The DaVinci Surgery System is an engineering marvel – and now in full 3D! (excuse the Hollywood like overtone). The purpose of this article is not to explain 3D surgery, but to encourage debate on why current 3D movie making is made to look as complex as though it were surgery.

Current 3D camera rigs used in movies and the camera crew look much like the scenario above with a 3D viewing station – needed in both cases, and a bulky “rig” (arguably, not needed in the movie making scenario). In the DaVinci scenario, a remote operator – The surgeon or, in Hollywood terms – the Director… is also the Cameraman, focus puller, DP and “actor”. One person doing the job of guiding the 3D camera to give him/herself the sense of “immersion” that only 3D can give, along with manipulating focus, guiding the camera AND the robot arms remotely – to perform the operation (acting)

In the case of making 3D movies, we have 3D Camera rigs that look like monsters and weigh in excess of 50 kgs in most cases, are intimidating, and look like they can only be used on productions that have millions of dollars in budgets. This may well be one of the reasons that stalwarts in the movie business prefer them, so as to keep the indie movie makers out of the game, while they cash in on the 3D craze. (This assumption is not proven scientifically!)

Consider this – there is a huge monstrosity of metal masquerading as a camera that calls for more than one person to operate efficiently. Thus we have Director, DP, focus puller, Convergence puller and sometimes an assorted array of assistants thrown in to capture a scene in stereoscopic 3D! Does it have to be this way to make a 3D movie in this day and age of high technology, miniaturization and falling prices?

Believe it or not the cameras shown above are capable of creating images that can be shown in full fidelity at IMAX screen sizes. These are the cameras and the configurations that should be used today when creating a 3D Camera rig. Imagine the the maneuverability possible with these rigs, and thus the elimination of one of the biggest excuses that lazy Cinematographers use to not shoot in 3D, or much worse, the justification of using 2D to 3D conversions. While every movie maker no doubt has a preference of Camera, we have to remember that it’s not the tool that counts, it’s how well it’s used to tell the story.

Beam-Splitters versus Side-By-Side rigs:

When shooting 3D, it is taken for granted that the “ideal” distance between the center of the two cameras lenses (interaxial) should be about the same as the distance between human eyes (interocular). While this is not strictly needed it has been generally agreed on. There are however, times such as shooting extreme close-ups or for macro-photography that the interaxial of the cameras needs to be much smaller. This would not be possible with most professional cameras due to their physical size limitations and lenses when placed side by side. Thus the need for a “beamsplitter” a two way mirror that allows one camera to shoot through an angled mirror, and the second camera mounted above or below that shoots the reflection of the same scene. Precise and minute adjustments are possible thereby allowing for an interaxial of zero to a much larger distance usually measure in inches.

This is the reason for the bulky looking “3D Cameras”. It does allow a movie-maker to use any camera pair, so it is not un-common to see larger cameras mounted along with the associated accessories for syncing the cameras, battery packs, motorized lens focus and sync etc.

However, what is important to note is this: In a typical movie, it is very rare to see interaxial or interocular separation of more than the human eye separation needed. With cameras that shoot raw such as the Si2K mini 3D system shown above and the soon to be released cameras from others, it really is not that valid an excuse to say that these cameras don’t have enough “latitude” etc.

It is also not uncommon to mix footage shot with different cameras when doing a 2D movie, so in the case of a 3D movie, a parallel side by side rig…i.e Two “professional cameras of choice” for the Cinematographer placed side by side and shooting parallel can be used for establishing Master Shots or land-scape shots where interaxials of more than 4 to 6 inches are a necessity for any 3D depth to be visible, and then usage of lightweight and easily maneuverable 3D rigs for the customary car chase and stunt scenes or for hand-held shots.

The one advantage of beam-splitter rigs has been mentioned above. The disadvantages can run into a few pages of explanation, but a brief summary of them are:

Disadvantages of Beam Splitter 3D rigs:

• Shooting through a 2 way mirror leads to loss of light (f stop) on one of the captured image
• Dust on the larger reflective surface
• Polarization of light issues when shooting reflective surfaces such as water bodies – both images will not match
• To overcome the above, polarizers are sometimes used on both cameras – this leads to more light loss
• Bulky and weight issues

One thought on Closeups in 3D: In a typical movie (3D) how many shots can you recall where there is a need for such closeups camerawork, that need interaxials where the nearest human talent / object is less than 10 feet without then violating other things like stereo windows, cropped heads etc. (all of which ruin the stereo 3d illusion?)

There are tricks to get decent close-ups in a “narrative” 3D movie with a suitable parallel rig- this is where an experienced Stereographer on-set is invaluable.

See Creative Scene Blocking for 3D movies. Software such as Nuke’s Interaxial Shifter and SGOs Mistika can in these cases recreate or re-compute interaxial in post, if an NLE’s simple horizontal image translation (H.I.T) does not solve the issue.

The RealFlex Secret – Do you really need a Beamsplitter?:

At Real Vision, we have been producing content for Autostereoscopic 3D screens since 2004. Autostereoscopic (no glasses needed) screens, need between 8 to 9 cameras either CGI (easy) or real – not easy at all! to create content to be displayed on these screens. We had to overcome these obstacles and thus have been “reshaping 3D volume” or creating virtual in-between cameras with in-house tools.

This secret can now be adapted to conventional stereoscopy, and allow for shooting with a standard parallel 3D rig. The 3D volume can then be re-shaped in post using the RealFlex method.

Of Convergence Pulling and Focus Pulling in 3D:

There is now a much debated issue in shooting 3D movies. The topic of Convergence or “Toe-in” of the cameras. Well known Directors such as James Cameron swear by converging on the area of interest or subject of importance in a scene, so much so that he has a custom camera rig to “follow” convergence to his region of choice, usually the subject of the scene. The reasoning is that he wants the camera to do the “hard work” of converging that our eyes normally do… so as to allow the audience to concentrate on the story being told.

Toe in on Camera = Trouble:

There are two camps on this – Those that agree and those that don’t. For those that agree, there is now a demand for a “convergence puller” on set – someone who either manually or via motorized control, toes-in the cameras so that the target subject of the scene is either at zero parallax (at screen depth) or is in somewhat shallow depth in or out of the screen.

The problems with this is – that as you converge on an area of interest (while also “pulling focus”), everything in the fore-ground AND the background in that scene goes out of focus and is blurry. There is another undesirable side-effect of this practice – Keystone errors.  Fortunately this error can be corrected by post production software such as NUKE, and there is expensive hardware that can also correct for this error in real-time for live shoots.

The biggest, most glaring error that is better explained in more detail is:  Semi-Cylindrical projection of the captured 3D space. Focus Pulling in 3D is another issue and is explained in a previous article.

** Update** Andrew Woods has a nice paper here on dealing with Stereoscopic Distortions. There is also an App called 3D Map to download for further detailed investigation.

While it may not be in every Movie makers interest to recreate the real world faithfully, when captured by a stereoscopic 3D camera, for those who want to add that much bit more to the so-called “immersive” or “suspension-of-disbelief” experience in their 3D movies, knowing a bit more on this subject is never a wasted exercise.

2D to 3D converted movies- good enough?

There is no doubt that due to the current craze and resurrection of 3D in Cinemas, every studio is working overtime to capitalize on it. From Cinematographers who are too lazy to learn how to tell a story in 3D and use it effectively, to post production houses who see a lucrative business in charging sums of around $100,000 per minute to “convert” a 2D movie to 3D. This tempts everyone to “think” in 2D, create a movie with that same 2D “thinking” and then expect it to cross-over as an excellent 3D movie.

This is the reason why the latest movies show some of the ill-effects of that mentality. Movies like Alice in Wonderland show how camera framing was done with 2D in mind, thereby cutting off the heads of characters and ruining the intimacy and sense of immersion and emotion engineering that would have been possible in some of the close up scenes with Alice and her father. Other scenes show a big tree “mapped” onto a “sphere” that gives it an un-natural look as Alice and her mother alight from the coach to the palace. The ‘crowds’ at the palace abruptly go “flat” due to non “conversion”.  The white flowers in the garden are all out of focus – something that can lead to headaches as people try in vain to “fuse” these blurry images with their eyes…

Some errors in older converted movies in a PDF file here.

… more recently one movie has received scathing reviews from the Hollywood community itself for the bad and rushed 2D to 3D conversion job done.

Is 3D a tool or a new Story telling Medium?

To make another analogy of why Shooting 3D movies should not be as complex as Open Heart Surgery, we should look at the case-study of a household name that had it’s beginnings as a cable tv company – CNN. They did not have resources or budgets like the established broadcasters. They just told the story on time and effectively, without the overhead of cumbersome heavyweight gear or established “practices”. Arguably it was CNN that pioneered the use of cell-phone like video quality when reporting live-in-the-field, without worrying that their cameras may not have enough resolution and pixel count or “latitude” etc.

This is not to say that a prosumer 3D rig should be used for producing a 3D movie, it is only to say that there is already satisfactory equipment available today to remove many of the obstacles that new movie makers may believe are a barrier to entry. One advantage that the fledgling 3D cinematographer has is the ability to learn how to “think” in 3D, and should thus investigate new methods of storytelling using stereoscopic 3d as a story telling medium.