>Hello! I've got a commercial shoot approaching in a few weeks and I want to simplify my end of things. This series of spots involves compositing actors, shrunken down to around 2", into various sets...like "The Incredible Shrinking Man" stuff. We all have done this a hundred times..same old stuff. But I am always amazed at how time-consuming the alignments always seem to take. We shoot the sets first, in this case a desk top or whatever, I bring various model human figures [all the model train scales: N, HO, O, G, etc..] and use these figures to help set up the shot and to also roll a few feet to help the post house know our scale. We also always record the video assist and use a switcher to help up 'blend' the pre-recorded background plates with the full scale 'live' actor set-ups. I use inclinometers to figure camera angle, record ALL of the data [focal length, T-stop, film plane to actor, film plane to stage floor, etc...], everything. But, if I know the intended scale, say 'O'-scale [1/4" to the foot], use an 'O'-scale model figure in the background plate to calculate the eventual actor's size, could I not also measure the distance from film plane to actor, and the height from the stage floor, all in 1/4" to the foot scale to help figure out the actual camera placement when we do the live action portions? I tried this a few times in the past and always had to go way beyond my calculations? If I use a 50mm lens on a 'miniature' set [1/4" to the foot] does the 50mm lens retain the same perceived field of view when shooting full scale actors? Example: an actor [full scale] at 16' from lens, a 1/4" to the foot model figure 4" from the lens. That seems to be my dilemma...does focal length retain size regardless of scale? Any thoughts on this? All of the books I've researched all talk about scale size & time, not scale size and optics....
>Jeff, I always get confused by the maths involved in trying to 'scale match' two shots which will be comped together - in fact maths is not my strongest subject at all... One of the most important facts to remember about this type of shooting is:
>Lens focal length does NOT determine perspective!
>That's gotta cause some big arguments here but it's true.
>Camera POSITION relative to the subject is what determines a perspective match - (within limits) it doesn't matter what lens you use! (apart from extremely wide lenses which may have their own built-in distortion)
>If you are shooting your people blue/green screen it is usually better to make them as big as possible in frame so you can pull a better matte.
>Just make sure your camera is at the correct position relative to the people (yeah you probably should at least start at the correct 'scale' distance away from them) then shoot with a longer lens to make them bigger in frame.
>This technique has the added advantage of REDUCING the size of blue/green screen required to cover the subject.
>Hope I haven't totally misunderstood your post and gone off at a tangent... Cheers Steve N.
>Lens focal length does NOT determine perspective! That's gotta cause some big >arguments here but it's true.
>Its both lens focal length (angle-of-view) AND the distance from the subject that determines true perpsective, distortion, etc. You wont get the same perspective on a set with a 50mm as you did with a 14mm (with distances compensated to allow equal field-of-view), unless you build a forced perspective set. Right ?
>I do agree about shooting oversized though and scaling down in post.
>Mark Doering-Powell LA
>(with distances compensated to allow equal field-of-view)
>I agree with that - but if you leave the 'equal field-of-view' factor OUT of the equation, the 14mm frame will contain the 50mm frame within it and both frames will share EXACTLY the same perspective! - In fact if you removed the 50mm field of view from within the 14mm frame and enlarged it, that would result in exactly the same frame as if it were shot with a 50mm from the same camera position. (if you ignore any distortion inherent in the 14mm - or grain from the blow-up)
>I know... sounds bloody pedantic doesnt it? The only reason I make the point here is that many people (and I include some cinematographers I've met) think that the lens itself creates the perspective of the shot regardless of camera position - it aint magic, perspective has simple rules.
>The only reason it's worth getting our heads around this is because when we are shooting two elements to be matched into one final composite, we rarely need to match the field of view - it is so easy and beneficial to reduce the size of one element in post - but the perspective MUST be damn close or it'll all look like bullshit!
>Sorry Mark, bit more of a rave than I intended... Cheers Steve Newman ACS
>Been too busy to jump in on this one till the weekend but I did want to make a few rambling comments.
>Jeff, numerically your question of scaling the set SHOULD work since you are matching your on-set shooting data so accurately. Focal length does retain scale. It's the LENS DISTORTION that is the wild card here and it's a big one too. I go through the same exercises that you do when I'm gathering info on full size sets. Mostly we are working from 1st unit's shooting notes. It's gotten better lately, but more than half the time we never get the info we need. On the recent Monkeybone shoot most of the 1st unit on-set data that came in was meticulous for the camerapersons shooting the bluescreen stop-motion monkey puppet, but an entire 2ND unit sequence that was shot with the lead actor on bluescreen for placement into my miniature set was basically ignored. (not to mention handheld)
>When we have the numbers, we line everything up and sometimes things fall into place and it's magical. Much of the time though we throw out all that info and just make it look good. (How many times have you done that with preproduction plans?) This is something I can do pretty quickly these days.
>On a recent Listerine commercial and actor walks on a bluescreen to be inserted on some computer generated teeth with whitewater rapids raging below. The post house supplied the model co with precise plans for constructing 4' tall teeth for us to shoot miniature rapids wrapping around and crashing into the molars. CG animatics had been done of the entire spot so to shoot the highspeed water we received the most detailed camera position data I had ever seen. Things fell into place like butter. STILL though, two of the dozen shots just had to be "found"!!?!
>When matching lens distances, I might move the camera into the subject and go a little wider on the miniature set's lens choice, (It is RARE that I would shoot a miniature BG with a lens as long as a 50mm. Especially at your incredibly small 48th scale 'O' model.) As a general rule I'm at less than 35mm most of time. Besides needing a wider field of view closer up, I'm also trying to simulate real world depth of field as well.
>When scooting things around to line up, distances may vary but the angles are sacred --unless they are just plain wrong. We also tweak the lighting to have more complexity than the light on the full size set. ("exaggerate all the detail" is a favorite model cry of mine, and it includes me too.)
>Steve and Mark have been right on in thier comments in maximizing the subject's size in frame to benefit compositing. It's easier to pull mattes with the tighter grain of a larger image and it's always easy to degrade the reduced final element after things have been fit together.
>Steve N. said:
>Just make sure your camera is at the correct position relative to the people (yeah >you probably should at least start at the correct 'scale' distance away from them) >then shoot with a longer lens to make them bigger in frame.
>To clarify :
>This is an advantage if the camera is locked and the lens is changed INSTEAD of changing the camera position relative to the subject. Otherwise the screen size would be constant for the subject (not frame) coverage.
>Mark D.P. said:
>Years ago there was a great example in a Professional Photography magazine >showing the above example..
>Mark can you dig that up? I'm always looking for ways to clearly explain concepts like this to others.
>Robert Rouveroy wrote:
>I wonder where the d.o.f has disappeared to if you do that long lens compression by >cropping in.
>It's still there. It sounds like your empirical test was from a frame that recorded a depth of field sufficient enough to hold both subjects in focus.
>The question we don't often have to ask is: Do we want to resolve with optics or grain?
>Eric Swenson VizFxDp
>Steve wrote :
>I know... sounds bloody pedantic doesnt it?...it aint magic, perspective has simple >rules.... Sorry Mark, bit more of a rave than I intended...
>Steve, I loved that rant.
>You're right that if you crop out the tighter 50mm frame out of the wider 14mm then you get the same thing as if you shot the 50mm from the same distance... but then you add: "if you ignore any distortion inherent in the 14mm" and to me therein lies the rub. Lenses do have distortion and when you deal with wide angle plates and VFX I think it becomes paramount to match that at times.
>Years ago there was a great example in a Professional Photography magazine showing the above example... and its a great way to think about lenses - that each focal length just picks a different angle-of-view through a sphere of glass. blow up the 14mm enough and its a 50mm shot. Even the long-lens compression is there if you crop in far enough !
>Its all a matter of angle-of-view.
>And my favorite mnemonic device on the subject I heard here from some wise VFX CML'er:
>"match your angles and scale your lines"
>Who was it that came up with that, short, elegant mantra ?
>Mark Doering-Powell LA
>Mark Doering-Powell wrote :
>blow up the 14mm enough and its a 50mm shot. Even the long-lens compression >is there if you crop in far enough !
>I wonder where the d.o.f has disappeared to if you do that long lens compression by cropping in. I tried to crop an 8x10 but the two subjects, 30 feet from each other, were still both sharp even at the largest zoom-in I could achieve. Explain? You are probably right, I don't dispute optical laws, yet an empirical test shows otherwise. -- Robert Rouveroy csc The Hague, Holland
>Mark quoted :
>"match your angles and scale your lines"
>I love it!... but I'm not too sure it explains enough to the people who we seem to constantly need to explain this stuff to!
>How many times have you guys had to repeat: 'Pleeeease, can we shoot the background plates first because they are exteriors!'
>How many times has the answer been: 'Why?'
>I know, sometimes that decision has been made (back to front) by scheduling circumstances and we MUST shoot elements in reverse order...on Babe II the sky plates were shot AFTER the bluescreen pelican and duck elements and I found myself briefing the chopper pilot about our orientation to the sun and then shooting with timings provided by an audio cue track which had been speed shifted to allow for the different frame rates used on both the f/g and b/g elements! Crazy! But, and this brings me back to the reason for the post, usually we are shooting foreground elements to be inserted onto backgrounds. Matching the 'field-of-view' rarely enters the equation but the perspective match is still critically important. Hi Erik! Cheers Steve N.
>...briefing the chopper pilot about our orientation to the sun and then shooting with >timings provided by an audio cue track which had been speed shifted to allow for >the different frame rates used on both the f/g and b/g elements!... Crazy!
>Which brings me to my 2nd favorite: "shoot your least controllable element first" (then match the other elements/plates to that)
>I also like Eric's "exagerate the details" ... that sounds like pure artistry and finesse.
>Mark Doering-Powell LA
>I also like Eric's "exagerate the details"
>Except the details in the distance of a miniature meant to look like full size...
>'You want to do what ?... fill the stage with fog ?
>It can sometimes be difficult to explain chiaroscuro when I'm not even sure how to spell it ! ...well, 16 feet has to look like 6 miles soo....
>Hey! just looked up the spelling and it is correct - but I'm not too sure it's the right word to use .. the definition is: 'the treatment or general distribution of light and shade in a picture' ..which isn't quite what I'm trying to say.
>6 miles of (dirty) air knocks out a lot of contrast like a diffusion filter except fog gives you diffusion-in-depth, something you can't get from any diffusion filter I've ever seen.
>We better quit this thread or geoff will relegate us to 'miniatures 101'
>Cheers Steve N.
>Steve Newman ACS wrote :
> .... "exagerate the details"
>> Except the details in the distance of a miniature meant to look like full size...6 miles >of (dirty) air knocks out a lot of contrast like a diffusion filter except fog gives you >diffusion-in-depth...
>No exceptions Steve my friend, you simply exaggerate the atmospheric density.
>It's all in your detail choices.
>-Spent 6 months last year in said smoke.
>.... "exagerate the details"
>>Sorry, but I missed the original discussion. Which details are to be exaggerated?
>My comment was, "exaggerate the details." It means that good miniatures are built and dressed to have more complexity the thier real life counterparts. The Miniatures DP's job is to continue that intricacy by adding extra embellishment to the light and shadows, going a little wider on the lens choice and carefully scaling down compound camera moves. When done in symphony, the illusion of scale is seamless.
>Knowing exactly which elements to enhance and when it's overdone is what comes with time.