Jump to content



steadicam and motiontracking


  • Please log in to reply
19 replies to this topic

#1 sebastian matthias

sebastian matthias
  • Guests

Posted 09 April 2005 - 06:32 AM

hi everybody !
i´ve got a big projekt coming up with loads of steadicam.
pretty much all shots are done with steadicam (in front greenscreen!) wich then have to be combined with models filmed by motion control.
we looked into a thoma "walkfinder",but it won´t work because of limitations in height and space (the set will be 800 square meters wide and 7 meters high !)[SIZE=7]
we are thinking about motiontracking programs now to feed the motioncontrol with the steadicamshots. has anyone of you done something like that before and knows which motiontracking program works best with steadicam ? will normal tracking points in the studio be enough or are there any other tricks with tracking steadicam-shots?

all infos would be great !

thanks for your help!

cheers
sebastian
  • 0

#2 thomas-english

thomas-english

    Advanced Member

  • Sustaining Members
  • 1165 posts
  • UK

Posted 09 April 2005 - 07:46 AM

My research on the subject, whilst with a slightly different aim led me to

Boujou Bullet by 2d3

Coincidentally I knew from ages ago a couple of friends that worked there so I dropped by and had a look at some of the demo s and it worked well.

Basically it takes points in the image and nodal/track matches them up ... Creating a 3D tracking chart of hundreds of 3D points. If you were compositing you would then match up your environment to this mass of objects in 3D space. If you were doing your shoot.

Mark up your green screen with lots of white balls . Even Have some hanging. Do your motion track. And export from Bullet your track something you can import into mocom.

Dunno how well it would work.... Im sure they would let you do a test, they re based in oxford

http://www.2d3.com/j...w.jsp?product=7

Let me know how it gets along
  • 0

#3 Howard J Smith

Howard J Smith

    Advanced Member

  • Sustaining Members
  • 279 posts

Posted 09 April 2005 - 08:24 AM

Hi

On Matrix (2&3) we used various methods of tracking - inc. the usual tracking balls (tennis balls hanging every 4 feet in a chain down against the screen, then tracking crosses.)
Then we also used a 3D tracking system using either 3 film cameras locked together or 2 lipstick video cameras on the steadicam about 12" apart this then gives 3D tracking - John Geta is the man for this.

There is also a software programe that can track from the rushes and then you can use this information to dive a motion control system.

We have also done motion capure of Steadicam with data capture balls mounted to the steadicam
please see - http://walterklassen...dioperators.htm there is a shot of me using this.
This was great as we were doing live comping - so I could see the comp on my steadicam monitor. - very useful for framing. (CG back grounds though)

Then there is also the grid pattern on the green screen as well. - the computer knows every inch of the screen and then can track a background to this.

I would be supprised if your VFX superviser has not got something arranged for you.

Hope the shoot goes well, sounds like fun.
All the best
Howard
  • 0

#4 thomas-english

thomas-english

    Advanced Member

  • Sustaining Members
  • 1165 posts
  • UK

Posted 09 April 2005 - 11:26 AM

The Boujou software I was on about would be similar to what Howard was comenting on " There is also a software programe that can track from the rushes and then you can use this information to dive a motion control system." as I know framestore and the like have bought boujou plugins.

What is very interesting is the two lipstick camera s 12 in apart on top of the camera. This, if used in conjunction with boujou would give a much more accurate motion track! even positioning a third lipstick to the bottom of the sled would be incredible. If you wanted to, simply use 3 DV cameras (preferably progressive scan) instead. And create an average track through 3d max/maya before exporting to mocom, maybe even by mapping the three tracks to a 3D model of your sleds dimensions, and positioning your camera track at exactly where the film plane was balanced up on your sled.

Software that gives a motion track from the rushes, (using white balls or not) plots the relative speed of movement of points across the frame, thus estimating how far they are from u and each other. Camera point A being the equivilant of left hand eye and camera point B being the equivilant to the right hand eye, building up a 3D image of the environment and thus the movement. similar to one eye d people driving, building up a 3d image of the environment by the movement of the environment.

The great tip with using these softwares is a long track into the start position and then keep the motion going after the end position (cut) to give the software as much environment data to work on.

Howard, I think thats incredible being able to frame up on a live composite! wow! wicked
  • 0

#5 sebastian matthias

sebastian matthias
  • Guests

Posted 10 April 2005 - 04:14 AM

they everybody !
tank you for that quick reply !
we had looked into bujou,too.this really seems to be the right software.there is a newer version of it,called bojou three and some moco supervisors have told me about it. seems to be even better than the bullet.
i knew that we would have to place hundreds of trackingballs on the walls,but the idea having them hanging from the ceiling as well never occurred to me.so thanks for that input.
i really like the idea of using lipstick cameras fixed at my rig.could you send me more infos about that?and send me some pictures if you have any???please? i will need to prove that it works and has been used before on a set because i know what the postsupervisor will answer:
"the quality of the picture is not good enough to use it for tracking !"
<_<
it would be great ifyou could give me detailed information about that.
i also will try look into walters capturing system with the sensors at the camera.do you how it works?
it sounds abit like the thoma"walkfinder" that works with led-balls and a set of infrared-cameras riggedat the ceiling..would be great if it would be
usuable for bigger sets.

i hope to hear from you again !
thank you for all that input !

all the best to you! i´ll do some loopings for you on the next set!

sebastian
  • 0

#6 thomas-english

thomas-english

    Advanced Member

  • Sustaining Members
  • 1165 posts
  • UK

Posted 10 April 2005 - 04:49 AM

to give your average track between 2 or 3 tracks, make your sled (3d model of dimensions) stretchy!
  • 0

#7 thomas-english

thomas-english

    Advanced Member

  • Sustaining Members
  • 1165 posts
  • UK

Posted 10 April 2005 - 08:02 AM

even If you only had one DV camera on the top of of your sled alongside your big camera, It would still be better because you could put a super wide lens on it (giving it lots more data). And it would have deep DOF. Which you would nt want on your shoot camera.

I am going to be super interested in how your gig turns out.
  • 0

#8 sebastian matthias

sebastian matthias
  • Guests

Posted 10 April 2005 - 02:50 PM

hey thomas !
i allready talked to the dop about the dv-camera option and he thought that could be a good idea but the cameras would have to have prime lenses,so all avaiable cameras wouldn´t really work.(beside the xl1,but who wants to have 3 of them mounted to his rig?!) :blink:
so howards idea of using lipstick cameras is probably a better option.
ifind it really great that you are sointerested in that topic and will all ofyou keep informed how it´s going on.(at least what i´m allowed to sayabout thias projekt)
it will still take a while till we actually start to shoot (maybe beginning of 2006).
i really appreciate all that good feedback !
i´ll stay in touch!
all the best

sebastian
  • 0

#9 thomas-english

thomas-english

    Advanced Member

  • Sustaining Members
  • 1165 posts
  • UK

Posted 10 April 2005 - 03:12 PM

why would the DV camera (tracking camera) need prime lenses? I reckon a pdx10 with a superwide lens would be perfect, or even a microDV camera!

You need not get too stressed about having progressive scan either, interlaced does work fine with boujou. Progressive better but interlaced is fine.

email ed bolton on e.bolton*at*2d3.com and ask him questions. He s sound. Check that you can use 2 or 3 camera s to create a track in boujou directly, otherwise you would need to create 3 tracks and then model them through 3dstudio/maya to create your camera track.
  • 0

#10 Stephen Murphy

Stephen Murphy

    Advanced Member

  • Sustaining Members
  • 261 posts
  • London

Posted 10 April 2005 - 05:33 PM

Sebastian,

There is plenty of footage of Andrew Rolands on the Matrix DVD boxset, using the lipstick rig that Howard mentioned.
He's flying it on a Millenium XL and the Lipstick rig looks similar to a set of small antlers on top of the camera.
Hope this helps,
  • 0

#11 RobVanGelder

RobVanGelder

    Advanced Member

  • Sustaining Members
  • 919 posts
  • Bangkok, Thailand

Posted 10 April 2005 - 08:47 PM

Some years ago I operated in a virtual studio where the camera´s were equipped with a kind of "antlers", a collection of rods with balls at the end (emitting infrared light) on top of every camera.
The studio ceiling had several infrared camera´s and they were used to track the antlers in 3D, after which the computers put real-time background behind the actors.
We could handhold the camera´s and also put it on a mini-jib without problem, so a Steadicam should be possible too.
But it is possible that this system is already obsolete now.
Could it be from Ultimatte? I know the guys were using that system too.
  • 0

#12 thomas-english

thomas-english

    Advanced Member

  • Sustaining Members
  • 1165 posts
  • UK

Posted 11 April 2005 - 07:28 AM

Yeah, These external motion tracks are always going to be much more accurate.

There is a cheap way to do this if your a little handy with 3d studio max.

Build some antlers, (as rob said) and put ping pong balls on either end as well as the middle. as well as ping pong ball at the bottom of your sled.

Get as many mini DV camera s as you can, do not use any lens converters and be aware as to what their widest lens is. (you may have to use their 35mm equivilant, found on the net, or calculate the lens by using reference marks through 3dmax)

Position The camera s at different heights and locations around the set, and carefully measure their point, height and angle. A good way to monitor this would be to have a meter rule in the middle of the set as a reference image.

Have all these camera s on when doing your shot. Posibly use one of those synch clapper boards.

In 3dmax, position each virtual camera at the same lens in the same position on your virtual set. Then overlay each camera s video sequence (if need be tweak your virtual camera s lens and position by usinng your reference image with meter rule), and track your points from each camera plane automatically and tweak it manually. This will give you a very accurate motion track of your sled. The more cameras you do this accross the more accurate it will be. Be sure to cover as many angles and altitudes as possible.

If you do want to splash out, instead of ping pong balls, use Infa red Led s (a coupla quid from radio shack/maplins) and use sony dv camera s with Infared night vision. This will be even more accurate because it would be a point source rather than a relativly large ping pong ball.

This will be cheaper and more accurate but a lot more work!
  • 0

#13 sebastian matthias

sebastian matthias
  • Guests

Posted 11 April 2005 - 11:03 AM

hey everybody !
how am i going to get all that input into my little brain ? :lol:
thanks for the info of the matrix-dvd box.i´ll definitely look into it.

rob,the system you used probably was a thoma"walkfinder".that was our first option but we soon found out that itwouldn´t work in our studio because of the dimensions.this led/ifrared sytem covers only a space of 9x6 meters,a minimum hight(ofthe led balls) of 1,5 meters (no low mode)
and a maximum height of 2,3 meters.even if we combined 4 systems(wich would make it very expensive,too) we only could cover a space of about 250square meters(we need about 800 !).but it´s still an awesome system.and it works pefectly!

the idea with the pinpongballs/antlers is good too but again there probably are limitations in dimension (esspecially when you use wide angle lenses beause the dots just become to small,then further you are away from the camera.but maybe with the right amount of cameras and the right focal lens that´s possible.

some of the things you guys wrote i couldn´t get into the windings of my brain jet.i´m just a little stupid steadicam-operator

:blink: but i will try to understand.

thank all of you for those many infos

i´ll be back soon


sebastian


ps:thomas i just read your message again an now i understand!this idea
with the nightshot-mini dv might be really good !
  • 0

#14 thomas-english

thomas-english

    Advanced Member

  • Sustaining Members
  • 1165 posts
  • UK

Posted 11 April 2005 - 12:19 PM

you still are going to have the space size problem though! especially if your going to be trying to recreate those subtle nuances that make steadicam steadicam.

shoot the whole thing on the mocom!
  • 0

#15 jay kilroy

jay kilroy

    Advanced Member

  • Sustaining Members
  • 222 posts
  • Pennsylvania, USA

Posted 11 April 2005 - 09:01 PM

Some years ago I operated in a virtual studio where the camera´s were equipped with a kind of "antlers", a collection of rods with balls at the end (emitting infrared light) on top of every camera.
The studio ceiling had several infrared camera´s and they were used to track the antlers in 3D, after which the computers put real-time background behind the actors.
We could handhold the camera´s and also put it on a mini-jib without problem, so a Steadicam should be possible too.
But it is possible that this system is already obsolete now.
Could it be from Ultimatte? I know the guys were using that system too.

<{POST_SNAPBACK}>


Rob,

We are in the process of having Ultimatte install a "Virtual Set" in our studio. We have 3 studio build ups, 1 jib and 1 handheld/steadi. We have yet to put it on the air with all that, but I have built our rig with the crazy antler ball and all those wires. I hated every minute of it!!!!! On a ProVid none the less. I will be interested to see how that goes when we finally get on the air with it. At least now we have a Vector to put it on.

How did you like your Ultimatte experience??

Jay
  • 0




Omnishot Systems

SkyDreams

Engineered Cinema Solutions

Boland Communications

GPI Pro Systems

Betz Tools for Stabilizers

IDX

PLC - Bartech

Ritter Battery

PLC Electronics Solutions

rebotnix Technologies

Media Blackout - Custom Cables and AKS

Teradek

Paralinx LLC

BOXX

Wireless Video Systems

Varizoom Follow Focus