Marcus-

My name is Michael Cioni.  I've worked on a number of 3D RED productions including Pirates of the Caribbean 4 (MX 3D – Pace) and Spiderman (Epic 3D – 3Ality).

The RED MX and Epic have built-in tools that make 3D capture and tracking very streamlined and efficient.  By utilizing Build 31, you can modify the RED file names in camera so that any and all captured media is tracked separately between LEFT and RIGHT eye.  For example: When you setup the camera ID in the PROJECT menu, you can now adjust the ID as "L" or "C" or "R."  In a typical RED capture, names display like this:

A001_C001_0130XY.R3D

With the 3D ID enabled, you can change the display to this:

A001_L001_0130XY.R3D

A001_R001_0130ST.R3D

When you are using a splinter camera, or any non-3D rig for a shot, you can leave the "C" in it's normal position, which can now stand for "CENTER" eye – this tell editorial and conform that any clip ID'd as "C" does not have a paired file.

For even more streamlined efficiency, you can also take a typical/standard "printer" USB 2.0 cable and plug it into both cameras.  By enabling this inter-link, one camera becomes the SLAVE and the other the MASTER.  When you roll the MASTER camera or adjust the menus, the 2nd SLAVE camera will follow.  This way, when you roll, both cameras can roll simultaneously and setups (such as ISO or framerate) will be mirrored between both cameras.

Finally, when working in REDCine-X, you have the ability to start creating 3D clips which will be merged together by either FILENAME (filenames are the same save for the "L" vs. "R") or by TIMECODE.  By jamming the cameras using a LOCKIT box (it is essential for 3D capture to have cameras jammed perfectly throughout production) you can merge L and R clips together in REDCine and create ALEs that contain the metadata required to conform they eyes back together for conform and DI.

Working with 3D is by no means simple, and it requires a lot of serious testing (all of the films I've been on openly admit they wish they could test longer!) so testing for 2-4 weeks is an inexpensive way to address problems with the cameras, rigs, offline, online, and conform as opposed to having to "dig" yourself out of a jam once you're not making your days.

Hope this helps.

best,

m

No Epics in the general population as of yet.  Only on RED-sanctioned pre-made deals are where Epic cameras can be found.  My best estimation is to wait until March before you'll see early numbers surfacing in the rental houses…

Joe-

I can say that the 3D work on Epic with Spiderman was very reliable and stable.  While I'm sure there are stories or rumors about problems on this project, overall the ability to work in 3D was and is fantastic, to the level that I am on my 4th 3D Epic project running smoothly.  Much of the success comes from 3ality's powerful TS5 rigs, their team and technology, which is why I am a big fan of the Epic being used in 3D and the 3ality company.

That said, there are issues that can contribute to balancing problems on any 3D production, to which Epic is not experiencing problems exclusive to RED, rather problems with rigs, cameras, SIP boxes and human error are all components of.

Alignment is a challenge that every 3D film must consider with great intention.  Weather it's Epic, F35 or Alexa, the simple act of achieving exact balance between 2 cameras that were not manufactured identically is an issue 3D  will face until we move towards single-sensore 3D stereo capturing.  It is true that Epic sensors, camera bodies and rig alignment will never (dare I say never) be exact to the micron.  This does have adverse affects on the fidelity of stereo images.  But the fact of the matter is the more time and experience spent on getting cameras as closely aligned as possible will go a long way to ensuring that an image is as near-perfect as possible.

I find that every sensor is different by some amount – similar to how even identical twins don't look EXACTLY the same when you see them standing besides each other.  To suggest that the microns difference between the seating of two sensors is the reason 3D capture isn't exact is to suggest that there is not a myriad of components that also contribute to the problem.  The problem exists, yes, but the exact seating of sensors (in my opinion) will actually have LESS of a negative perceivable affect on 3D artifacts than a misaligned rig or a lower quality mirror.

They say that in order for the space shuttle to take off, nearly 5 million components have to be working in perfect harmony…or the launch is not considered perfect.  So is a similar scenario with 3D capture: so many moving parts, cables and settings, both physical and electronic, need to be considered to make the best image possible.  Even if the sensors are perfectly aligned, a simple 1/16th of an inch offset in physical rig camera alignment will create a much more distinct and destructive result, nullifying the issue of manufacturer sensor placement.   

One of the challenges of Spiderman was also in the fact that the filmmakers chose to shoot 3D at high speeds – speeds higher than that of 30fps.  The interesting thing is that with a CMOS sensor that uses a rolling shutter approach to image capture, locking cameras together exactly at say 96fps is not something that can be technically achieved because there is currently no lockit box that clocks at greater than 60fps.  This means sometimes there may be one component of technology that is literally years ahead of ancillary technologies required to execute capture correctly.  So is the Epic broken because it cannot clock 2 cameras perfectly at 96fps?  Of course not!  It's the edge of technology where compromises have to be made in order to move forward.  (And this issue was solved through some tools and talents that made it work in spite of an initial technological limitation). 

My advice to you is to practice getting those cameras as perfectly aligned as possible in the physical space you are capable of.  Learning how to align quickly and correctly is an art in itself that can keep production moving and (more importantly) ensure that images are the best they CAN be.  Like the identical twins: if they stand next to each other in the same outfit with the same haircut and in the same pose, the differences between them becomes much harder to see.  In other words, MINIMIZE the discrepancies between the two eyes as best as the technology will let you and you won't end up painting the production in a corner.  In my opinion, no camera designed to capture 2D and modified for 3D is ideal.  In my opinion, there is no way to actually get two cameras to be perfectly aligned to the micron.  But you CAN get them so close that some simple adjustments with the precision of post production tools will do the job of making a picture work so that neither the audience nor the filmmakers would ever be aware of any discrepancy in the first place.

m