Originally published in FCPUG Supermag #4 (April 2010)

It’s 2010 and 3D back in style, again, for the fourth time in 150 years.   Will its popularity last this time?   If the actions of TV manufacturers like JVC, Sony, Panasonic and Samsung are any indication the answer is a resounding “yes!”

Obviously Hollywood has also jumped on the bandwagon and the overwhelming success of recent 3D films like Avatar have proven 3D is a money-maker.

So how can you get involved by shooting your own 3D content?  It’s actually quite easy to get started and learn the basics of stereoscopic 3D photography.  You won’t be able to sell yourself as a stereographer after reading this beginner’s guide (it literally takes years to learn all the aspects of shooting and build the experience to shoot good stereoscopic 3D) but I guarantee you will have some fun and impress your friends.

The basic principle behind shooting stereoscopic 3D is to capture and then present two slightly different points of view and let the viewer’s brain determine stereoscopic depth.  It sounds simple enough but the first thing any budding stereographer should learn is some basic stereoscopic terminology.  These few terms may seem daunting at first but they will form the basis of your stereoscopic knowledge.

Terminology

Stereoscopic 3D a.k.a. “Stereo3D,” “S-3D,” or “S3D”
“3D” means different things to different people.  In the world of visual effects it primarily refers to CGI modeling.   For example, every Pixar film ever made was created through the use of “3D” animation techniques, but only “Up” and the re-releases of Toy Story 1&2 were also presented in stereoscopic 3D.

This is why stereographers refer to the craft specifically as “stereoscopic 3D” or simply “S3D” to differentiate it from 3D CGI.

Interaxial (a.k.a. “Stereo Base”) & Interocular (a.k.a. “i.o.”) separation
The interocular separation technically  refers to the distance between the centers of the human eyes.  This distance is typically accepted to be an average of 65mm or roughly 2.5 inches. Interaxial separation is the distance between the centers of two camera lenses. The human interocular separation is an important constant stereographers use to make calculations for interaxial separation.

Beware that Interaxial separation is often incorrectly referred to as “Interocular” and vise-versa.  In the professional world of  stereoscopic cinema it has become the norm to refer to interaxial separation as “i.o.”

Stereo Window and Screen Plane
Simply put, the “Stereo Window” refers to the physical display surface. You will be able to visualize the concept if you think of the HD screen as a real window that allows you to view the outside world.  Objects in your stereoscopic scene can be behind the window, at the window (the Screen Plane,) or in front of the window.   Most of the S3D in Avatar happened behind or on the stereo window and there were only a few “off-screen” gags.

Convergence, Divergence & Binocular Vision
Binocular Vision simply means that two eyes are used in the vision system. Binocular Vision and Convergence are the primary tools mammals use to perceive depth at close range (up to 300 feet for humans.)  The wider an animal’s eyes are apart (its interocular distance) the deeper its binocular depth perception or “depth range.”

At greater distances we start to use monocular depth cues like perspective, relative size, occlusion, shadows and relation to horizon to perceive how far away objects are from us.

Here’s an example of how your binocular vision works.  Hold a pen about one foot in front of your face and look at it.  You will feel your eyes both angle towards the pen in order to converge on it, creating a single image of the pen.  What you may not notice is that everything behind the pen appears as a double image (diverged.)  Now look at the wall straight ahead and the pen will appear as two pens because your eyes are no longer converged on it.  This “double-image” is known as a retinal disparity and it is the distance between these two images (horizontal parallax) that helps our brain determine how far away an object is.

In terms of stereoscopic projection when we “converge” on an object (either through angling the cameras towards the object or shifting the two images in post) that object appears to sit on the screen plane.  Any “double-images” resulting from unconverged areas of the scene are perceived to be at a different distance (in front or behind) than the converged object.  We refer to objects as having “positive” (background), “negative” (foreground) or “zero” (screen plane) parallax.  Human eyes are not meant to diverge more than parallel so it is the stereographer’s responsibility to prevent excessive positive parallax. See the sidenote on the 1/30 rule Native Pixel Parallax later in this article.

This is the basic principle behind stereoscopic shooting and emulating human binocular vision with two cameras.  However, be aware that a large percentage of the population (some estimates are 15%) suffer from “stereo-blindness” and cannot perceive depth through strictly horizontal parallax cues.

Disparity
Disparity is a “dirty word” for stereographers.  In fact the only “good” type of disparity in S3D is horizontal disparity between the left and right eye images.  As mentioned before this is known as horizontal parallax.

Any other type of disparity in your image (vertical, rotational, zoom, keystone or temporal) will cause the viewers eyes to strain to accommodate. This can break the 3D effect and cause muscular pain in the viewer’s eyes.  Every stereographer will strive to avoid these disparities on set or use special software in post-production to correct it.

Ortho-stereo, Hyper-stereo & Hypo-stereo
I already mentioned that the average interocular of humans is 65mm (2.5 inches.)  When this same distance is used as the interaxial distance between two shooting cameras then the resulting stereoscopic effect is known as “Ortho-stereo.”  Many stereographers choose 2.5” as a stereo-base for this reason.  If the interaxial distance used to shoot is smaller than 2.5 inches then you are shooting “Hypo-stereo.”  This technique is common for theatrically released films to accommodate the effects of the big screen.  It is also used for macro stereoscopic photography.

Lastly, Hyper-stereo refers to interaxial distances greater than 2.5 inches.  As I mentioned earlier the greater the interaxial separation, the greater the depth effect.   An elephant can perceive much more depth than a human, and a human can perceive more depth than a mouse.  However, using this same analogy, the mouse can get close and peer inside the petals of a flower with very good depth perception, and the human will just go “cross-eyed.”   Therefore decreasing the interaxial separation between two cameras to 1” or less will allow you to shoot amazing macro stereo-photos and separating the cameras to several feet apart will allow great depth on mountain ranges, city skylines and other vistas.

The trouble with using hyper-stereo is that scenes with gigantic objects in real-life may appear as small models.  This phenomenon is known as dwarfism.  The opposite happens with hypo-stereo, where normal sized objects appear gigantic. (gigantism.)   For the rest of this beginner’s guide we will ignore the effects of dwarfism and gigantism and instead focus on just creating 3D depth.

Passive Polarization, Active Shutter Glasses, Anaglyph & Autostereo
There are three basic types of glasses used for presenting stereoscopic 3D material.  In most of the Real-D and Imax theatres in North America the common method is passive polarized glasses with either circular or linear polarizers.  Most of the consumer 3DTVs on the market use some form of active shutter glasses to flicker the left and right images on and off at 120Hz.   Autostereoscopic displays use lenticular lenses or parallel barrier technologies to present stereoscopic material without the use of glasses.

Anaglyph glasses will work with almost any display but use color filters to separate the left and right images.   The most common configurations are red/cyan, blue/amber, and green/magenta.  Dolby 3D uses a form of spectral color separation to filter the left and right eyes.

Now that you are familiar with some of the basic stereoscopic terminology it’s time to select your gear.

Gear

Camera and stereo rig selection
In order to create a stereoscopic image we will need to photograph two images from slightly different perspectives.  We can use a single camera for still-life stereo-photography or two cameras for motion-picture stereo-cinema.  The simplest way to mount one or two cameras is with a “slider bar.”   For example, a single camera can take one photo for the left eye perspective, and then simply be slid down the bar to take a photo from the right eye perspective.   When shooting with two cameras they can be mounted side by side.

Interaxial separation is an important factor when shooting S3D so therefore the width of your two cameras will determine the minimum interaxial separation.

If your cameras are too wide and a smaller interaxial separation is required then a beam-splitter rig can be employed.   Beam-splitters use a 50/50 mirror (similar to teleprompter glass) that allows one camera to shoot through the glass and then other to shoot the reflection.   The interaxial can be brought down to as little as 0mm with beamsplitter rigs.

Beamsplitter rigs are expensive, and since most beginner stereographers will be working with a limited budget and a slider bar, selection of “thin” cameras will benefit the ability to achieve small interaxial separation.

Special Stereoscopic Lenses
There are special stereoscopic lenses on the market designed for various digital SLR cameras.  These lenses will work with a single camera but capture a left and right point of view in the same frame.   The concept is intriguing but the lenses are very slow (F/11 – F/22) and the aspect ratio is vertically oriented.

Purpose-built Stereoscopic cameras
Stereoscopic film cameras have existed for decades.   I personally own a Kodak Stereo camera from the early 50’s that I’ve shot hundreds of 3D slides with. Recently manufacturers like Fujifilm and Panasonic have recognized the demand for digital versions of these cameras and released new products to market.

Genlock capability
If you plan to shoot stereoscopic video with any action then it will be beneficial to use two cameras that can be genlocked together.  Cameras that cannot be genlocked will have some degree of temporal disparity.  However using the highest frame rate available (720p60 for example) will reduce the chance of detrimental temporal disparity.  There are also some devices capable of synchronizing cameras that use LANC controllers.

Interlace vs. Progressive
Every frame of interlaced video inheritably will have some degree of temporal disparity between the fields.  It is recommended to shoot with progressive formats whenever possible.

Lens & Focal Length selection
Wider lenses will be easier to shoot with for the beginner and will also lend more “dimensionality” to your subjects.   Telephoto lenses will compress your subjects flat so they appear as cardboard cutouts.   Stay away from “fisheye” lenses because the distortion will cause many geometrical disparities.

OK, so you’ve learned your terminology and selected your gear.   Now what?  It’s time to get out there and shoot.   We haven’t discussed the various calculations or the rules of S3D but I encourage you to shoot now so you can learn from your mistakes.

The Math & The Rules
Over the years I’ve read and re-read every book I could find on stereoscopic photography.  Within those books I have found and attempted to understand a myriad of different stereoscopic equations created by really smart people with names like Rule, Bercovitz, Davis, Lipton, Levonian, MacAdam, Wattie, DiMarzio, Boltjanski, Komar and Ovsjannikova.  Please feel free to google them.

I’m not even going to attempt to explain all the intricacies of the various algebraic equations available to calculate things like depth range and depth budget and instead introduce you to the only calculation beginners should concern themselves with: the 1/30 rule.

The 1/30 Rule
The 1/30 rule refers to a commonly accepted rule used by hobbyist stereographers around the world.  It basically says that the interaxial separation (stereo-base) should only be 1/30^th of the distance from your camera to the closest subject.  In the case of ortho-stereoscopic shooting that would mean your cameras should only be 2.5” apart and your closest subject should never be any closer than 75 inches (about 6 feet) away.

Interaxial x 30 = minimum object distance

or

Minimum object distance ÷ 30 = Interaxial

If you are using a couple DSLR cameras that are 6 inches wide the calculation would look like: 6 x 30 = 180 inches or 15 feet.  That’s right… 15 feet!

Optional sidenote: Does the 1/30 rule apply to all scenarios?
No, the 1/30 rule doesn’t apply to all scenarios.  In fact, in feature film production destined for the big screen we will typically use a ratio of 1/60 or 1/100.  The 1/30 rule works well if your final display screen size is less than 75 inches wide, your cameras were parallel to each other, and your shots were all taken outside with the background at infinity.  When you are ready to take the next step to becoming a stereographer you will need to learn about depth range and the various equations available to calculate maximum positive parallax (the parallax of the furthest object,) which will translate into a real-world distance when you eventually display your footage.   Remember that photo of the eyes pointing outward (diverging)?  Well it isn’t natural for humans to diverge and therefore the maximum positive parallax when displayed should not exceed the human interocular of 2.5 inches (65mm.)   In the post section I will briefly cover how you can readjust the convergence point bring the maximum positive parallax within the limits of the native display parallax (2.5 inches.)

Don’t Break the Stereo Window
We discussed briefly how the display screen represents a window and objects can be behind, at or in front of the window.  If you want an object to appear in front of the window it cannot touch the edge of the frame.  If it does the viewer’s brain won’t understand how the parallax is suggesting the object is in front of the screen, but at the same time it is being occluded by the edge of the screen.  When this contradiction happens it is referred to as a window violation and it should be avoided.   Professional stereographers have a few tricks for fixing window violations with lighting or soft masks but it is best for beginners to simply obey this rule.

Turn off Image Stabilization
If you are using video cameras with image stabilization you must turn the feature off or the camera’s optical axis will move independent of each other in unpredictable ways.   As you can imagine this will make it impossible to tune out disparities.

Use identical settings on both cameras
It is very important to use the same type of camera, same type of lens and exactly the same camera settings (white balance, shutter speed, aperture, frame rate, resolution, zoom, codec, etc.) on both cameras.   Any differences will cause a disparity.  It is also a good idea to use manual focus and set it to the hyperfocal distance or a suitable distance with a deep depth of field.

Use a clapper or synchronize timecode
If your cameras are capable of genlock and TC slave then by all means use those features to maintain synchronization.  If you are using consumer level cameras it will be up to you to synchronize the shots in post.  In either case you should use a slate with a clapper to identify the shot/takes and easily synch them.

If your cameras have an IR remote start/stop it is handy to use one remote to roll & cut both cameras simultaneously.   If you are shooting stills with DSLRs there are ways to connect the cameras with an electronic cable release for synchronized shutters.

Slow down your pans
However fast you are used to panning in 2D, cut the speed in half for 3D.  If you are shooting in interlace then cut the speed in half again.   Better yet, avoid pans altogether unless your cameras are genlocked.  Whip pans should be OK with genlocked cameras.

Label your media “Left” and “Right”
This might seem like a simple rule to remember but the truth is that most instances of inverted 3D is a result of a mislabeled tape or clip.  Good logging and management of clips is essential with stereoscopic post production.

To Converge or Not Converge… That is the question.
One of the most debated topics among stereographers is whether to “toe-in” the cameras to converge on your subject or simply mount the cameras perfectly parallel and set convergence in post-production.  Converging while shooting requires more time during production but one would hope less time in production.  However “toeing-in” can also create keystoning issues that need to be repaired later.   My personal mantra is to always shoot perfectly parallel and I recommend the same for the budding stereographer.

Post
So you’ve shot your footage and now you want to edit and watch it.  For Final Cut Studio there are basically two options for editing and mastering: Cineform’s Neo3D and Dashwood’s Stereo3D Toolbox.

Neo3D
To use Neo3D first transcode all your footage to the Cineform codec.   Then use Neo3D to mux your left and right clips into new Cineform 3D quicktimes.  The in points of each clip and be adjusted by up to 15 frames to synchronize the left and right eyes perfectly. In Neo3D set your screen type to match your 3D display.  If you don’t have a 3D display then use an anaglyph mode with anaglyph glasses and adjust the controls to fix any disparities introduced while shooting.   When your clips are ready you can start editing in FCP.

Stereo3D Toolbox
Bring your clips into Final Cut Pro and create a new sequence for each pair of clips.  If your clips don’t already start at the same frame it is a good idea to subclip them to the clapper.  You can stack the left clip on V2 on top of the right clip on V1 and apply the Stereo3D Toolbox filter to just the left clip and double-click or hit Return to open it in the filters tab. Then just drag the right clip into the right image well.  Set your output mode to match your 3D display or use Anaglyph.  Repeat the process for each pair and then you can edit in a new sequence with these paired sequences as clips.

Fixing Disparity and Setting Convergence
Both Neo3D and Stereo3D Toolbox have sliders to adjust vertical, rotational, zoom, color & keystone disparities.   Fixing these disparities requires skill and practice but my recommendation is to start with rotation and make sure any straight lines are parallel to each other and then adjust zoom to make sure objects are the same apparent size.   Next adjust the vertical disparity control make sure all objects next to each other.   Finally adjust the horizontal convergence to perfectly align the object you wanted to be on the stereo window.

Native Pixel Parallax
There is one last thing you should check after aligning each shot.  You must make sure that your background doesn’t exceed the Native Pixel Parallax of your display screen or your audience’s eyes will diverge (which is bad.)   The idea here is that the maximum positive parallax (the parallax of your deepest object/background) does not exceed the human interocular distance when presented.

You can determine the Native Pixel Parallax (a.k.a. NPP) by dividing 2.5 inches by the display screen’s width and then multiply the result by the amount of horizontal pixels (i.e. 1920 for 1080p or 1280 for 720p.)

I present my S3D material on JVC’s 46” 3DTV.  It is 42 inches wide and 1920 pixels wide so the calculation is 2.5/42×1920 = 114 pixels.   This means that the parallax of the background should not exceed 114 pixels.

In Stereo3D Toolbox you can enter your screen width and the filter will automatically calculate NPP and display a grid.   If the parallax in your background does exceed this limit then adjust your convergence to move the depth range back away from the viewer.

Share your S3D Masterpiece on YouTube with the yt3d tag
Now that you have finished editing and mastering your S3D movie it is time to share it with the world.  YouTube added the capability to dynamically present S3D content in any anaglyph format.   All you have to do is export your movie file as “side by side squeezed” and encode it as H264 with Compressor.  I recommend using 1280×720p for S3D content on Youtube but not 1080p.  The workload of rendering the anaglyph result is handled by the viewer’s computer so 1080p will decrease the frame rate on most laptops.

Upload your movie file to YouTube and then add the tag “yt3d:enable=true” to enable YouTube 3D mode.   If your footage is 16×9 aspect ratio also add the tag “yt3d:aspect=16:9”. YouTube 3D expected crossview formatted side by side so if you exported as side by side parallel instead of crossview you will need to add the tag “yt3d:swap=true” to ensure the left and right eyes are presented correctly.

I think I’ve covered the basics of shooting & posting stereoscopic 3D but we’ve really just scratched the surface of what a professional stereographer needs to know.   If you want to continue your education in this area I recommend you pick up Bernard Mendiburu’s 3D Movie Making or search your library for the “bible” of stereoscopic 3D, Lenny Lipton’s classic “Foundations of the StereoScopic Cinema.  A Study in Depth.”