he said "the interior of Hal the computer" he must mean shots of the brain room set, which would have necessitated a very wide angle lens to take in fully. Unfortunately he apparently died in 2020, so we can't ask him for details on this lens.

That said, is it possible to adapt a wide-angle lens to simulate a circular fisheye? Yes, it is. You could attach a lens extension tube between the lens and the camera to do just that. And that would create a simulated circular fisheye effect. However, this process would not increase the field of view of the lens. It would simply result in a round image on the film, which is not the same thing.

Second, we have primary evidence in the form of continuity reports and behind the scenes photos showing that the Fairchild-Curtis was used on-set to film 2001, but in contexts that are not compatible with any known HAL POV shots.

For example, the Fairchild-Curtis is known to have been used on the rubber-wheeled cart to shoot centrifuge scenes. These shots are extreme wide-angle full-frame shots in the film. They are not circular fisheye shots.

Since the lens did not have an adjustable field of view (it wasn't an extreme wide-angle zoom, which weren't invented until years later anyway) it would have been impossible to film both full-frame shots and circular fisheye shots with the same lens. Therefore the Fairchild-Curtis was likely not used to film circular fisheye shots.

Finally, there's at least one photograph of a Todd-AO camera, with the Fairchild-Curtis lens fitted, in the alien hotel. There would be no reason to haul this setup on-set for a HAL point of view shot, because obviously HAL never left the Discovery. This is the closest we have to a smoking gun, in my view.

Given these pieces of information, and given that the sole argument in favour of the theory is questionable, I believe it's improbable that the Fairchild-Curtis lens was the circular fisheye lens used to film the HAL POV sequences.

But the fact that the lens is unlikely to have been used to shoot HAL's viewpoint doesn't, of course, make its historical value any less. It's a unique lens that was used to film a lot of key moments in one of the most remarkable films ever shot.

Finally, this whole question could be answered relatively easily with access to the lens. If the lens were attached to a 65mm camera it would be possible to put a ground glass inside the camera at the focal plane to see whether or not the lens projected a full-frame semi-rectilinear image (as I believe it does) or whether it projected a circular fisheye image, cut off on the sides (as would have been needed for the HAL POV shots). A test like that would answer the question definitively.

What about the 8mm Nikkor?

So. Is it possible that the Nikkor 8mm f/8 still photography lens, which was used as the prop HAL 9000 camera eye, could actually have been used to capture the footage used for HAL POV shots as well?

For a long time it was not generally believed that the static prop lens was used for actual filmmaking. After all, both the optical and mechanical properties of the Nikkor lens were matched specifically to 35mm still photography, and weren’t designed to work with the 65mm cameras used to make the movie.

Unfortunately, we have no proverbial smoking guns at this point - no photos of the Nikkor lens in use to film a scene, and no continuity report listing the lens. We also don't have access to the wisdom of the filmmakers themselves - Kubrick and his cinematography staff are no longer with us. Douglas Trumbull sadly died in 2022.

What we have is circumstantial evidence. To start, let's consider if there are any technical reasons why the Nikkor lens couldn't have been used.

Is the image area compatible?

This is key. Would the image area projected by the Nikkor lens actually match the area of 65mm film? If the Nikkor's projected image area is too big or too small for 65mm film, then the lens obviously could not have been used. Well, let's take a look.

So, purely in terms of image area, yes. 35mm still imaging film and 65mm spherical movie film are almost compatible on the vertical axis, which is what matters here. In fact, the Nikon lens would be slightly cut off at the top and bottom, much as it appears in the film - its image area didn't quite fit 35mm still film, and was about 4% too big for spherical 65mm film.

Since 2001 was shot spherical and not anamorphic, the lens image from the Nikkor would work – a square would appear as a square.

Note that this doesn't mean that any 35mm still camera lens could be used with 65mm film, because the latter's image area is much wider than the image circle of the lens. But since we're only considering the case of a circular fisheye lens here, the horizontal axis width doesn't really matter!

Could a Nikkor lens be physically mounted on a Panavision or Todd-AO camera?

Out of the box, no. Nikon's “F” lens mount isn't compatible with any motion picture lens mount.

But look back at the comment by Douglas Trumbull - he mentioned that Kubrick had experimented with lens adapters to get Nikkor lenses to fit his movie cameras. Although Trumbull was not a cinematographer on the film, he did work in many areas of special effects in the film, and of course developed a reputation for industry-leading technical abilities, so his word counts for a lot.

Is there adequate back focus clearance?

The Nikkor 8mm f/8 lens is an early optical design for 35mm still photography wide-angle lenses. And it's super annoying to use, because it has this long tube sticking out the back, which houses the rear element. This is a problem for any SLR, since the moving mirror will collide with this part. So you have to physically lock the mirror up to use it on a Nikon F camera. Later fisheye lenses are built on the "retrofocus" principle, which makes it possible to have increased distance between the back of the lens and the film/sensor plane.

But what about movie cameras? Is the distance from the lens mount (lens flange) to either the gate or the surface of the film (focal plane) adequate to accommodate such a weird lens? Well, I'm no expert, but it seems the lens might fit. According to this table, the flange focal distance of a Panavision 65mm is 2.030", which would clear just fine. The Todd-AO FC camera was apparently 1.905", which would also be OK.

Could you focus the Nikkor lens on a movie camera?

The Nikkor lens was designed for still photography, and so does not have a follow focus gear ring the way cinema lenses do. But it doesn't actually matter, since the Nikkor 8mm f/8 has no focus capabilities! It's permanently fixed at infinity focus, and since its depth of field is so large only things really close to the lens will be unacceptably out of focus. And follow-focus would be unnecessary anyway since all the HAL POV shots were static.

Could you compose scenes visually?

You wouldn't have been able to compose scenes through the viewfinder unless the movie camera in question were a reflex (mirror-equipped design), and I don't think Super Panavision cameras were in the 1960s. And in the case of the Nikkor 8mm f/8, which has a long tube sticking out the back of the lens, a mirror couldn't have been used anyway.

So you'd have to point, roll film, and check the rushes (dailies, in US parlance) the next day to see what worked.

Is the quality adequate?

As I noted at the start of this article, the image quality of the lens used in the HAL POV scenes is not brilliant. It definitely looks shot wide open, and the fastest speed (ie: widest aperture) attainable by the Nikkor 8mm f/8 is, well, f/8. That's not great. The other three f-stop settings on that lens are f/11, f/16, and f/22. So shooting wide open would make sense, even at the cost of the optical aberrations visible in the POV scenes. The lens also wasn't multicoated.

So - was it the Nikkor?

Basically there are no technical reasons that I can see why the Nikkor 8mm f/8 lens couldn't have been used to film the HAL POV scenes, assuming Trumbull's assertion that the 2001 team experimented with adapters to fit Nikkor lenses to their 65mm cameras is correct.

There is also the fact that the 8mm's image circle would closely match what we see in the movies - a circular image that's cut off at the top and bottom.

However, no primary evidence has yet come to light confirming the Nikkor theory.

What about add-on lenses?

Still photographers operating on a budget often fasten supplementary lenses to get a fisheye effect. These fisheye adapters screw onto the filter threads of a longer focal length lens, converting it. They tend to be of extremely low optical quality, of course.

Cinematographers also use add-on adapters, though of much higher quality. So as a point of conjecture, perhaps the 2001 team added a supplemental lens, such as a Schneider-Kreuznach Century adapter, over one of their wide-angle lenses to achieve this circular fisheye effect.

Another idea is that the POV shots were actually filmed on 35mm stock using a different lens altogether, then blown up to 70mm. This seems extremely unlikely, as the image quality would be just terrible. And, while the lens used in the POV shots is clearly quite soft, the film grain does not change for those scenes, invalidating the 35mm theory.

These are possibilities, I suppose, but there's really no evidence for either theory.

Point of view errors

There are some occasions where we are shown HAL’s view of a given space. However, in three of those views the HAL view does not correspond to the actual position of the known lens. For example, during the pod bay bench test scene the HAL view doesn’t make optical sense to the position of the faceplate! The HAL lens is looking out towards the pod bay doors, but the POV view is rotated 90 degrees to HAL’s left/camera right. That would only be possible if the camera and lens were positioned in front of the faceplate, and turned.

I suppose Kubrick is showing the audience that HAL can see 180° around the room, even if the visual portrayal of such doesn’t always correspond to reality. Just like the astronauts have images projected on their faces in the EVA pod makes no sense – Kubrick knows it doesn’t, but he’s going for something that looks good and also carries a certain visual symbolism.

There’s a related issue with the HAL view of Bowman entering the brain chamber. In fact, the shot below is particularly problematic since only one HAL faceplate is shown in the whole brain chamber, and you can see it in the shot. But the shot is supposed to be a reflection! So that would only make logical and optical sense if an invisible camera were shooting a lens positioned next to the panel with six video monitors on it.

Another thing of note is that the shots where scenes are visible reflected in HAL’s glowing eye, like the one above, are actually masked double exposures. You certainly don’t see the reflection of the camera in the eye.

The other views

One final thing is that the HAL point of view is shown two different ways in the film. There's the subjective point of view, shot with a circular fisheye lens, that's supposed to be showing what HAL is actually seeing. Then there are a couple of scenes which are double-exposed as above and below, and which represent an exterior view of HAL as he gazes back towards the viewer.

Now it's entirely possible that the shot of the brain room above, with the camera looking up, was done using the Fairchild-Curtis lens. That shot was matted in a circle and composited over top of a view of the HAL lens shot separately. Conjecture: perhaps that's where the confusion over the Fairchild-Curtis lens and HAL's POV arose. After all, it is a shot of HAL's eye, and part of it was shot using the Fairchild-Curtis. Just not one of the true subjective POV shots.

Conclusion

There is no hard evidence thus far, in the form of continuity reports or otherwise, documenting what specific lens was used to film the HAL POV shots. I've looked through stacks of these reports at the Kubrick Archives and have come up empty-handed, but I didn't have time to look at everything, of course.

Nonetheless, it appears that the notion that the Fairchild-Curtis 160° f/2.0 ultra wide-angle lens was used for filming the HAL POV sequences may be incorrect. First, 2001 effects man and legendary filmmaker Douglas Trumbull specifically told the LA Times that it wasn't. And second, we have photos of the lens being used to film interior sets in a fashion that is not consistent with a HAL POV shot. The only way I believe the lens could have been used for the circular fisheye effect is if had been used with additional optics.

The collector who purchased the lens and sold it to the Academy Museum appears to have promoted the idea that it was used for the HAL POV shots, but I would argue that that assertion needs to be backed up by more evidence.

Given the information before us today, I'd say that the Nikkor 8mm f/8 lens is the most likely candidate for the lens used by production on the HAL POV shots, but this is not definitive.

IV:  The Different HALs

VI:  Replicating HAL

Contact

© 2021-2023 NK Guy, contact@theageofplastic.info