Published : 9th Dec. 2008
What tricks are available to visually smooth the story telling when one actor plays a set of twins. Apart from the basic ones are there other tricks used to help subtly blend the two characters.
Regards in advance
Both films made extensive use of motion control to
duplicate the actor.
In lieu of full motion control, you can do quite a bit with a combination of a memory head (Hot Gears, an Aerohead, and others should be fine) and clever blocking. I've done a lot of shots for a number of television programs this way (Without A Trace, One Tree Hill, Veronica Mars, and others). You'll have to use some green screens on occasion if one character is crossing the other. The primary things that sell such shots are a combination of proper eyelines (very, very important) and some camera movement on at least one angle of the scene, which tends to make it all seem a lot more natural. You might also consider structuring some of the coverage so that there can be a "hero" moment when the "double" is revealed - for instance, begin the scene in practically shot overs (using a double for the FG character), and the have the FG character turn into the camera (this would have to be a green screen shot, of course), matching the VFX shot to the actual scene coverage. Or have one character put something down, and have the other one pick it up. It will be useful and probably necessary to have video and synced audio playback of previous takes, you should sync this with a bloop light for the second pass either automatically or by hand.
In the end, much of the success of these kind of shots is dependent on the performer's ability to match themselves, and that's something you can't control. In any case, you should have a visual effects supervisor on hand who's done these kind of shots before. The VFX supervisor can help you achieve the shots your director wants, as well as possibly help to embellish them and suggest alternatives if they're appropriate. You should also arrange to have the director, yourself, the VFX Supervisor, and the actor talk prior to getting on the day with this, so that everyone is on the same page and the actor knows what's going to happen. The last thing you want is an actor who's paying too much attention to the mechanics to the detriment of their performance. Some will get there quickly, some won't, it's just one of those many weird talents (like looping).
Chief Technologist (and still sometimes a VFX Supervisor)
Cineworks Digital Studios
Also Star Trek Nemesis.
We used motion control for Brent Spiner to play two characters, Data and B4 androids. I was the motion control operator. Mr. Spiner did some amazing performances totally in sync with each character. His long time stunt man played opposite for him for both sides of the shot. You can see all the twinning scenes on the special version of Star Trek Nemesis, one was deleted from the movie, but is somewhat restored in the deleted scenes section.
Worked well overall.
I have always been a little suspect of "repeatable" heads in that they typically don't take a camera sync pulse (great if they do) and therefore are not really in the same place for each frame in the way that a properly setup motion control rig does.
I always think of these for repeating difficult camera moves for the operator's sake, not so much for accurate camera positioning on a subpixel level. They may be adequate for television, but I wouldn't think they would be that great for theatrical. Depends on the shot, if you intend to shoot the second character totally on a green screen then fine. I would make sure to get a clean plate just in case.
Check out the General Lift company for motion control gear and some examples.
VFX Supervisor, Digital Artist and Motion Control Operator
Los Angeles, CA
Chris Dawson" wrote:
class="style19">> I have always been a little suspect of "repeatable" heads in that they typically don't take a camera >sync pulse (great if they do)
The new version of Hot Gears does. I believe the Aerohead can, too. Dave Stump, are you out there?
class="style19">> They may be adequate for television, but I wouldn't think they would be that great for theatrical.
I don't necessarily disagree, but I have had a reasonable degree of success with this approach, even with 16mm (I didn't want it but I had to deal with it....).
Cineworks Digital Studios
Chris Dawson wrote:
class="style19">>>I have always been a little suspect of "repeatable" heads in that they typically don't take a camera >>sync pulse (great if they do)
A playback system on set were the images can be combine to see if the shot works on its own as well as a cut of the scene to see if it works with in a scene, become very important.
Would HD be preferred over 16mm due to its image stability or is this a case of theoretically yes, but in practice it doesn't matter. This is in relation to a theatrical release.
Chris Dawson wrote :
>> I have always been a little suspect of "repeatable" heads in that they typically don't take a camera >>sync pulse (great if they do)
Mike Most replied :
> The new version of Hot Gears does. I believe the Aerohead can, too.
Just putting in an observation. I don't believe the Aerohead actually tracks the shutter pulse, it just triggers the move start to it. This means that the moco move will START with the shutter in the same phase on each pass. If there is the slightest variance between the frame rates of the camera and the computer, the sync can drift... meaning that the take can be of different length (by a few frames) or can drift out of shutter phase sync.
I'm pretty sure that Aerohead and Mo-SYS work this way, starting on a pulse but not tracking the subsequent pulses. I don't know about the other repeat heads (Hot Gears, Scorpio, ArriMotion, etc).
Why this is important (in case it's not obvious) is that the movie camera's shutter is closed half the time. You don't want one element of the shot to be made with the camera shutter open in the move positions where it was closed on matching elements. The motion control system I'm most familiar with, Kuper Controls, actually tracks the shutter pulse --- essentially the computer slaves to the pulse and maintains the same shutter phase relationship on all passes.
If you are shooting split screens, this is important. Green screen composites, less important. As Mike says, often the "random motion" of the motion control rig is greater than the phase offset, but if the phase offset becomes an issue, it is a very big job to fix in post... a much bigger job than stabilizing "random motion".
Gear+Rose Motion Control
Torstein Dyrting wrote:
class="style19">>> Would HD be preferred over 16mm due to its image stability[...]but in practice it doesn't matter. >>This is in relation to a theatrical release.
The latter. In my personal experience most post people aren't particularly concerned about 'stability' these days. It's a relatively simple fix.
If you're going to print and can shoot on as slow a stock as every situation will allow you then you'll have the best possible facsimile of 35mm. HD is HD is HD. The Genesis is - so far - the only digital camera that seems to produce an image that looks 'right' as opposed to something alien and (to my eyes at least) rather wonky.
Has anyone here done a comparison between shooting greenscreen elements on 7212 and HD?
Don Canfield wrote:
class="style20">>> You don't want one element of the shot to be made with the camera shutter open in the move >>positions where it was closed on matching elements.
class="style20">>>but if the phase offset becomes an issue, it is a very big job to fix in post...
Can you give an example Don? I've used one of the systems you mention a lot and I've become quite 'laissez faire' about it.
class="style20">>> Can you give an example Don? I've used one of the systems you mention a lot and I've become >>quite 'laissez faire' about it.
I'm not Don, but as most of us VFX nerds will point out, ALL the judgement calls about what's OK or not with greenscreens, or how accurately multiple pass shots are highly situational. In fact, the situational nature of these judgements is why we think we provide a useful service to the industry phase is important - example:
Let's say you are shooting at 24fps and the shot includes a fairly rapid pan from right to left... not strobe fast, but fairly fast.
Let us assume that the camera speed control is modern and accurate.
Let us assume that the "repeatable head" time-base is also modern and accurate.
Let us further assume that the "repeatable head" is mechanically accurate in the case of clamp-on motors for geared heads, this is dependant on the lash adjustments and wear and tear on the head) With these favourable assumptions, it is reasonable to assume that over a short move, you will have the same number of frames for each move.
Even so, if on pass one, the camera was "centre shutter open" when the head was in the middle of head move frame 73, for instance and on pass two the camera was "centre shutter closed" in the middle of head frame 73 (to be arbitrary about where we are in the move, but exhibiting the worst case of 180 deg out of phase) then pass 2 camera frame 73 will be exposed EITHER half way between the physical camera position when it exposed pass 1 camera frame 72 and 73 OR half way between the physical camera position when it exposed frame 73 or 74.
If the shot is a lock-off it doesn't matter, but if you are in a fast pan, the camera might be pointing several degrees differently from the previous pass...which means the elements will not and CAN NOT line up. There is no frame of film exposed in pass two that corresponds to the camera position in pass one.
Now: If your second pass is to provide a clean plate for rig removal, you are probably just fine... the relatively small offset is not a big deal when choosing frames to find the necessary material to paint out the rig.
If your second pass is for a greenscreen-backed actor insertion (as with twins) and not touching anything in the first pass, you are likewise probably OK... your second pass element is isolated from the background and can be slid.
If your second pass is for a soft split using the 2nd pass background and you are in the desert somewhere with a non-descript background, you might or might not be OK, but remedial work will have to be done to make it work and a pixel-pusher somewhere will use your name in vain.
If your second pass is for a soft split where there is either close interaction with foreground elements or a fairly specific background (like a city street for instance with all that architecture) you have now created a massive roto job at best, and, depending on lens distortion, (which affects images size and shape depending on which part of the field the objects are in) maybe even more serious headaches.
It is quite likely that the time and money spent to make the shot look good will vastly outweigh the slight additional cost of using a proper phase-accurate motion control head.
If you are scaling the time base for a shot where you shoot one pass with the head and camera going at one speed and the second pass where the head and camera are both sped up or slowed down by the same amount, you will discover that many servo-driven "repeatable heads" do not do precisely the same move at different speeds. I could write a ten page treatise on why this is sometimes so, sometimes not, and situational with regard to the nature of the shot, the particular design of a particular head (and its software) and important or not with regard to sticking the elements together in a shot...but I won't.
I certainly don't have a horse in this race... horses for courses and all that... and I am all for doing whatever it takes to make it fast and easy to shoot on set...even if that sometimes makes life more miserable for the post-folk...but these decisions should be informed ones. A simple pan and tilt motion control head is not all that much more expensive to rent than a repeatable remote head... especially since most crews in the US will hire a specialist head technician to work with the head... he might be a little less expensive than a motion control programmer, but not orders of magnitude so.
There are a number of vendors of motion control equipment whose packages are reliable, well packaged for work in rugged conditions, and efficient to rig and run... and there are other vendors also:-) The words "motion control" scare veterans who remember the good old days when the gear was essentially stage equipment dragged out into the field and scare people who have not worked with motion control and have heard stories. No one tells the stories about the thousands and thousands of shoots where the motion control equipment worked properly and the shoot was uneventful, because those stories are no fun to tell.
sometime Motion Control Supervisor
Luddite at heart
Mark Weingartner wrote:
>>In fact, the situational nature of these judgements is why we think we provide a useful service to >>the industry
Mark - Just wanted to say thanks a bunch for your 'real world', hands-on, detailed insight into this mo-co situation. That is exactly what makes this a great forum - the open-minded sharing of great knowledge gained by our peers with a truly 'professional' attitude and a love of getting to the real truth about things - not just suppositions and marketing hype!
class="style20" >>I've often found that mechanical movement - i.e., the boom arm, the floor, the dolly if used, the >>mocon track if used, etc. - is usually more significant than the in camera pin registration.
You would be wise to assume that tracking and stabilization in post will be required, regardless of shooting technique or media.
SFD vfx & creative post
Santa Monica, CA
From Peter Pilafian:
class="style20">...That is exactly what makes this a great forum - the open-minded sharing of great knowledge gained by our peers with a truly 'professional' attitude and a love of getting to the real truth about things - not just suppositions and marketing hype!
Boy, I'll second that, even if it puts me in the company of people I may not deserve to be considered with:-)
At the risk of sounding like I am sucking up... which I ain't...that Geoff Boyle fellah really started something...
I had a long conversation with a friend of mine a few days ago who wanted to sort through his options on a particular shooting problem... and right at the beginning of the conversation he admitted that the only reason he called
me was because there would not be time for him to harvest the answers had he posted on cml.
This has become the first reference point for a number of people faced with technical issues or new territory... and considering who few DP's work standing next to other DP's and how few gaffers work with other gaffers, this has become such a rich bazaar wherein informed, experience is traded...
...it is SO nice to be able to learn from other people's mistakes once in a while:-)
We need to keep the attitude of positive sharing and discussion - no question too stupid or naive to ask and all that - because a great number of us allegedly experienced pro's learn a great deal from reading the answers to some of those "obvious" questions.
LA based VFX DP/Supervisor
> Can you give an example Don? I've used one of the systems you mention a lot and I've become >>quite 'laissez faire' about it.
Tom, et al....
First of all, I think I know the system you speak of, and I have nothing but respect for Michael's gear. I would buy one myself if I thought I could market it effectively.
Mark wrote a terrific treatise on your question.
The first time I saw the problem, I had never seen it before, but I knew instantly what it was. My mouth when dry and my knees went weak. I knew what the problem was and understood the "fix" before anyone else in the room did. The only thing that kept me "off the hook" was that I had recommended to the fx super (who also owned the moco gear) that we had to sync the passes. This was many years ago, and I didn't completely understand why, but this was the "prevailing wisdom" of those more experienced than I. He insisted that it didn't matter, and we didn't really have the time to build what it would take to read shutter pulse for sync (connectors and a simple circuit had to be built to interface with the RP4 computer... there, that will date me to those who know what RP4 is...) So, we shot with an Arri BL4 "wild"... that is, without reading shutter pulse.
The shot was a simple quick tilt down shot of 2 copying machines as the printout shot out of the machines at the same time. The trick was that we used only one machine rigged with motion control motors to the paper mechanism. The tilt down and the paper movement was synchronized so that the paper would move out of each copier at exactly the same rate. To make the gag work, it was essential that the desktop surface align exactly, so it appeared that there were 2 machines side by side on one surface. The shot started static, tilted down, and then stopped and held.
On the first rough composite, the static start was spot on, then in the tilt, the desk surface edge separated and then realigned at the hold at the end. The fx super said, no big deal, we fix it with a shift in post. What he didn't realize initially was that the fix was a frame by frame reposition for the length of the camera move, as there was not a constant speed. Each moving frame needed to be repo-ed. This was not a simple task, as the slightest misalignment is immediately evident in the running footage. And this was 20 years ago, before digital. It was a nightmare.
I imagine that in this day and age, it's less of an alignment nightmare with tracking and digital tricks. But if you are going to use a repeat head to accomplish this, it makes sense to use one that will eliminate these kinds of problems.
Typically, repeat heads cost about $1500-2000 / day. Motion control head only packages are typically $1000-1500 / day. Repeat head techs are $500-700 / day. Motion control operators are $700-900 per day. Cost wise on set, it's pretty much a wash cost wise. It's always best to get the gear that will serve you best.
As Mark said "the situational nature of these judgments is why we think we provide a useful service to the industry:-)" He also said "The words 'motion control' scare veterans who remember the BAD old days when the gear was essentially stage equipment dragged out into the field and scare people who have not worked with motion control and have heard stories." This is now a mature technology, and there is little to fear from it.
For what it's worth....
Gear+Rose Motion Control
I have had success using the Arrimotion system for this application on two TV series for DP Blake T. Evans in Los Angeles. One was the drama. The District and the other was the comedy Jake in Progress in which there were two sets of "twins", something to do with alter egos I believe.
The Arrimotion does indeed take a sync pulse from the camera, if it's an Arri, and drives focus, zoom, iris and frame rate as well as pan and tilt. With the proper camera, say a 435 Extreme or Arricam with Lens Data System and LDS lenses, you can accomplish most of what a proper motion control setup would. I say most because unless you are in NY, London or Germany you will not be able dolly or boom the camera. I believe those are the only places that have the converted Panther dolly.
With the Panther, it is a quite portable and rugged motion control system, without it is a grand, rock-solid repeatable head that will trigger any number of "events" such as lighting, motor control turntables, bloop lights, etc. It can ramp and trigger via frame numbers or timecode and I believe now has the capability to connect to a PC laptop via Ethernet and use a graphic display, enabling the tweaking of beginning and ending of moves, zooms, etc. just like the big boys would do it. The big advantage as I see it is the way the system may be used on set as a standard gear head with remote FIZ until the setup requiring multiple passes comes along and you simply record moves with your regular operator at the helm. No waiting to setup moco head and strike it when shot is done
Although I have used most of the bells and whistles on other shoots, I think at least on one of the shows I mentioned was with Panavision cameras which will run "wild" with no sync pulse. We did empty plate passes and everything worked out fine but Mark W. pointed out how that may or may not work depending on the nature of your shot.
Finally, I must say that it was wisely pointed out that the financial end of it with rental and technician rates doesn't provide a savings. Ease of use and portability are the advantages.
I have no horse in this race, only time spent learning the system and a fondness for Arri engineering.
camera operator and sometimes Arrimotion guy
I know I am a little late to this thread but I feel it necessary to mention another system which has not been spoken about but let me disclaim, we are rental agents for this system so there is a vested interest, or as it has become more commonly known, we have a horse in this race.
The RevolveR is a Motion Control System designed by JWC (Jackson Woodburn Controls) which IS slaved to Camera Shutter Pulse, avoiding issues so excellently described by Mr. Weingartner.
The beauty of the system is its simplicity especially with regard to integration with standard set equipment.
It is a highly affordable alternative to some of the more costly less accurate systems available (I am referencing repeatable heads rather than any other true motion control system).
I did not mention it earlier because the units which have been readily available for the last couple of years have been strictly Pan, Tilt and Lens function repeatable, until now, the new develops include remote handles and the addition of a track axis taking it into real MoCo territory - it has just been on a big job in Tanzania.
Apparently along with the systems in London, SA and Paris, there is one available in LA too, although I have no idea where it is or whose care it is under.
For more information please check:
Dean Slotar | One8Six Cape Town
t +27-21-555-1780 | f +27-21-555-1828 | m +27-82-895-2620
Tom Townend wrote :
> Can you give an example Don? I've used one of the systems you mention a lot and I've become >>quite 'laissez faire' about it.
I know nothing about the Aerohead but I do know a little bit about the Mosys. As I understand it (and I may be wrong) the mo-sys rig is triggered by the pulse and then each incoming pulse is compared with the internal pulse from the rig .. if they are more than 20ms apart then he rig will come up with an error which is a flashing part of the Display and used to be a honking buzzer.
I have to confess I prefer the Kuper system of letting you know the number of degrees out of Sync the system is at any one time.
When you are using a Crystal Sync camera the speed is not in fact constant (if you were using a stepper motor then it would run at a constant speed) the variations are miniscule .. if you put a 435 into engineering mode or look at the phase output on a Kuper rig then you will see a slight variation so a 435 output will vary between 25.025 and 24.975 if running at 25fps. The reason for this is because the servo motor is continuously adjusting and correcting ..
The problem is that this "Error" is larger (In Millseconds) the slower the motor is run so it can go over 20Ms.
Specialist Camera Assistant
07973 317 241
Portable Pan Tilt And Track Moco rig
Funny you should ask but this just happened on Monday. The Australian Society of Cinematographers put on a Chroma Key workshop and we shot tests on many cameras that included the Sony F900 and Super 16 with 7212. We even had a surprise visit from a F23. Hopefully we will include the F23 in our tests. It may take some time to ingest and process everything but the ACS will make a "report" ASAP.
0412 677 541
class="style20">> I know nothing about the Aerohead but I do know a little bit about the Mosys. As I understand it (and >>may be wrong) the mo-sys rig is triggered by the pulse and then each incoming pulse is compared >>with the internal pulse from the rig
Last time I spoke to Michael Geisler about this, he said it only triggers the move start to the shutter pulse. It was one of the reasons I felt the system was not a truly viable motion control head. I asked him to look into making it track the pulse, or better yet, to build an interface that would allow the Kuper system's step/direction output to control his CANbus based electronics. If this has changed, it's good news.
Gear+Rose Motion Control