MIO3D push for stereo rigs with 3 or more cameras

The principle behind stereoscopic 3D is relatively simple. Two cameras record the same image/scene from two different angles. When each eye sees the corresponding angle, a stereoscopic image is formed by the brain.

 

There are two main variables within a stereoscopic camera set up; how far apart the cameras are which is known as the inter-ocular distance (or sometimes referred to as inter-axial) and the convergence of the cameras, that is at what point the images intersect each other.

The inter-ocular (IO) distance affects the ‘depth’ of the scene. A very short IO distance of a few millimetres produces a 'weaker' 3D affect. It would only be ideal for capturing stereoscopic 3D images very close up. A wider IO produces the opposite result.

The convergence point influences where an object appears – in front or behind the screen. It could also be on the plane (the screen) where the two cameras intersect (the convergence point). Before the convergence point, the objects will appear in front of the screen (known as negative z space and is used less often in ‘modern’ 3D) and objects that appear after the convergence point will appear behind the screen (positive z space).

To the most point this set up works. However, on a professional shoot, the IO distance is often adjusted for each scene and for other shoots, so is the convergence point. Some directors will shoot in parallel mode and others will shoot in convergence mode. Shooting in parallel means the convergence point is set during post production. Both cameras are set on the rig pointing forward parallel to each other to record the scene. This allows a lot of freedom on set which can be especially useful if there are a lot of SFX scenes and there is an adequate post production budget.

Shooting in parallel is known as a ‘purest’ method and shooting in converged mode (where the two cameras our towed in) can create an artefact known as keystoning (to understand what keystoning is click here) but shooting converged only creates keystoning if the entire imager is used which professionals don't do. Shooting in parallel mode eliminates keystoning but forces the maximum movement to reach convergence.

The tighter the cameras, the ‘weaker’ the 3D effect is but the convergence can still be adjusted to be as far or near as required. The wider the cameras are, the ‘stronger’ the 3D effect. Shooting converged is good if the project is very simple and post production time and budget is limited. Shooting in this way is also guaranteed to be closer to where you want to be than parallel thus wasting less of the information on left and right.

Either way, both techniques use fixed IO distances. So why is this problem? Well 3D is highly influenced by screen size. A video specifically produced for a platform such as the Nintendo 3DS will require more of an ‘extreme’ 3D effect (wider IO distance) than a theatrical motion picture for example.

Also, adjusting the IO distance on a stereo rig for each shot takes up time (and money) during production. If the IO is not set correctly for the given shot it cannot be adjusted after the fact so the stereographer only gets one chance to set the IO for one screen Size.

However a company founded by Sean Fairburn in 2009 called MIO3D (Multi-Inter-Ocular)  claims to have solved the problem by creating plates suitable for more than two cameras; from three up to potentially hundreds!

The idea is, by installing at least three cameras on the same rig, the stereographer can choose between a tight IO, a medium IO and a wide IO so the editor gets multiple convergence points in post-production to choose from. There needs to be no adjustment between each shot. Each camera is recording video simultaneously and it is the editor’s choice as to what he or she uses.

Sean told 3D Focus “MIO 3D was set up to solve the problems that were consistent and not fixable in post, namely getting the IO right for shots that have a lot of motion and screen sizes that need to be fed correct scale.”

The company now has twelve separate rig designs which allow for shooting parallel or converged. Either way, a user has the ability to adjust the convergence point after the fact wherever required.

Another use for the versatile Fairburn 3D MultiPlates could be for autostereoscopic capture. Filming for presentation on a glasses free 3D display requires a minimum of between five and eight different angles (depending on the display) unlike conversion software is used such as Triaxes.

“I would say that first MIO 3D gives the production houses more distribution opportunities in markets for 3D they would not have had such as the iPad and home theatre.” said Sean Fairburn. He continued…

“Having multiple IO for a given screen size makes for better 3D and thus bigger audience but the cost savings on set are marginal. However I would say that once the cameras are set in the morning you don't realign anything the rest of the day and just shoot, so the time in production is the same as it would be in 2D HD. In post, the alignment fine tuning is done once and that correction is applied to the whole day’s work. Just like a colour decision list, we have an alignment decision list and apply it to everything shot that day or that scene. This saves a massive amount of time and gets closer to 2D HD post with one extra step.”

Sean Fairburn has a patent pending on the use of three or more cameras to get multiple inter-ocular out of it in the field and is looking to licensing it to other camera companies and production companies interested in shooting the MIO3D method with their own cameras.

It is a methodology rather than a complete system so is compatible any cameras from cheap and cheerful GoPro cameras right up to high end RED cameras.

The video below from NAB 2011 will explain more and you can click here to visit the MIO3D website.

FREE WEEKLY 3D NEWS BULLETIN –