public final class MDLCameraProjection
extends java.lang.Object
[@summary] MDLCamera models a physically plausible camera.
Values are represented as float in MDLCamera as it offers sufficient precision, and because calculations will be identical on any processor architecture. This consistency is a requirement of the model.
MDLCamera provides the a model of the parameters governing the physical process of transforming a scene into an image.
This process is modeled as a series of steps, each governed by the physical properties of real world cameras.
1. The position and orientation of the camera [@property] nearVisibilityDistance The minimum distance from the camera of a visible object [@property] farVisibilityDistance The maximum distance from the camera of a visible object
Those objects existing between nearVisibilityDistance and farVisibilityDistance are considered to be visible. All lights are considered, irrespective of nearVisibilityDistance and farVisibilityDistance, to compute scene luminance.
[@property] worldToMetersConversionScale
Some calculations, such as the calculation of stereo view matrices, require calculations to occur in world space. Significant quantities measured in mm therefore use this conversion scale to perform the calculation. The default value is 1.0.
3. Scene Luminance through the lens
[@property] barrelDistorion [@property] fisheyeDistortion
Illuminated objects result in scene luminance, which passes through the lens. All lenses impose some amount of radial distortion which can be computed from focal length. However, some lenses introduce error, and radial distortion can be used as an aesthetic control as well. Therefore radial distortion is provided as a property. If r is the radial distance of a pixel from the center of the projection, then radial distortion is computed as
r' = r * (1 + barrelDistorion * r^2 + fisheyeDistortion * r^4)
radialDistortion sufficiently describes the distortion characteristic of most lenses. In order to simulate certain other lenses, such as those found in security cameras, fisheye lenses, plastic toy lenses, sport cameras, or some VR headsets, radialDistortion2 is introduced.
The default for the radial distortion parameters is zero, resulting in a rectilinear projection.
[@property] opticalVignetting
Optical vignetting occurs to some degree in all lenses. It results from light at the edge of an image being blocked as it travels past the lens hood and the internal lens apertures. It is more prevalent with wide apertures. A value of zero indicates no optical vignetting is occuring, and a value of one indicates that vignetting affects all locations in the image according to radial distance. Optical vignetting also occurs in head mounted displays, and the value here can be used as an intended amount of vignetting to apply to an image.
[@property] chromaticAberration
Chromatic aberration occurs to some degree in all lenses. It results from a lens bringing different wavelengths of light to focus at different places on the image plane. A value of zero indicates no chromatic aberration is occurring, and one indicates maximum. Chromatic aberration affects all locations in the image according to radial distance. Chromatic aberration also occurs in head mounted displays, and the value here can be used as an intended amount of chromatic aberration to apply to an image.
4. Geometry of the lens
This is a thin lens model.
[@property] focalLength
The default focal length is 50mm, corresponding to a field of view of 54 degrees, and vertical sensor aperture of 24mm. Changing focalLength will update the field of view property.
[@property] focusDistance
The distance, in meters, at which the lens is focused. The default is 2.5m.
[@property] fieldOfView
The field of view is calculated from the focal length and sensor aperture. Changing the field of view will update the focalLength property with respect to the sensor aperture. The default is 54 degrees, corresponding to a focal length of 50mm, and a vertical sensor aperture of 24mm.
[@property] fStop
The f-stop is the ratio of the lens' focal length to the diameter of the entrance pupil. The default is 5.6. It controls the amount of light that reaches the sensor, as well as the size of out of focus parts of the image. The diameter of the entrance pupil, is therefore obtained by dividing the fStop by the focalLength.
5. Lens exit aperture
[@property] apertureBladeCount
The shape of out of focus highlights in a scene is commonly known as "bokeh". The aesthetic quality of a lens' bokeh is one of the characteristics that drives the choice of a lens for a particular scene. To a large degree, the appearance of bokeh is governed by the shape of the lens aperture. Typical lens apertures are controlled by a series of overlapping blades that can be irised open and closed. A lens with a five blade aperture will yield a five sided bokeh. The default is zero, which is to be interpreted as a perfectly round aperture.
Note that the effect of a filter on the front of the lens can be modeled equivalently at the exit aperture. The MIOCamera does not explicitly provide specification of such effects, but a simulation could incorporate them at this stage.
[@property] maximumCircleOfConfusion
Although the size of an out of focus bokeh highlight can be computed from other camera properties, it is often necessary to limit the size of the circle of confusion for aesthetic reasons. The circle of confusion is specified in mm, and the default is 0.05mm. The units are mm on the sensor plane.
[@property] shutterOpenInterval
The length of time in seconds the shutter is open, impacting the amount of light that reaches the sensor and also the length of motion blur trails. The shutter time is not the same thing as scene frame rate. The rule of thumb for movies is that the shutter time should be half the frame rate, so to achieve a "filmic" look, the shutter time choice might be 1/48 of a second, since films are usually projected at 24 frames per second. Shutter time is independent of simulation frame rate because motion blur trails and exposure times should be held constant in order to avoid flicker artifacts.
6. Sensor illuminance
[@property] sensorVerticalAperture
The default aperture is 24mm, corresponding to a 35mm stills camera. _____________________ [][][][][][][][][][][ ^ \ | | \ ^ | \ | | \ 24mm 35mm \ | | \ v | ][][][][][][][][][][] v --------------------- <- 36mm -->
Changing the aperture will update the field of view property with respect to the focal length. The horizontal aperture can be calculated from the aspect property.
[@property] sensorAspect aspect ratio of width versus height of the sensor aperture. The default is 36mm/24mm, ie: 1.5.
[@property] sensorEnlargement
The sensor enlargment property scales the active region of the sensor. A zoom factor of 2 will result in the central portion of the image being cropped and expanded to fill the image plane. The default is 1.
[@property] sensorShift
The sensor shift property moves the center of the sensor relative to the lens. This is useful in certain calculations, such as shadow or reflection projections, as well as to provide vergence in a parallel stereoscopic camera.
7. Exposure
Finally, the sensor values need to be converted for display.
[@property] flash
Flashing is applied to a film stock some time before processing. It is a small and even level of exposure added to the entire image intended shift the brightness and color of darker areas of the image. Since exposure is logarithmic, flash does not affect midtones or highlights.
The equivalent to flash in a color grading system is known as lift. Typically red, green, and blue can be controlled independently.
A negative value for flash can be used to subtract color.
[@property] exposure
Finally, exposure should be applied to the compressed value. Red, green, and blue exposure can be specified separately. The default is 1.0.
[@property] exposureCompression
Gamma curve compression where values below the x value are to be passed through, about the y value, values are to be clamped at maximum display brightness, and a function such as a logarithmic ramp is to be applied in between. The default is (1,0, 1.0).
A displayable value is therefore obtained via pow(exposureCompression((sensor value + flash) * exposure), displayGamma)
transform property
The MDLTransformComponent on an MDLCamera is an MDLTransform.
2. Visible Objects and Lights
| Modifier and Type | Field and Description |
|---|---|
static long |
Orthographic |
static long |
Perspective |
public static final long Perspective
public static final long Orthographic