At the Band Pro 3D event today, some industry vets discussed tips, tricks, good, bad, and ugly of stereoscopic imaging and post production, with a huge number of useful tips and suggestions on how to successfully navigate the 3D minefield. My raw notes begin after the jump.
BandPro3D_Panel – thrilling pics of four guys talking geek. Hey, I’m into it….
Jeff Blauvelt, rental company – pre-pro through distro when dealing with clients on low end for 3D
Howard Postley, CTO/CEO of 3ality (high end live 3D) – tools as well to fix 3D
Bob Kertesz – Bluescreen guy super engineer, horror stories from set
Tim Sassoon of Sassoon Film Design –
Jeff – clients at low end were being pushed to get 3D stuff out last January to work with EX3s and Reds – you couldn’t get’em yet. 3D market has been blocked up by Pace, 3ality, and Paradise at the high end, nothing on the low end. It seems simple at first, but….
so what they did – they own EX3s, Reds, F900s – the weight to build a custom rig was better for lighter, EX3s have twice the light sensitivity of Red – deeper depth of field was better for 3D, so EX3 was it. Ignition in the UK built a stereo brain – genlock your EX3s, feed it into stereo brain, get the output to a 3D monitor – instead of $10K Transvideo – got a $2000 Hyundai gaming monitor that worked with HDMI inputs. Live View with EX3s for callibration and alignment, no playback – had to go into post, sync it up, render to a stereo player, etc. to get it done. As a post facility, have a couple of SRW VTRs, since was their deck, just hauled their 5800s out into the field to be stereo capable.
Howard – the good news – 3D is growing rapidly, seems to be here to stay this time – Avatar is looking like it might do some significant box office, CES will be a watershed – bad news – the 3D is hard, hard to do right, a lot of the things people are doing, a lot of spaghetti being thrown at walls and not much of it is sticking (3ality quote) – eye strain and disorientation etc. were mistakes in making the 3D, not necessarily in the equipment. Need more folks trained on how to do 3D right. Hopefully it will continue to go up. They make tools to work on it . In business to make good 3D and manage it capture to exhibition. A bit known for making rigs (not their core biz), made them because they needed to exist and do certain things. Heart of their system is the image processor – their stereoscopic image processor – does everything they can think of for doing 3D for realtime analytics of any signal coming in, measurement at the nanometer level, correction of keystone, rotation, geometry, heuristic control over rigs, metrics, metadata generation, geometric harmonization for multiple systems
Bob Kertesz – working on set w/crews and producers – done about a dozen 3D shoots up to now, one live with 2 ultimattes – about 2/3 of the rigs they’ve used from 3ality, 1/3 from Pace- was all top end gear. Doesn’t worry about the 3D IO or that stuff, concerns himself with the engineering aspect to match the cameras as well as can be, making sure cameras are genlocked (world of hurt later), the little technical things ya gotta do double up for 3D. If your story is crap, its gonna be crap in 3D. The story sells it ALWAYS. Get as techie as you want, doesn’t matter, 3D is only going to make a bad story worse.
Tim Sassoon – involved with 3D before digital – has done 2D to 3D conversions with over/under on a down shooter. Rotoscoping with a pencil on bond paper. Has been doing stereo in 4K for the past dozen years for IMAX films. In last 5 years, IMAX has gone predominantly 3D. The rest of the film business needs to steal stadium seating, high res, big screens, more immersive feel. That is where they need to go. Been doing DI and VFX, a lot of 2D to 3D VFX, just finished Alice in Wonderland. Was always planned for 3D release, but was intentionally shot 2D. Mostly daytime exteriors in UK in springtime – had to sky replace everything. Just as well they shot in 3D.
shooting 2D for 3D instead of 3D for 3D.
Lot of things you can shoot in 2D can’t in 3D.
Binocular stereo is an artifact or our eyes and of the camera the same way that depth of field is – we’re taking pictures of a world in 3D. Our binocular model is not necessarily going to last for 20 years – may have a different model driven by types of display. We want something that doesn’t require glasses and is more holographic – this argues against stereoscopic capture. We’re going in a lot of different ways with that – but we’re moving more towards photogrammetry with an entire scene in the long term future. Images becoming more computational.
Howard – involved live and motion picture productions – live vs something posted:
several right off the bat – dealing with multiple cameras simultaneously you’ll be cutting between at random – in a film, well planned, have some idea of what shot comes after what shot – good luck with that in sports (or a concert?). Whateer you’re gonna in post, post is half a frame afer it is shot – so no time for correction! “some fun challenges” combined with audience expectation issues in terms of graphics – subtitles are a subset of live sports graphics. Sport Vision talks about doing the 3D world and laying it in there – one of the great misunderstandings of what is 3D vs modeled in 3D (he agrees with Tim that binocular capture is a point in time that’ll last 5+ years).
Tim – with Moore’s Law in full effect, everybody has to replace all their consumer tech every 10 years at MAX
Q: Jeff – from budgets for your clients, what advice for producers if don’t have the budget to do proper 3D?
A: for a feature film, if under $1M, can’t do a good feature film at this stage. People are trying at the $300-500K range are trying it direct to DVD. cineform’s tools are pretty awesome for FCP for minor fixes. single camera single directed narrative can be easier than live for 3D. A whole new creative area to be aware of as you edit the film – for the stereographer, how much IO per shot, how quickly is it cut (brain shear), negative to positive parallax for the editor, etc. Is an evolving language.
Howard adds – 3D requires 2 things you can’t do without – a lot of expertise, a lot of precision. If somebody has enough expertise, can shoot good 3D with two cams and a 2×4. If somebody ISN’T a real expert, can go real bad real fast. At the other end of the spectrum, the precision and expertise are moving into equipment and software, so that the requirements of what people need to know is going down – don’t have to have 30 years of experience to make it work. The costs should come down as the technology allows a broader range of folks to do it more reliably and more affordably. but that precision/knowledge has to be there SOMEWHERE.
Q: Bob – what do we need to pay attention to now to avoid the same situation we have with HD?
A: if you mess up the shoot to point where you need a lot of post fixing, almost all of that is destructive. Pablo is pretty good for minimal destruction. If doing on a Mac in your basement, it is going to soften it if not using a great deal of care, your picture is going to fall apart. Get the technical parts RIGHT. Fix It Later is not a big deal in 2D, IS A BIG DEAL in 3D. If one of your cameras is a little off, if you don’t have pristine tools to fix it, it’ll beat it all up. If you’re shooting SxS 8 bit 35mbit, it is going to fall apart – just be aware of it. The cheaper the system you’re shooting with, the more you need to be perfectly mechanically aligned. Genlock – it HAS to be dead on – it’ll hose you BIG time. If you’re going to external recorders, almost all his stuff has gone to SR recorders in stereo video since no issues there – if the two cameras are off by more than 16 pixels, it will error out and not roll!
If you do it yourself or hire an engineer, HAVE A GOOD 3D PERSON THERE to cover the engineering and mechanical.
Howard – fix it in post mentality – belief that 3D effect comes from parallax, spacing between lenses in cameras. WRONG. 20-30 different depth cues establish 3D. Various positions, geometry of shot, etc. Not only do you have resolution loss from rotating the image, the color angle and color parallax moves the depth cue will conflict with the parallax depth cue – that all by itself could cause nausea. That process will blow up or effect the other 20 or so depth cues will have a compounding effect. In gneral, fix it in post is NOT the approach to take! There is no way to fix it easier/better/cheaper in post, get it right on set instead!
Tim – why do 2D/3D conversion and not shoot it 3D inthe first place?
A: some things can’t shoot very effectively – in 2D/3D conversion, normal stereo foreground and hyper stereo background. In 3D, is immersive – immersiveness is experiential – when we experience a scene, build up a sense of the depth of the scene by moving around – manipulating the stereo base throughout the scene – you can’t shoot that!
At the extremes of lenses, binocular stereo doesn’t work well – very wide or very zoom – 3D exagerates foreshortening (makes seem cardboard cutouts) – moreso than you experience in 2D. Divergence barrel distortion is almost impossible to fuse (see it right). Less of an issue with the middle focal lengths.
A lot of the shows done recently (G-Force and Alice) asked to do much milder stereo than you would photograph – flatter projection than actual reality for creative reasons. Both shows were parallel, not converged – this means infinite parallax
A big argument – do you keep the cameras parallel, or do you toe in to converge them
bob on lenses – shallow depth of field is NOT your friend
-7/10/14 on 2/3″, up to 50mm for 35mm
dont shoot wide open, the 3D will fall apart – when depth cue becomes a tertiary depth cue, it falls apart
depth of field will become a deliverable in the future – 2D shallow depth of field, 3D deeper depth of field
the creative aspect of 3D is different than that of 2D – different pacing and timing – not a great thing to shoot one thing and say it’ll deliver in a bunch of different formats – one of the best examples is what to do – shoot it as coverage – as if shooting SD and HD
Producers HATE multiple deliverables – every film released has had multiple deliverables – is the way it is
96 DCP deliverables for Avatar
Christmas Carol – dealt with 5 streams – IMAX 3D, traditional 3D, 2D – all different cuts and
A: glasses gotta go! Fine for 10 minutes at a park, but wrong for watching a bunch of stuff
More 3D HDTVs sold next year than there will be IMAX customers next year
Lenny Lipton’s download from stereoscopic.org for PDF
Autodesk has a good PDF
more is better – learn all you can
the interesting parts they don’t all agree – bettter off reading it all and arguing with it rather than read one and nod OK
things are changing quickly in front of your eyes
Tim – at SOME point watch a chunk of what you’re doing on a big screen – when you see it on a big screen, your sense of scale plays differently – gotta shoot for the screen you intend.
which means beware of what you take away from looking at any other screen – the 24″ screen you edit on, the 9″ you shoot on, etc. More than 50% of shots on episodic TV are soft this season because being cut on 24″ 6 bit panels
I SO AGREEEEE!!!!!!!!!!
Delete the first take of every close-up automatically! The issues you see on a 24″ monitor aren’t the same on a 40 foot screen.
Can’t shoot for every screen you’re going to deliver on – shoot for the BIGGEST screen you’ll deliver on, make that your master. One eye out of focus is a problem on 3D sets – with regard for multiple monitors –
Scaling down nonlinearly will lose resolution somewhere in a nonlinear way – home has different master than theatrical, more scaling happens on the TV, the way it was scaled, lighting, seating position, lighting factors, etc.
you can harmonize the geometry for overall solution – CAN shoot for 80 foot screen and for 40″ home screen and have them both be OK if you account for it.
most films are being played behind the screen, in IMAX it plays out front – you don’t notice edge violations
in regards to editing (Steve, Poster?) – can make mistakes on a 24″ monitor – conform periodically and watch on a big screen – you’ll be surprised
take two regular projectors and put polarizing imagers and make it go
Matrix TripleHeadTo Go to go into two projectors for QT Player or whatever
people are gonna nail HVXs to a 2×4 and go for it.
Q: in 2D people have learned camera tests are important, producers are pushing that down – camera tests is just the beginning of the project -gotta run it all the way through post to make sure it is gonna work, beat the crap out of it long before you’re gonna shoot. Tough to do with 8 bit long GOP MPEG.
If stuck with that, add 1/2% noise in 16 bit space.
The SIF is underdescribed as a lifesaver (the 3ality toolkit)
Fox has deleted the position of DIT on dramatic TV shows. Aside from fact they think they are saving money, a big issue is that there are unqualified DITs on the roster working these shows. There’s an opportunity to have qualified people working the technical positions. Producers’ job is to save money, but gotta talk to guilds and unions and have some pushback. Making sure that the communication gets back to money deciders will help. ASC/PGA tests got folks realizing gotta have qualified folks on set.
Stereo filmmaking classes in film school? (Yes? Somebody said?)
3ality has a certification program
USC will have, next term, 3D instruction, Columbia comign out with it, NYU has it now, one in Seattle
too many holes to fill, in UCLA
Q: the whole 3D thing (gimmeckery) is tech based driven – driven by ability to supply it, not so much there is demand for it?
A: Tim – driven by changes in movie biz – 2012, Transformers – things that would have been theme park rides not entertainment – (ride films) – ripe for 3D – Avatar is a splendid case in point
Technicolor guy’s answer – Chicken Little was a while back – My Bloody Valentine – 6x as many people saw that one in 3D vs 2D. That ratio is changing, but an overwhelming result – 3x as many want to see it in 3D with a $3 premium.
3D isn’t going to make a poor story good or better, but 3D can take a compelling story and make it better. The theatrical side is that if you have a good story, this’ll enhance it
Q: how many consumers go to a SECOND 3D feature?
A: does the novelty wear off?
CES expected a LOT of announcements around 3D
The first market for quality autostereoscopic displays will be sports bars
LG just announced jumping into 3D
()TripleD works amazingly well for conversion as well)