Friday, February 25, 2011

Part 2 of 2 understanding CGI integration

The point of this blog is to get you to start thinking 3D integration rather then educating you on any particular technique.  Compositing and 3d Computer graphics design takes years to master so I wright this with no intention of saying this is the end all be all of how to view and tackle CGI.  However, I will say that understanding some simple concepts can help you tackle even the most challenging of visual effects.  



So where do I start? Lets start with the most common method of compositing and that is keying.    Also known as color keying, croma keying, blue screen, green screen etc.  The idea behind keying is to cut out a real world object from an image and pasting it on a background and possibly overlaying it with a foreground.  That is about as complex as the idea gets.  However when things get more complex we start talking about depth of field, motion tracking, color spill from the green screen etc.

The more movement you have in the shot or the final composite the more challenging it will be to composite it and make it look believable.  Even a simple hand held camera can be a nightmare for some post people.  So if your going to shoot green screen ask your self how much movement does this shot really need and does it fit the story.  Any good compositing takes as much time as say a traditional animator would because you may have to manipulate each frame individually with a garbage matt.  cutting out any unwanted elements or what have you.  Good keying will solve any major problems in post.  The type of keying you go with will ultimately determine how you deal with it in production.  keying in post stays relatively the same sense your telling the computer that the croma keyed background is to be cut out and made in to an alpha matte.  I would explane all this stuff but I'd be writing a whole book just on explaining rather then actually expressing how to tackle CGI from a real world perspective.    

Back in the days when compositing was relatively in its infant stages studios would use an all black back ground to key out there subjects.  Disney did this when they made Mary Poppins.  You can see a good example of this in the DVD version of the movie where they show the behind the scenes.  They used black to minimise spill and it was easier to key out for the most part.  Later did we start using blue screen and green screen.  The idea behind blue and green screen is that it is an odd color and does not really match anything in the shot.  hopefully.  The major problem with any color key is spill and unwanted color reflection.  there is a number of ways to tacke the situation.  The first of witch is to put your subjects farther from the background second is to light the croma key from the back, but you need a special fabric that allows light to illuminate it from behind.  Its not a %100 but it works for the most part.

when color keying the matt that your using must be perfectly lit with no wrinkles or stans or any thing.  using a light meter and soft keno lights will do the trick.  You can also use daylight balanced shop lights  but remember the entire screen must be lit balanced for the keying to work.  Bad lighting on a croma key back drop will show up like a sore thumb in post. Its better to reshoot bad lighting on a color key then to fix it in post.

One of the bigest problems I see in making keying realistic is matching the back ground match the subject in compositing.  One of the reasons these shots don't match typically is color temperature.  The color temperature and direction of the light must precisely match the back ground.  Even the quality of light must mimic the background.  If you shoot a shot outside then odds are you will need to set your croma key out at the exact time of day at the exact position of the shadows shot in the back ground.  This is the least complicated way of eliminating the problem of lighting continuity.  As I said it before every thing must be controlled and precise if you do any level of CGI or compositing.

I might be nit picking on this but I believe image size will effect the realism of the final composite.  The main reason is that the larger the image size on the camera the more depth of field it will generate.   Also image size can change prospective a little bit even if the lens was matched to be the equivalent of the background for the final composite.  So grabbing an image off the web in hopes that it will work may not work at all realistically.  The reality is your audience can and will most likely notice that the image has been composited in some way.  If you don't care or if the content your compositing does not need that level of realism then go for it.  The ideal situation is to stick with the same camera and medium throughout the shoot even for convenience sake just work to make sure both cameras are the same shooting the same medium with the same lens.

lens focal length has a huge effect on the final composite.  You can't expect to use a 100 mm lens and go in to the studio and use a 50 mm lens and think it will work.    The back ground and the subject must have the same lens and focal length and the focus distance must be identical to that of the back ground.  If your subject is 5 feet away shooting with a 50 mm lens then focus at 5 feet and and do the same for the composite background.  I don't care if the back ground is out of focus or not it just has to match.  The same goes for the lens aperture.  The more open your aperture the more depth of field you will have.  So take extensive notes when filming for CGI integration.

Also lens hight and angel will greatly effect camera perspective.  If you shoot a high angle shot for an action sequence then expect to match camera hight angel and distance from the subject to the background exactly the same.   So both shots must be shot identically with the same medium and camera if there is to be any compositing.  

So if your shooting a long shot then expect to do the same when keying out your characters.  lets say your shooting a car jumping over a bridge and you want your character to be under that car and the shot is 50 feet away.  Then shoot the car jumping over the bridge and at a separate more controlled location shoot the same shot 50 feet a way with a green screen larger then the subject and key it out in post.  It sounds simple but every thing must match.

In after effects there is a way to change image size and focal length so match that with your cameras lens size and image size if you want to maintain visual continuity.

If your shot requires movement at all no matter how complex you must decide what will move the camera or the subject.  If you have both the camera and the subject move then you are going to have a more complicated time in post geting the shot composited right.  The more movement you have the more complex it will be in post to get the final composite.  If you are shooting movement of a simple object such as a spaceship it is better to have the model spaceship stationary and have the camera move but if you are shooting an object like a watch on an arm then it is better to have the camera stationary and let the watch on the arm move a little.  Movement that is more controlled with less movement is ideal.  But be ware complicated shots will cause a lot of stress in post especially if the shot is not planed. Keep any shot you do as clean and plan as possible.  That means no fog or random objects in the way or what have you.  You can add the effects later in post if you want.

Sometimes doing effects in camera is easer, but at the end of the day you have less control in the final composite.  The main advantage is that you can controle prospective.  The early star wars did this and it worked.  Especially when using miniatures.  In the case of miniatures, the rules change because you need to compensate for prospective.  That means lens size lens hight and angel must be relative to real world prospective.  Humans are used to seeing things from a certain perspective so you have to compensate for that perspective when in production.  Both shooting miniatures and shooting for CGI cost about the same no matter how you look at it.  There are pros and cons but its all preference.  Personally I have no qualms about ether or.  In the end its all about telling the story.

In the old days when CGI did not exist people would use glass and have artest to paint on that glass to make the effect look real like adding a city in the middle of a desert.  Or you can have a person over looking the ocean and seeing pirate ships off in the distance.  However, Those days are long gone so don't expect to see that FX any time soon.

There is new croma screening technology available that uses a gray screen with led lights attached to the camera.   It eliminates a lot of the problems as described before, but the led must be attached to the lens.  The light emitted is green and or blue so it could cause some color spill.

CKL-200 SD Dual-color Chromakey Light System

  
I believe "Van Helsing" the movie used this technology, but the draw backs is that you have to have the screen some what on a straight angle and the camera has to be relatively close.  There are more specs on this product.  From experience this product works, but it is not perfect.  Just an alternative.  

you can find out more information here.

LD-1 Dual Color LED Chromakey Ring

Another method of CGI in compositing when integrating the real world to the real is using digital wire meshing.  Essentially digital wire meshing will creating a digital plan you can put any where you want the effect to be placed.  for example you can put a wire digital mesh on a persons arm and make it look like he or she has something growing out of it.  Using markers to track the image you can keep that digital computer generated image on the subject.  

BLACK SWAN Featurette: Visual Effects Reel

There is nothing wrong with combining different visual effect techniques to as long as you can tell the story and the audience does not become distracted by the effects.  If the audience does not know the effects are there then you have done your job.  Causing attention to the screen rather then the story will cause the audience to become mentally jared and it could take 20-30 minuts for them to get back in to the rhythm of the movie.  And yes one bad or jaring cut or graphic can cause an audience to loose focus of at least 20 - 30 minuts of the film.  Even for a split second a bad cut or graphic can be devastating to a scene.  

if you ever get a chance read Walter Murch "in the blink of an eye"

In the Blink of an Eye Revised 2nd Edition

He explanes the phenomena of a bad cut on a psychological scale very well.  "In the blink of an eye" is probably one of the best editing books I have ever read and I don't think there is another book like it.  I know this blog is about CGI integration but the reality is if the audience feels any negative disturbance in the image then then the audience will in fact loose focus of the story as a whole.  and there is nothing worse then loosing the audience during a movie when the objective is for them to become engaged with the story. 

There are so many combinations to approach CGI integration and compositing.  The trick is to break it down in to very simple concepts.  The VFX your story calls for will determine how you will shoot the scene.  If you are a visual effects artest or are working with a visual effects artest let him approach the situation in the best way he knows how.  Trying to reinvent the wheel will cause all sorts of problems in post, time and money. Best idea is to start planing with your VFX guy early and have him on the set and along with you every step of the way.  The VFX guy is the last person you want to leave out of the loop.  The more involved your VFX guy is the better the end product will come out.  Technology is always changing and sometimes for the better, but believe it or not these seemingly complex concepts are fundamentally simple when broken down.  So worry about the story first and formost and when it comes time see what adjustments to the scripts make them then.  Because finding a creative visual solution to challenging problem is always better then working with a bad script.      

By 
Robert Sawin   

0 comments:

Post a Comment

Total Pageviews