Monday, February 13, 2012

Friday, February 10, 2012

The Hobbit Production Logistics

3D Production

Lately, I have been asked by a number of productions about the feasibility of shooting in 3D. 
Below is a Q & A, I have put together on the subject.

Q; How much will it cost to shoot in 3D?
The short answer is that 3D will add approximately 35% to your production budget .

Q: How does 3D filmmaking work?
A: Most of us humans see life in 3D.  That’s because we have two separate, distinct eyes that naturally create two, slightly different perspectives.  In the brain, this small variance in image perspectives is interpreted as depth and dimension.  In the world of 3D filmmaking, we replicate this process by using two cameras.  One camera to represent the left eye and a second camera for the right eye.  Then we sync, place the cameras on a 3D rig, align and offset the two cameras.  We’re then able to film a subject, deriving two images with a slight perspective shift. When these images are overlaid in post, the perspective shift produces the illusion of dimension, or 3D.  What’s miraculous about our human eyes is the ease, speed and accuracy at which they’re able to re-focus, converge, and shift from one object to another, delivering a 3D picture all in under a millisecond.  It’s much more difficult to re-create these precise actions in a 3D filmmaking environment. But human eyes, although amazing, do have a few setbacks when it comes to stereo viewing. 

Since the distance between our eyes is ‘fixed’ on our faces, as the distance to objects increases or decreases from us, our eyes can’t change their position, or perspective.  For example, notice that mountains in the distance appears flat. This is because the perspective shift (distance) between your two eyes is fixed and slight, so that at such distances you see virtually the same image and therefore no discernible dimension.  In such an instance, 3D filmmaking has an advantage over human eyes.  Filmmakers are able to increase the IO (inter-ocular) distance.   In this context IO refers to the distance from the center of one camera lens, or eye, to another.  An increase in IO creates a greater shift in the image perspectives and thus accentuates the 3D effect.  So when filming mountains at a great distance (miles away) in 3D, your two cameras may be 10-20 feet apart or more, depending on your shooting ratio and/or desired 3D effect.   This ability to change the IO is the 3D filmmaker’s primary tool for manipulating dimension – depth and pop.  Conversely, pushing two cameras closer together (decreasing the IO), diminishes the perspective shift, thus reducing the 3D effect.  Beyond IO manipulation, stereographers (3D experts) can incorporate secondary tools like convergence (tow-in) and image alignment (xyz axis) to achieve a comfortable, eye-catching 3D images that ultimately serve the story.  As a side note, when filming mountains in 3D at a distance, typically NO convergence (tow-in) would be used, but rather the cameras would be fixed straight ahead.
Q: Why do I need a beam-splitter (mirror) rig to film 3D, can’t we just mount two cameras side-by-side?
A: The short answer is – you CAN shoot 3D using two cameras mounted side-by-side.  As a matter of fact, this is the preferred method for filming scenic 3D landscapes, wide shots and any longer focal lengths.  But unfortunately, for shots where the focal length is about 25 feet, or less (medium and close-up shots), you need a beam-splitter, or mirror box rig.  Why?  Because objects filmed at closer proximities require only a slight or small perspective shift between both cameras to achieve comfortable 3D images.  Put simply, your 2 cameras must literally share part, or most of the image area being filmed.  Unfortunately, with a side-by-side setup the camera bodies “get in the way” and won’t allow your lenses to get any closer than about 5 or 6 inches, making 3D uncomfortable.  Conversely, a beam-splitter rig allows the cameras to actually get on top of each other, sharing the same image area.  In the simplest terms, beam-splitter rigs let one camera typically positioned horizontally shoot “through” a piece of optical-grade mirror glass -- while a second camera set perpendicular and vertical, films “off the reflection” on the front side of the mirror glass.  In this posture the shift in perspective between the two cameras can be very, very slight.  We highly recommend that your beam-splitter mirror glass be an optical-grade, very flat, very clear variety with highest-grade 50/50 coating.  If your glass quality is in the least bit substandard (and most inexpensive mirror glass is) – your images will be unsatisfactory and your 3D images unusable as professional 3D.
Q: Why not use a dual lens camcorder to shoot stereoscopic 3D, like the Panasonic 3DA1?
A: The “fixed lens” nature of a dual lens 3D camcorder restricts the IO distance (it’s fixed) and therefore the control over the 3D effect.  Rather 3D camcorders rely completely on camera convergence (tow-in) to manipulate the 3D effect.  Convergence is typically set after the stereographer sets his or her IO.  Convergence is the icing on the cake, so to speak, not the main course.  So using a dual lens camcorder, limits the types of 3D scenes you can capture – no medium or close-ups, flat landscapes and doesn’t allow for things like polarizing filters or lens choices.  So with camcorders like the Panasonic 3DA1, with a lens separation of about 3.5″ (70mm), you’re restricted (for comfortable 3D) to subjects between approximately 12ft (4m) to 40ft (12m). Beam-splitter rigs don’t have this same limitation as the camera separation can be adjusted all the way down to zero if necessary for objects directly in front of the rig.  This is of major importance when you consider that most theatrical stereoscopic 3D productions use inter-ocular (IO) distances of between 1″ to 2″ (25-50mm) and shoot using high-grade lenses – something only attainable on a beam-splitter rig.

Q: Do the cameras need to be gen-locked (synchronized)?
A: Yes - you will need to have at least one gen-locked camera to sync it to another camera, or use LANC equipped cameras with a 3D LANC controller.  If you don’t gen-lock your cameras, it’s very likely that some of your frames may be out of phase with each other.  Please understand that you CAN NOT correct this phenomenon in post and it may render all, or some of your footage useless in 3D.  In such a case, you’d have to use the 2D version. The bottom line is, if you’re shooting a professional 3D you need to use two identical cameras that gen-lock. 

Q: What is the light loss?
A: With a true 50/50 beam-splitter mirror each camera only gets half the light.   Therefore the light reaching each camera is reduced by approximately 1 f-stop. It is impossible to design a mirror rig with less loss, the light is after all being divided by half. Be wary of any rig with anything other than a 1 f-stop per camera loss.
Q: Can I zoom my cameras?
A: Yes, but you might not want to.  First you will need to find a way to synchronize the two zoom lenses. This can be done with most newer 2/3” digital broadcast zooms. Fujinon now has special matched pairs of lenses available for 3D applications. But it’s difficult to use with smaller compact camcorders, as lenses do not normally zoom in sync.  You should be aware that as a zoom lens changes focal length, the center point shifts slightly which will misalign both cameras.  Even more importantly, is that zooming “in and out” in 3D, is not a pleasurable effect for the viewer.   A zoom foreshortens the image, while the 3D depth increases.  In reality, it’s a funky visual effect that takes the rattles the viewer out of the picture.  It’s best avoided, if at all possible. It’s best to shoot your master shots, then push in, re-frame and shoot your medium and close-ups.  It is recommend shooting “in zones”, where the camera remains equidistant from the subject, but moves in a variety of side-to-side fashions.
Q: Is the mirror glass delicate?
A: No more than any other piece of 3mm glass.  The mirror is an optical-grade, clear glass, so drop it and it will break. When filming 3D - always have a spare piece of mirror glassine your set.  If your single piece of glass breaks, without an extra - your shoot comes to a halt.

Q: Should I use a 3D monitor when shooting 3D?
A: Yes, you should be monitoring (watching) everything your shooting in real-time, stereoscopic 3D.  Filming in 3D without using a 3D monitor – is like driving blindfolded.  A multiplexor is a device that combines the signal from 2 cameras in real-time, then translates them into a single 3D image – usually anaglyph or polarized. The monitor you select is dependent on several factors.  Are you shooting in the field without a 120V power source, or in studio?  In the field, a fully contained 12V 3D viewer system is recommend. In the studio, a multiplexor and large screen HD monitor is preferred.
Q: Can I make my own 3D rig?
A: It's not a good idea in any case. Cost savings verses catastrophic failure are not warranted. If you are shooting a micro-budget film stick with native 3D cameras.