Joint Camera and Object
Tracking
カメラとオブジェクトの両方がショットの中で動いているならば、あなたは各々をトラッキングすることができ、同時にそれらを解析して、3Dシーン中で動き回っているカメラとオブジェクトでシーンを製作できます。高品質のソースならば、いくつかのオブジェクトは、カメラと同時にトラッキングされるかもしれません。最初に、あなたはオブジェクトと背景を区別するロトスコーピングまたはアルファチャネルを設定しなければなりません。または、両方とも手動トラッキングで実行しなければなりません。いずれの方法でも、あなたはオブジェクトのトラッカーと背景(カメラ)のためのトラッカーのセットに行きつくことになります。
You must set up a complete set of constraints- position locks, orientation,
and distance (scale)-forboththe camera and object (a set for
each object, if there are several). Frequently, users ask why a second set of
constraints for the object is required, when it seems that the camera
(background) constraints should be enough.
あなたはカメラの完全なセット(もし複数あるなら、各々のためのセット)の制約位置ロック、方向と距離(スケール)のセットアップを完了していなければなりません。しばしば、ユーザーはオブジェクトのための制約の第2セットがなぜ必要かについて尋ねます。そのとき、カメラ(背景)制約で十分でなければならないようです。
However, recall a common film-making technique: shooting an actor, who is
close to the camera, in front of a set that is much further away. Presto, a
giant among mere mortals! Or, in reverse, a sequel featuring yet another group
of shrunken relatives, name the variety. The reason this works is that it is
impossible to visually tell the difference between a close-up small object
moving around slightly, and a larger object moving around dramatically, a
greater distance away. This is true for a person or a machine, or by any
mathematical means.
This applies independently to the background of a set, and to each object
moving around in the set. Each might be large and far, or close and small. Each
one requires its own distance constraint, one way or another.
The object's position and orientation constraints are necessary for a
different reason: they define the object's local coordinate system. When you
construct a mesh in your favorite animation package, you can move it around
with respect to a local center point, about which the model will rotate when
you later begin to animate it. In SynthEyes, the object's coordinate
constraints define this local coordinate system.
Despite the veracity of the above, there are ways that the relative
positioning of objects moving around in a scene can be discerned: shadows of an
object, improper actor sightlines, occasions where a moving object comes in
contact with the background set, or when the moving object temporarily stops.
These are assumptions that can be intellectually deduced by the audience,
though the images do not require it. Indeed, these assumptions are
systematically violated by savvy filmmakers for cinematic effect.
However, SynthEyes is neither smart/stupid enough to make assumptions, nor
to know when they have been violated. Consequently, it must be instructed how
to align and size the scenes in the most useful fashion.
The alignment of the camera and object coordinate systems can be
determined independently, using the usual kinds of setups for each.
The relative sizing for camera and object must be considered more
carefully when the two must interact, for example, to cast shadows from the
object onto a stationary object.
When both camera and object move and must be tracked, it is a good idea to
take on-set measurements between trackable points on the object and background.
These measurements can be used as distance constraints to obtain the correct
relative scaling.
If you do not have both scales, you will need to fix either the camera or
object scale, then systematically vary the other scale until the relationship
between the two looks correct.