The n stages of VFX post production
If you’re a student, you’ll probably have a rough idea of the stages that a film goes through before it hits the cinema screen. The one thing that a school or college always seems to miss is the sheer work that goes into every single frame.
Production is where you do the dirty nasty hard work of actually filimng everything. They typically ponce about in posh cars, have chefs on tap, and genrally live the life. If you’re interested in post production then you must be attracted by the long hours and the lack of natural light.
However there is a need for someone well versed in VFX to be on set to help consume the champagne, and also put the markers in the right place.
Stage 1: Ingest
So the director has shot everything, the VFX supervisor has made sure that the green screen is green, and the light probe images really do match up. The film’s producers have found some companies to bid for the shots that need VFX.
Now you need to make sense of the images and assets you’ve been given. This stage is called ingest, as you need to figure out how to force these images down the throat of your pipeline without giving it, or your self indigestion.
Its at this stage you perform a scan the negatives, clean up and fix any damage or dust that are on the frames (also known as plates). With the rise of wholly digital film cameras you replace dust with codec munging.
All the footage is then dumped into an asset management system, along with anything else that you were given (lidar scans, light probes readings, prebuilt models, etc)
Its during this stage (or whilst bidding for shots) that you do your research, like figuring out how to model a realistic looking 90 meter standing tidal wave. Its also the time where you get your environments modeled and main CG elements knocked up.
Stage 2: Pain and Roto with Tracking
As a student there is a good chance that you’ll end up doing a spell here. This is the stage legions of people do exciting things like painting out zits from bespotted teenage actors. This is a very manual process, I cannot emphasize how manual it is. Every single frame which has visible zit/lines/bags/scars/bald patch must be corrected by hand. There is no magic “hide Mr Cage’s massive widows peak” button.
Rotoscoping is where someone traces round the edge of an object to separate is from the rest of the image. That wikipedia article is talking about animation rotoscoping, which is a form of cheating. However a former colleague of mine Conrad Olson has a much better explanation:
Rotoscoping involves ‘cutting out’ parts of a live action plate that the compositors need separately from the rest of the shot. This needs to be done if the objects couldn’t be shot as separate elements in front of blue or green screens.
The process involves tracing around the objects on a particular frame and then animating the shape to match the outline of the object as it moves through the shot. This can be a slow and manual process.
Painting is where items like safety rigs, tracking markers, personal deficiencies and deficient personnel, are removed. This stage can be also known as the “Fuckit fix it in post” stage. For some tasks there are magic plugins that do things like remove rigs automatically. For more complex tasks like marker or body part removal, you’re left with cut and paste (clone stamp.)
Tracking, is the process of marrying up the real world camera motion with the 3d world inside maya (or 3DS max). Basically you select some points in the video that you know don’t wobble or move (walls, fences houses, etc). You then run the tracking software which assuming you’ve picked enough points will work out whats moving where.
Although there is an “automagic” track feature in most tracking programmes, you’ll still need to do a lot of manual tweaking to make sure that there are no wobbles in the camera movement. Ritch has some very good worked examples of how tracking works in practice
Oh dear, there are quite a few more stages:
There are many more stages, and they will have to wait till part two…