I am prepping a 3D movie where the director, who has a lot of experience in shooting miniatures (I don't) wants to do that for this movie, but talking to a stereographer about this, he warned me that it is very hard to make this actually looking good in 3D. We will test this in the coming weeks but I was wondering if there are people who can chime in on this. On a side note, while watching "Valentine" I was distracted by one wide shot of the mining plant which looked miniature while it was shot for real.
Thanks in advance.
Mick van Rossum, NSC
Jakarta (for now)
>>he warned me that it is very hard to make this actually looking good in 3D.... while watching "Valentine" I >>was distracted by one wide shot of the mining plant which looked miniature while it was shot for real.Whatever interocular setting you shoot, the audience will relate to their own physical interocular. In the case of your miniatures, this means that the interocular must be scaled to the scene, which generally means a mirror rig, but then you have a mirror box in the way. Though it can and has been successfully done, examples being stop-motion animation like "Coraline", and in an extreme case "Bugs 3D!" (specialized optics). In the second case, that's the effect of hyperstereo, or greater than normal interocular, which miniaturizes the scene. Hypostereo, less than normal, gigantisizes. Tim Sassoon
Coraline was great, right?
Timothy Sassoon wrote:
>>“Though it can and has been successfully done, examples being stop-motion animation like >>"Coraline", and in an extreme case "Bugs 3D!" (specialized optics)”Bugs! Producer here... In addition to Tim’s remarks...Coraline was a single camera, on a motion control rig, to shoot left, move camera, change convergence and shoot right. Bugs! was a dual mirror-snorkel system, to get very wide angle field of view with large apparent dof. So, if the whole set is full of miniatures the Coraline way works if nothing is moving. If things are moving on wires/tracks etc you will need a dual camera set up of some description. In the second case, that's the effect of hyperstereo, or greater than normal interocular, which miniaturizes the scene. Hypostereo, less than normal, gigantisizes.” We have recently shot some 3D aerials over London with very wide interaxials (about 32-36”) and little or no miniaturisation. The trick so as NOT to get miniaturisation is to keep ALL of the image behind the screen plane. Things will look a little smaller, but if you try and bring things into theatre/audience space then the scale will instantly become apparent and the brain will go, “That can’t be a real size factory, as it is coming through a window only 12m wide!”.
The problems with shooting miniatures in 3-D include the impossibility of scaling down your camera system to create a small enough Inter axial (interoccular) to prevent.... miniaturization with a side by side, and the likelihood that a mirror rig would not get your lens close enough to the subject to work right, even if you have the io and convergence scaled properly.
One solution for this which is relatively easy to deal with conceptually, would be to us a motion control rig to shoot your two eyes as separate passes with the offsets built into the program. This would absolutely take a "real" motion control programmer who can do all the arithmetic, not just a so-called repeatable head... but it can be done. With stop motion work, people use rigs with little linear bearing bars and just push the camera back and forth (that is to say left and right) to get the two eyes.... this will not work with explosions, obviously.
I hope this helps a little bit, at least
Mark H. Weingartner wrote:>> ....motion control rig to shoot your two eyes as separate passes with the offsets built into the program. >>This would absolutely take a "real" motion control programmer who can do all the arithmetic... Seems to me this would be a good candidate for a previz approach, with simple geometry (built to scale) in CG as stand-ins for the miniature set pieces, and without the physical camera restrictions. The previz would allow for a full 3D preview as well. When the move is locked, it could be translated to mocon coordinates (a reasonably common practice today) and shot in either one or two passes (a single pass using a L/R interleave). Come to think of it, isn't that the way one would likely do this? Mike Most
Michael Most wrote:>>Seems to me this would be a good candidate for a previz approach, with simple geometry (built to >>scale) in CG as stand-ins for the miniature set pieces, and without the physical camera restrictions. I recently attended the Dimension 3 expo in France and saw a demo by Ken Schafer of a VERY COOL stereo 3D previz tool :
>>>and the likelihood that a mirror rig would not get your lens close enough to the subjectWe broke a beam splitter on Capt. Eo shooting miniatures up close and in your face ... so it IS a real problem.
My suggestion is to call Fantasy II film effects in Sun Valley CA. Gene Warren Jr.(owner) has a great deal of experience in all things miniature including 3 D. Among others I can think of he did "Spacehunter Adventures in the Forbidden Zone" http://www.imdb.com/title/tt0086346/.
818 843 1413.
On Terminator 2 3D (the ride film still playing I believe), we shot miniatures using motion control, single camera. Each pass had right eye and left eye elements, easily calculated by the Kuper motion control software, the additional 3d commands written specifically for that show.I don't recall there being any major issues with compositing. We did 2-3 different scaled models within a single shot. The work was done at Digital Domain back in 1995-96. I was a 1st assistant during the live action, pulling the 3d effect per Peter Anderson, ASC and later, the motion control camera operator for miniatures. Additionally on our stages, the Skotak brothers shot larger scaled models using a smaller lightweight over-under rig developed at Digital Domain by Scott Salsa with supervision by Peter Anderson.
>> ... and shot in either one or two passes (a single pass using a L/R interleave).As with everything... totally situational... with stop motion, left/ right alternate shots would be the obvious choice. With go motion, depending on frame rate (shooting speed) and rig speed and accelerations, it might make more sense to shoot the two eye passes separately or it might make more sense to shoot alternate frames... and it would depend on whether you were shooting film or files as to the convenience and advantage of having two strips or one - simply from a data-management standpoint for scanning.... etc Mark H. Weingartner
I'll second Phil:
I saw FrameForge 3D too in Dimension3 and it is really cool!