| worldToNDC
|
M44f |
For images generated by 3D computer graphics rendering, a matrix
that transforms 3D points from the world to the Normalized Device
Coordinate (NDC) space of the renderer.
NDC is a 2D coordinate space that corresponds to the image plane,
with positive x and pointing to the right and y positive pointing
down. The coordinates (0, 0) and (1, 1) correspond to the upper
left and lower right corners of the OpenEXR display window.
To transform a 3D point in word space into a 2D point in NDC space,
multiply the 3D point by the worldToNDC matrix and discard the z
coordinate.
NDC space in OpenEXR is the same as in Pixar's Renderman.
|
| sensorCenterOffset
|
V2f |
Horizontal and vertical distances, in microns, of
the center of the light-sensitive area of the camera's sensor from
a point on that sensor where a sensor surface normal would intersect
the center of the lens mount.
When compared to an image captured with a perfectly centered sensor,
an image where both horizontal and vertical distances were positive
would contain more content holding what was at the right and what was
at the bottom of the scene being captured.
|
|
sensorAcquisitionRectangle |
Box2i |
The rectangular area of the sensor containing photosites
the contents of which are in one-to-one correspondence with
the captured sensels, for a monochrome sensor, or with the
reconstructed RGB pixels, for a sensor covered with color
filter array material in a Bayer or a similar pattern.
Because understanding the above formal definition is
critical for many applications, including camera solvers,
some short definitions:
-
a photosite is that optoelectronic component on the
sensor which, when light hits it, accumulates or
otherwise registers electric charge
-
a sensel is the read-out contents of a single
photosite
-
color filter array material is material deposited
on top of an array of photosites such that each
photosite is discretely covered with a material that
passes photons of certain wavelengths and that blocks
photons of other wavelengths
-
an RGB pixel contains red, green and blue components
indicating relative exposure values
-
RGB pixel reconstruction is the process of taking sensel data from a neighborhood
of a particular photosite, including that photosite itself, in a photosite covered by
either red, green or blue CFA material, and combining the read-out sensel data from a
particular photosite with that of surrounding photosites, said surrounding photosites
being covered by a variety of red, green or blue CFA materials, to produce an RGB pixel.
The Wikipedia article on demosaicing covers the basics
of these ideas quite well.
In the case of sensels read from a monochrome sensor, the idea of a
one-to-one relationship between sensels read from the photosite array
and pixels in the data window can be straightforward. Often there is a
conversion of the sensel data read from the photosites to a different
representation (e.g. integer to float) along with scaling of the
individual values.
A common spatial scaling is from a 2880 x 1620 acquisition format to a
1920 x 1080 HD format. In this case, a camera solver will want to know
the original set of photosites that contributed sensel values to the
downscaler: their number, and their position. Through a combination of
sensorAcquisitionRectangle, sensorPhotositePitch,
sensorOverallDimensions and sensorCenterOffset, the application can
know exactly the area on the sensor on which the light fell to create
the sensel values that produced a monochrome image.
RGB images are more complicated. RGB pixel reconstruction is a form of
filtering, and kernels are square, with relatively small span, e.g. 5x5
or 7x7. Edge handling for the kernel is important; the Wikipedia
article describing an image processing kernel covers it well.
Elements of the reconstruction kernel that are never at the center of
the kernel are not counted as part of the
sensorAcquisitionRectangle. Recalling the simple case above of a
non-spatially-scaled 2880 x 1620 monochrome image being in 1:1
correspondence with an array of photosites on the sensor, if we are
instead reading from a CFA-covered sensor to reconstruct a 2880 x 1620
RGB image, the actual array of all photosites whose sensel values were
fed into a 5x5 reconstruction kernel would not be 2880 x 1620, but 2884
x 1624. Nevertheless, the size of the sensorAcquisitionRectangle would
be 2880 x 1620.
Camera systems differ on how to handle the case where the position of
the RGB reconstruction kernel is such that one or more elements of the
kernel do not correspond to physical photosites; these are edge cases
in every sense of the phrase.
|
| longitude
|
float |
For images of real objects, the location where the image was recorded.
Longitude and latitude are in degrees east of Greenwich and north of the
equator. Altitude is in meters above sea level. For example, Kathmandu,
Nepal is at longitude 85.317, latitude 27.717, altitude 1305.
|
