2D23D
If, when, television demands 3D content – wouldn’t it be great if you could feed all your archives into a black box and have it converted to 3D.
Television executives dream of new revenue streams like this. Perhaps they might be able to convince us to watch Friends – yet again – if it was 3D.
Sorry. There is no magic box, not yet anyway.
But all the same the new JVC 2D to 3D on-the-fly conversion box does work and it is pretty clever. There must be a room full of very bright software engineers/geeks behind the thing.
It is a not Velcro camera accessory that you would pop on your credit card at B&H. For one thing the IF-2D3D1 will set you back thirty grand in greenbacks. It is also a device that goes in the post house machine rack or into the DIT truck.
The DIT could find it pretty useful if you are shooting stereo because it is not just a conversion device. It also mixes left and right camera signals for monitoring and alignment of the cameras.
When you shoot with a mirror rig the vertically mounted camera’s image is upside down, and the 2D3D1 will take care of that while keeping sync between the left and right cameras.
But if you see some publicity material suggesting the device will also correct mis-alignments, take it with a grain of salt. What is does do, is offer ways to split screen between left and right eye images to make it easier to check crucial camera alignment and colour matching. There are a dual vectorscopes if you want to get technical about it.
The 3DTV flavours it delivers are line-by-line, side-by-side, above-below, checkerboard and – should you be desperate – anaglyph. If you are doing a conversion job you output independent 1080p left and right eyes.
The conversion of 2D to 3D is both clever and limited.
“I would say in the truest sense, 3D has to be captured in 3D,” agrees Noel Oakes, national sales and marketing manager JVC commercial products Australia.
One reason for his caution over terminology is that the way the device works it is limited to positive parallax, that is, it can’t make objects come out from the screen. Not that this is too critical, since viewers are likely tire of any gimmicky shots.
Nevertheless after watching it convert footage I was left with the impression that it didn’t seem to be able to bring convergence of the main subjects – actors around a pool – up to the screen plane. This is probably a function of the amount foreground in the shot, but the ‘looking out though a window’ result lacked intimacy.
This points to the hazards of assuming you can simply repurpose your 2D archives automatically to 3D. Shots composed for 2D can look odd in 3D.
Oakes wants to keep expectations realistic.
“We don’t want people to think that this is a box that will take any signal in do a perfect 3D job on every scene. It isn’t a set-and-forget across all scenes, you must spend some time to adjust different parameters,” he cautions.
What the device seemed to do well, and without any great fiddling, was inject a sense of 3D into landscapes and aerials, raising the possibility that it might be used to convert some 2D footage shot in conjunction with stereoscopic footage.
To shoot 3D looking landscapes with a 3D stereo rig requires a wide inter-axial distance (distance between lenses) which has the effect of making things look too small. This conversion algorithm doesn’t suffer from the same limitation.
Another problem of stereo shooting is where long lenses produce a flattening effect – objects appear to be cardboard cut outs. According to the IF-2D3D1 manual, the device’s sub-intensity function will avoid this effect, allowing close ups on long lenses.
The industry is still at the early stage of the 3D game and experience is still fairly thin on the ground with this kind of thing.
That’s one reason why manuals like the one issued by JVC for the IF-2D3D1 tend to have useful theory sections – worth reading because each one seems to cover slightly different things.
This manual suggests aerials can be a problem for rigs. “It is not realistic to use two separate lenses for airborne recordings as the lenses tend to vibrate individually.” Naturally, is has the solution “It is much more economical and time saving to record in 2D and convert the images to 3D with the IF-2D3D1.”
They also suggest there is a place for the IF-2D3D1 at sports events where box lenses are way too big for rigs, and in any case risk flattening.
Solution: “Combine a conventional 2D camera with the IF-2D3D1 to create a system that enables real-time, on-site recording with 2D-3D conversion.”
Imagine that. Cheap 3D footy.
