Perspective Window
The perspective window allows you to go into the scene to view it from any
direction. Or, you can lock the perspective view to the tracked camera view.
You can build a collection of test or stand-in objects to evaluate the
tracking. Later, we’ll see that it enables you to assemble tracker locations
into object models as well.
The perspective window is controlled by a right-click
menu, where different mouse modes can be selected. The middle mouse
button can always be used for general navigation using the control and ALT
variations. The left mouse button may be used instead, with the same
variations, when the Navigation mode is selected.
Image Overlay
The perspective window can be used to overlay inserted objects over the
live imagery, much like the camera view. Select Lock to Current
Camera to lock or release, or use the ‘L’ key. Note that when the view
is locked to the camera, you can not move or rotate the camera, or adjust the
field of view.
The perspective window is designed to work only with
undistorted imagery: the perspective view always shows the
view with an ideal, undistorted camera, even if SynthEyes has calculated a lens
distortion value.
This reflects the essential difference between the camera and perspective
views: the camera view shows the source footage as is, and distorts all
3-D geometry to match. The perspective view shows the 3-D world as is, and
would have to dynamically un-distort the footage in order to make everything
line up. Rather than trying to do that while you use the perspective window, it
requires that you un-distort the footage previously, if necessary.
Navigation
In navigation mode, with the left mouse button, or any time using the
middle mouse button, dragging will pan the display. Control-dragging will cause
the camera to look around in different directions, without translating.
Control-ALT-dragging will truck the camera forwards or backwards.
ALT-dragging will cause the camera to orbit. The center of the orbit will
be the center of any selected vertices in the current edit mesh (more on that
later), around a selected object, or around a point in space directly ahead of
the camera.
The mouse's scroll wheel will change the view's field of view if it is not
locked to the camera, or if it is, it will change the current time. If locked,
shift-scrolling will zoom the time bar.
If you hold down the Z or ‘/” (apostrophe/double-quote) key when
left-clicking, the mouse mode will temporarily change to be navigation mode;
the mode will switch back when the mouse button is released. You can also
switch to navigate mode using the ‘N’ key. So it is always pretty easy to
navigate.
Creating Objects
Create objects on the perspective window grid with the Create mesh
object mode. Use the 3-D panel or right-click menu to control what
kind of object is created. Selecting an object type from the right-click menu
launches the creation mode immediately. If the SynthEyes user interface is set
so that a moving object is active on the Shot menu, the created object will be
attached to that object.
The Duplicate Mesh script on the Script menu can clone and offset a mesh,
so you can create a row of fence posts quickly, for example.
Moving and Rotating Objects
When an object is selected, handles appear. You can either drag the handle
to translate the object along the corresponding axis, or control-drag to rotate
around that axis.
The handles appear along the main coordinate system axes by default, so
for example, you can always drag an object vertically no matter what its
orientation.
However, if you select Local-coordinate handles on the
right-click menu, the handles will align with the object's coordinate system,
so that you can translate along a cylinder's axis, despite its
orientation.
Additionally, for cameras or moving objects, you can select Path-relative
handles, so you can adjust along or perpendicular to the path.
Placing Seed Points and
Objects
In the Place mode, you can slide the selected object around on the surface
of any existing mesh objects. For example, place a pyramid onto the top of a
cube to build a small house.
You can also use the place mode to put a tracker's seed/lock point onto
the surface of an imported reference head model, for example, to help set up
tracking for marginal shots.
For this latter workflow, set up trackers on the image, import the
reference model. Go to the Camera and Perspective viewport configuration. Set
the perspective view to Place mode. Select each tracker in the camera view,
then place its seed point on the reference mesh in the perspective view. You
can reposition the reference mesh however you like in the perspective view to
make this easy—it does not have to be locked to the source imagery to do this.
This work should go quite quickly.
If you need to place trackers (or meshes) at the vertices
of the mesh, not on the surface, hold the control key down as you use the place
mode, and the position will snap onto the vertices.
Grid Operations
The perspective window's grid is used for object creation and mesh
editing. It can be aligned with any of the walls of the set: floor, back,
ceiling, etc. A move-grid mode translates the grid, while maintaining the same
orientation, to give you a grid 1 meter above the floor, say.
A shared custom grid position can be matched to the location of several
vertices or trackers using the right-click|Grid|To Facets/Verts/Trackers menu
item. If 3 trackers (or vertices) are selected, the grid is moved into the
plane defined by the three. If two are selected, the grid is rotated to align
the side-to-side axis along the two. If one is selected, the grid slides to put
that tracker at the origin. So by repeatedly selecting some trackers(vertices)
and using this menu command, the grid can be aligned as desired.
You can easily create an object on the plane defined by
any 3 trackers by selecting them, aligning the grid to the
trackers, then creating the object, which will be on the grid.
You can toggle the display of the grid using the Grid/Show Grid menu item,
or the ‘G’ key.
Shadows
The perspective window generates shadows to help show tracking quality and
preview how rendered shots will ultimately appear.
The 3-D panel includes control boxes for Cast Shadows and Catch Shadows.
Most objects (except for the Plane) will cast shadows by default when they are
created.
If there are no shadow-catching objects, shadows will be cast onto the
ground plane. This may be more or less useful, depending on your ground plane;
if the ground is very irregular or non-existent, this will be confusing.
If there are shadow-catching objects defined, shadows will be cast from
shadow-casting objects onto the shadow-catching objects. This can preview
complex effects such as a shadow cast onto a rough terrain.
Shadows may be disabled from the main View menu, and the shadow black
level may be set from the Preferences color settings. The shadow enable status
is “sticky” from one run to the next, so that if you do not usually use it, you
will not have to turn it off each time you start SynthEyes.
Note that as with most OpenGL fast-shadow algorithms, there can be shadow
artifacts in some cases. Final shadowing should be generated in your 3-D
rendering application.
Note that the camera viewport does not display shadows by design.
Edit Mesh
The perspective window allows meshes to be constructed and edited, which
is discussed in
Building Meshes from Tracker Positions. One mesh can be selected as an edit
mesh at any time―select a mesh, then right-click Set Edit Mesh or hit the ‘M’
key.
Preview Movie
After you solve and add a few test objects, you can render a test
Quicktime movie (except on Win64) or a BMP, Cineon, DPX, JPEG, OpenEXR, PNG,
SGI, or Targa sequence (also TIFF on Mac). While the RAM-based playback is
limited by the amount of RAM, and has a simplified drawing scheme to save time,
the preview movie supports anti-aliasing. The movie playback can later run at
the full rate regardless of length.
Right-click in the perspective window to bring up the menu and select the
Preview Movie item to bring up a dialog allowing the output file name,
compression settings, and various display control settings to be set. Usually
you will want to select square pixel output for playback on computer monitors
in Quicktime; it will convert 720x480 source to 640x480, for example, so that
the preview will not be stretched horizontally.
If you are making a Quicktime movie, be sure to bring up the compression
settings and select something, Quicktime has no default and may crash if you do
not select something.
Also, different codecs will have their own parameters and
requirements.
Important Tip: the H.264 codec requires that the Key every N frames
checkbox be off, and the limit data-rate to 90 kb/sec checkbox be off:
otherwise there will be only one frame.
Similarly, image files used in a sequence may have their own settings
dialog. Note that image sequences written from the Preview Movie are always 8
bit/channel with no alpha. You can re-write image sequences at 16 bit and
including an alpha channel using the
Image Preprocessor (again depending on details of the source and output
file format).
Technical Controls
The
Scene Settings dialog contains many numeric settings for the perspective
view, such as near and far camera planes, tracker and camera icon sizes,
etc. You can access the dialog either from the main Edit menu, or from the
perspective window's right-click menu.
By default, these items are sized proportionate to the current “world
size” on the
solver control panel. Before you go nuts changing the perspective window
settings, consider whether it really means that you need to adjust your world
size instead!