Friday, December 17, 2010

More details on the small 3D beam splitter rig

The small beam splitter with a couple of hacked GH13s
It seems the beam splitter I designed generated more interest than I initially imagined, so here is some more info on how I built it, but before I´d like to mention a few things, such as why I decided to built this rig in the first place. I am a director and film maker by trade, I have shot commercials, music videos and documentaries with all sorts of sets ups and cameras from Sony´s F35, Red, 16mm film and 5D mk2, as well as a recent project in 3D: a commercial for a German bank´s charity, for Cinema release. The rig was from P&S technik if I remember well, we used a couple of Sony 900, the rig weighted well over MY OWN weight, it needed 3 guys to pass it from dolly to tripod. 

Versus the standard size 3D beamsplitter we used...
I wanted to have a 3D rig that could be hand-held, similar in some respect to the DSLR revolution, small, compact and enabling me to shoot easily wherever I wanted without the need of a crew to lug the gear around, pull focus, and so on. Imagine that at any time during that shoot, there was around the camera, at least the director (me), the DP, the stereographer, the camera assist, and fortunately no additional focus puller, but still 4 guys stuffed around that rig at all time (check the making of on 

When building the beamsplitter, what I wanted to achieve was a rig with the smallest footprint possible, with proper image quality. I am not yet totally happy with the results, but I am getting there. I am thinking of adding a 3D field monitor (bit heavy) or possibly pluging a pair of those LCD 3D glasses instead, so as to be able to check depth while shooting. The rig weight about 4kg, (that´s about the weight of a normal professional camcorder like the HVX 200) which is, as far as I know, the lightest rig with broadcast quality image output.

It took me a while before getting to build the rig, first trying to find technicians to help me, or build it for me, and after not finding anyone ready to help, wondering if I was up to the task myself. I am a pretty handy person, I have a few tools but then iron, drills and screws are not much part of my trade. I finally layed a basic blue print: the concept is in itself very simple really, but the devil is in the details. I bought the mirror online (it took me a while to find the right size as custom made mirrors are really expensive), I went to see a local company that builds aluminium window frames, and asked them to cut me a few pieces according to my specs.

The blue print was essential in getting the placement of the cameras correct, one of the most important things I´d say was to get the lenses as close to the mirror as possible so the least "amount of mirror" is needed for each camera, so as to maximise the useful interaxial distance of the rig. Let me remind you that the whole point for me was to build the smallest beam splitter possible, since after 10cm of parallax distance, it becomes much easier to shoot parallel (basically beamspiltters are a pain, side by side is MUCH easier, more about that in a moment). 
The whole point of a beam splitter is to be ablevto shoot 3D close-ups that you would´nt be able to shoot side by side due to the size of the camera. All I needed was a beamspiltter to accomodate the GH13 (in itself one of the most powerful camera around) for smaller interaxial distance.

The main problem with this rig is that it is VERY basic, so aligning the cameras can be a pain, I can change the placement of the upper camera for interaxial distance unscrewing and re-screwing at the right distance. I needed as well the possibility to accommodate two (very) different lenses, namely the stock 15/150mm and the 20mm, which are at oposites in terms of size. the 20mm is about 2cm thick, the 150 is a good 10 cm long, and weight 4 times as much, both are fantastic lenses. I have a pair of  zeiss FD 55 too, but I don´t bother much with them. I decided to stick to Panny lenses that offered me AF, since I´d be shooting quite a bit hand-held and moving around, and certainly without a focus puller to assist, what´s with the additional remote focus gear on the rig. Since the lenses are not synced, sometimes there are focus discrepancies (especially with the extremely slow focus of the 20mm lens) which obviously mess up the shot. I would usually shoot more takes of the same shot because of that. 
I am well aware that the guys at 3Dvision have spent quite some energy trying to sync both cameras perfectly, but there solution is too cumbersome for me, and I have found out that with a simple dual remote shutter, the sync problem was never more than 8/1000th of a second when shooting 50p, which is perfectly reasonable in most cases, except when there is loads of camera movements (not advisable in 3D in anyway).

The first results where horrendous, I have to admit, I spent quite many hours half sick in front of the 3D screen, and that is certainly not a feeling you wan tot inflict to your audience. It took me a while to nail down all the little details, such as the best depth of field and focus strategy, adding barn doors because flares on the mirror where a disaster.

The thing is  beam splitters are a transition tool, until small enough cameras at reasonable prices are available to get rid of beamsplitters altogether, they are a pain in the bum to calibrate, the mirror will eat your light (2 or 3 stops) and bring color shifts between cameras, it is one more surface/lens to keep really clean, bigger mirrors introduce yet more problems because the glass is never absolutely flat and bends slightly because of gravity, therefore squashing your view, which introduces more corrections in post, etc, etc. It is MUCH easier to shoot side by side, simple. Very efficient cameras are getting smaller all the times, and (shallow) depth of focus is not so much an issue in 3D, so bigger lenses/sensors are not so important. Most likely we wont hear of beamsplitters in five years except for macro really. My own two cents.

On this blog you´ll be able to see some demos and tests made with the rig, you can also have a look here at my own site for more stuff:,(remove the /3d/ to see the 2d stuff) I hope you find this as interesting as I did building and experimenting with the rig, and please, leave some comments, let me know what you think of the results achieved.

Thursday, December 2, 2010

Interesting rig side-by-side nikon based 3D rig...

An interesting picture has emerged of a Nikon based 3D rig, which was used during the Tour de France, I have been trying to gather some info about it without much success. Looks like the left camera is upside down, presumably to reduce the interocular distance, it seems both zooms are paired through an arm linking them. Still looking for infos.

Sunday, November 28, 2010

Autostereoscopic displays right round the corner?

As reported on iPONT International, a creator, converter and distributor of glasses-free 3D, and Palace Cinemas, the largest cinema chain in Hungary, will screen 3D movie trailers without the need for 3D glasses - giving movie-goers the first-ever opportunity to preview movies in 3D while in the lobby. Theywill be able to see autostereoscopic movie trailers on a 65-inch 3D display.
Based in Budapest, Hungary, iPont International ( specializes in the creation, conversation and distribution of 3D content. 

panasonic´s 3DA1 versus a beamsplitter and a couple of gh13s

That camera is definitely quite a nice bit of kit, but it is certainly not worth the huge amount of money they ask.  Filming with a couple of hacked gh13s on a beamsplitter, I can change the interocular distance (the 3DA1 can´t), shoot 50fps in 720p (the 3DA1 can´t as far as I know) up to 80mbps, color space 422 when shooting mjpg (3DA1 is 420 therefore not broadcast compliant for many 3D channels). I can change lenses, have shallow depth of focus if I want to (shallow DOF can be awsome in 3D as well), the rig can be handheld, barely weighting more than the 3DA1, and , oh! it cost me a quarter of the price of the 3DA1. It looks like the 3DA1 interocular distance is about 5or 6cm, there is basically a "sweet point" distance of 1m tp 1, 50 at which the 3D effect works best, closer and you´´ll induce headaches to your audience, further and the 3D effect will rapidly fade.

Friday, November 26, 2010

Promotional 3D film of Ibiza shot with the small beam splitter

Extract of a promotional 3D film for Ibiza which I shot with the beamsplitter I designed, to be used by the tourism board of Ibiza and a local production company with whom I work regularly.

Making of Deutsche Bank Skyliners 3D commercial directed by Bastien...

This is the making of the the 3D commercial I directed in May 2010 for Deutsche Bank Skyliners (through Neue Sentimental Frankfurt), you can see the commercial in an earlier post (below). As you can see the rig is huge, it could barely enter the cubicles where I had to film the kid, which obviously limited enormously what and how we could shoot. This is partly the reason why I decided to design the smallest 3D rig possible, using a couple of hacked Gh1s.

Wednesday, November 24, 2010

Interesting system from SinaCam:

this system is probably one of the most promising I have seen around for a while, tiny but with proper HD specs, and small enough to eliminate the need of a beamsplitter for most shots, this is pretty much the future of 3D filming:

Consisting of  a couple of  2/3“ CCD Sensor 1920×1080 Pixel, 10 Lux/F2.2 Sensitivity, shooting 1920×1080, 1440×1080, 1280×720 interlaced and progressive up to 60 fps, Output Interfaces up to 2x dual HD-SDI with optional Memory Card Module for high speed H.264 recording. With that type of small cameras, the interaxial distance could be reduced to 41 mm (width of the camera), which would allow for close ups down to 120cm or so, close enough not to need a beamsplitter for most shots, exept extreme close ups.

I would love to know the price of this little wonder.