Home of Professional Cinematography since 1996
Composites Of Different Scales
class="Paragraph" Published : 20th April 2004
I'm preparing for a shoot in a few months that presents a problem I haven't yet dealt with : compositing elements of vastly different scale.
To explain briefly, there is a character in the film that is just six inches tall, and he needs to interact with a real world environment and characters that are normal-sized. None of what we've planned is too complicated (characters never touch one another, for example). Basically we just need to have the small man running around the kitchen table, and the floor, and in a shoebox. The logistics of each shot I'm enjoying figuring out on my own, but the overall process of how to go about doing it is what is stumping me.
In a nutshell, I'm concerned about some things that I just can't be sure of, like matching camera angle and focal length characteristics.
My thoughts so far :
1/. I should use the same focal length lens for both shots of a composite and simply move the camera back until the actor is small enough in frame to match when shooting the "little man" in the studio.
2/. I should compose each shot with the actor (playing the "little man") as large in frame as I can afford to make him, as we can always shrink him down in post.
3/. I should have a protractor on set and measure every angle I can think of (and then some), and be sure to replicate each exactly when we go to the studio to shoot "the little man."
4/. All shots must be locked-down, sans movement, because we have no access to motion control.
Am I right in these lines of thought? Any advice on efficient, practical ways to match angles, particularly the tilt angle? We're shooting, processing, and transferring all of the "real world" footage before we shoot any of "the little man" so that we can make comparisons on set with regards to lighting, angles, etc. We'll be sure to take meticulous notes regarding angles and heights and focal lengths etc. and follow these religiously. We'll be shooting 16 (or perhaps Super16) but finishing on BetaSP or DigitBeta, with perhaps the possibility of a 35 blowup from DI (but that's mainly a pipe dream).
Sorry for the long post. Am I correct in any of these assumptions, or am I looking at this all wrong?
Radio-TV-Film Senior, Northwestern University
>To explain briefly, there is a character in the film that is just six inches >tall, and he needs to interact with a real world environment
>In a nutshell, I'm concerned about some things that I just can't be sure >of, like matching camera angle and focal length characteristics.
All the stuff Mark "analog inclinometer " W. said, plus this :
Get yourself a doll which stands up by itself and is very close in size to the ratio you will want in your final composites, and use it liberally while shooting your backgrounds. Always squeeze off a few seconds with the doll in the shot in position for every setup. You will be amazed at how much this helps with your line up issues later.
I recommend the Barbie line; her new spring fashion and accessories collection is to die for.
I vaguely remember an article in (I think) American Cinematographer several years back about the film "Indian in the Cupboard". If memory serves, the article went into great detail about how the filmmakers dealt with the very issue you're having. There is even a description of how they recreated a semi-circle dolly move that in the "normal" sized plate shot was 10 feet or so, but in the "little man" shot would have been a 100 foot or more move in just a few seconds. They did this by putting the actor on a turntable and using a ring of lights around him controlled with shutters to fade them up and down to make the direction of the source appear stationary in relation to the actor as he turned. Not sure if you're interested in anything so complicated, but it might be fun.
Maybe some on this list has a box of old AC's that they might be able to dig out. Or maybe the AC has an archive of old articles online?
Hope you can locate the article.
>I vaguely remember an article in (I think) American Cinematographer >several years back about the film "Indian in the Cupboard".
August 1995 issue.
David Mullen ASC
Cinematographer / L.A.
>1/. I should use the same focal length lens for both shots of a composite >and simply move the camera back until the actor is small enough in >frame to match when shooting the "little man" in the studio.
Focal lengths and angles match, distances scale, so in a sense, if your 6' tall man is supposed to be 1 foot tall, for instance, (making him 1/6th scale) you could measure the distance that he should have been at in the background plate, and multiply it by six when you line up the shot in the studio which should get him the right size...but remember, you have to multiply ALL the measurements by 6, so, for instance, if the camera was3 ft off the ground and at horizontal, you would want to raise the camera to 18 ft off the ground to put your little guy in the right place in frame. Now, if your little guy is standing on a table, for instance, and the table is 21/2 ft off the ground and the camera is 3 feet off the ground for the room element, then when you shoot your "little guy" element on the stage floor, the camera would want to be3 feet off the ground (which would be 6" off the table times 6x for the scale.)
With me so far?
>2/. I should compose each shot with the actor (playing the "little man") >as large in frame as I can afford to make him, as we can always shrink >him down in post.
Well, yes and no.
If this is your first time doing this, you should start by figuring it all out as I have outlined above which would get your little guy the right size and in the right place on the film. Perspective is determined by position, not by focal length, so once you have the camera in the right place so that everything matches, you could MAGNIFY the image with a longer focal length lens to have a bigger neg to work with, but it will make the post end more complicated - by making him bigger, you also make his translational movement in frame bigger - two steps makes him move further across the smaller magnified frame than it would with the "right" lens. While this is all fixable, it is more complicated - you should talk this over with whoever is going to do your post compositing to make sure they can handle it. It might be simpler to shoot the right size and know that it will all fit together. Spending the time to get the lighting right and the bluescreen or greenscreen lit as evenly as possible might do you more good than magnifying the little guy...if he has to be "tracked in" to position in post because of the size change, it might look more fake, even if the edges are better. It is hard to make a strong specific recommendation on this without knowing more about the specific shots.
Bear in mind that in the imperfect real world of lenses, changing to a longer lens to magnify the image may also make for superimposing two images shot with different kinds of lens distortion causing another mis-match issue.
>3/. I should have a protractor on set and measure every angle I can >think of (and then some), and be sure to replicate each exactly when we >go to the studio to shoot "the little man."
Most critical is the camera height and tilt angle and distance to subject - making the little man floor plane match the plane of whatever he is walking on in the background plate is critical - if he is walking around, for instance, you really don't want him to start floating or sinking through the background plate floor. Pan line up can be by eye
>4/. All shots must be locked-down, sans movement, because we have >no access to motion control.
Safest way to go unless you build yourself some "poor man’s motion control" stuff, but consider what sort of difficulties you would be adding to your life and how much production value you would be adding to the shot...simple is better, especially if you are of limited resources.
Most of my colleagues use digital levels to find tilt angle, but in my perverse, analog way, I use a Suunto clinometer (approx $100 in the US, slightly more in "high VAT countries."). The clinometer allows you to interpolate to a quarter of a degree or so which is close enough and as accurate as the digital levels, and the clinometer itself is small enough to keep in a shirt pocket and small enough to go on a flat spot on the camera (I like the matte box rods for this). Harder to read - you have to get your eye right up to it, but it also never runs out of batteries...It's also good for sun angle, shadow length, height estimation etc. The model I use is a PM-5 360 PC. High end camping/climbing stores sell them, but I bought my last one at http://www.forestry-suppliers.com
Record your video tap and use a small cheapo mixer with wipe and dissolve capabilities to match your "little man" element shot as accurately as possible to you background plates. Note that if you do not use the same camera with the same tap, or if the tap has been adjusted or moved, you will not be able to line up accurately without repositioning one image and/or resizing it slightly. Using the telecine dailies won't work because their scale will be off - use the tap image from the first shoot with the live tap image in the studio.
Remember depth of field!
If your little guy walks 6 real feet towards camera it is only 1 foot on the background plate - if your depth of field would keep that whole 1 foot in focus , then to be perfectly righteous about it, you should be working at a stop that lets him stay in focus on the stage for his full 6 foot walk. You can cheat here if you have to by pulling focus on the "little man" element, but do keep the issue of "what would be in focus and what wouldn't be in focus" in mind as you design the shots
Disclaimers - others on this list may disagree about the magnification issue, and etc Your mileage may vary.
Good luck, and tell us how it works out...
VFX Super/DP LA based
(and ex NU faculty brat)
Bob Kertesz writes :
>I recommend the Barbie line; her new spring fashion and accessories >collections to die for.
Definitely falls into the "too much information" category, Bob.
Randy "put away my dolls years ago" Miller, DP in LA
Well Mark has basically said it all...
The only area I'd disagree with is about pulling focus, it can work but a lot of lenses shift image size/position as you focus and this will cause you horrendous problems later.
I use a Suunto as well, the dual one that include a compass in my case.
For this job :
class="Paragraph" we were shooting on what was effectively 8 sets on 3 stages,
to match the lighting I made a drawing with all the angles
of the lights on it together with their intensity in FC and
their colour temperature.
To make sure that the key was exactly the same in every set I put a pointer at the top of this drawing that had to be matched to a set reference point, there was then a position that a piece of dowel had to be placed and the light adjusted until the shadow of the dowel filled the shadow that was drawn on the paper.
This was a remarkably crude technique but worked extremely well, this drawing was mass duplicated and every member of the 3 lighting crews had one as did all the camera crew and there were dozens of spares pinned to the studio wall!
We also did rough drawings of what each level of the shot was supposed to be and pinned a Polaroid to them as we did each one, just to make sure we didn't miss anything.
Geoff Boyle FBKS
Director of Photography
Quoth Geoff :
>The only area I'd disagree with is about pulling focus, it can work but a >lot of lenses shift image size/position as you focus and this will cause >you horrendous problems later.
I don't disagree with Geoff at all on this : if you can make the stop to avoid having to pull focus on the "little man" element, you will be in much better shape...the fewer cheats and adjustments you have to make in making the two elements work together, the fewer little things might creep in to give away the gag.
If, for instance, you start cheating the lens height or the tilt angle "just a teeny bit" you may find that the floor plane doesn't quite line up, and while you can look at the video tap images mixed together and talk yourself into saying "it's close enough" and when you are looking at the shot on a monitor while it is being composited it might look "not too bad" when it is blown up and projected, it might end up with the perspective mismatch becoming much more obvious and painful.
As Geoff has pointed out, a lot of lenses breathe and shift as you rack focus, and since the object of the game is to have the "little man" element and the "room" element look like they were captured as a single element on one piece of film, there is danger that ANY breathing of the "little man" element with respect to the room element will be visually jarring - much more jarring than if the whole image were breathing or shifting together.
Knowing ahead of time that you have the depth of field issue to deal with, you can design the shots to minimize the challenge.