readme-jai.txt

jai-1_1_2.rar

     Optimized Rect and Random iterators exist as well as a non-optimized
     version of the Rook iterator.

 12. Snapshotting of External Sources

     SnapshotImage provides an arbitrary number of synchronous views of a
     possibly changing WritableRenderedImage.

 13. Meta-data Handling

     Meta-data handling is provided via a name-value database of properties
     associated with each JAI image. Mechanisms are provided by which such
     properties may be generated and processed in the course of image
     manipulation. The ability to defer computation of such data is also
     provided in a manner conceptually equivalent to that available for
     image data. Please refer to the DeferredData and DeferredProperty
     classes for more information.

 14. Serialization Support

     SerializerFactory provides a framework is provided to assist in
     serializing instances of classes which do not implement
     java.io.Serializable. Such objects must be serialized by extracting a
     serializable version of their state from which the original object may
     be extracted after deserialization.

Operators

Java Advanced Imaging extends the imaging functionality provided in the Java
2D API by providing a more flexible and scalable architecture targeted for
complex, high performance imaging requirements. In this context a large
number of imaging operators are provided.

   * Native Acceleration

     Pure Java implementations are provided for all image operators and
     imaging performance is addressed for some of these by providing C-based
     native code. Native C-code based acceleration for a certain subset of
     operators under specific conditions (listed in the table below) is
     available for the Sun/Solaris, Win32 and Linux (x86 only) platforms. On
     Sun UltraSPARC-based platforms, additional performance is gained with
     hardware acceleration via the VIS instructions for most natively
     supported operators. On Win32 platforms which support MMX instructions,
     hardware acceleration is gained for a subset of the natively supported
     operators.

     If a native implementation is present it is, by default, the preferred
     implementation. But if the nature of the sources and parameters of the
     operation are incompatible with the native operation then processing
     will revert to Java code. In general the following minimum requirements
     must be adhered to for the mediaLib native implementation of an
     operation to be executed:

        o All sources must be RenderedImages.
        o All sources and destination must have
             + a SampleModel which is a ComponentSampleModel and
             + a ColorModel which is a ComponentColorModel or no ColorModel
               (i.e., it is null).
        o All sources and the destination must have at most 4 bands of pixel
          data.
        o If Interpolation type is one of the arguments to the operator,
          then native acceleration is available only for Nearest, Bilinear,
          Bicubic and Bicubic2 cases. Additionally for byte images
          InterpolationTable is also supported for native acceleration.

     Further restrictions may be imposed by individual operations but the
     above are the most common requirements.

   * Imaging Operators

     The following image operators are implemented in this release. Only a
     brief description of each operator is provided here. For detailed
     information on these operators, refer to the package
     javax.media.jai.operator in the full documentation available at
     http://java.sun.com/products/java-media/jai/docs/index.html.

     All operations are performed on a per-band basis, unless specified
     otherwise. C acceleration applies to all platforms whereas VIS is
     specific to Sun UltraSPARC and MMX to Windows.

       1. Point and Arithmetic Operators

                                                     Native Acceleration
              Operator Name       Description
                                                 C VIS MMX      Notes
                                Computes the
           absolute             absolute value   X  X   X
                                of the pixels of
                                an image.
                                Adds the pixel
           add                  values of two    X  X   X
                                source images.
                                Adds a
           addcollection        collection of
                                images to one
                                another.
                                Adds a set of
                                constant values
           addconst             to the pixel     X  X   X
                                values of a
                                source image.
                                Adds a set of
                                constant values
           addconsttocollection to the pixel     X  X   X
                                values of a
                                Collection of
                                source images.
                                And's the pixel
           and                  values of two    X  X   X
                                source images.
                                And's the pixel
                                values of a
           andconst             source image     X  X   X
                                with a set of
                                constants.
                                Computes a
                                linear
           bandcombine          combination of   X  X   X  3x4 matrix only.
                                the bands of an
                                image.
                                Creates an image
                                consisting of
                                all bands of all
           bandmerge            sources
                                concatenated in
                                the order
                                encountered.
                                Selects a subset
                                of the bands of            Only if the band
           bandselect           an image,        X  X   X  selection is
                                possibly                   monotonically
                                reordering them.           increasing.
                                Thresholds a
                                single-band
           binarize             image to two     X  X
                                levels to
                                generate a
                                bilevel output.
                                Set all pixel
                                values below the
                                low value to
                                that low value,
           clamp                set all the      X  X   X
                                pixel values
                                above the high
                                value to that
                                high value.
                                Converts an
           colorconvert         image to a given
                                ColorSpace.
                                Generates an
                                optimal LUT by
           colorquantizer       executing a
                                color
                                quantization
                                algorithm
                                Combines two
                                images based on
           composite            their alpha      X  X   X
                                values at each
                                pixel.
                                Creates an image
           constant             with constant
                                pixel values.
                                Divides the
                                pixel values of
                                the first source
           divide               image by the     X
                                pixel values of
                                the second
                                source image.
                                Divides the
                                pixel values of
           dividebyconst        a source image   X
                                by a set of
                                constants.
                                Divides a set of
           divideintoconst      constants by the X
                                pixel values of
                                a source image.
                                Computes the
           exp                  exponential of   X
                                the pixel values
                                of an image.
                                Performs
                                reformatting on
                                an image,
                                including data
           format               type casting,
                                replacing the
                                SampleModel and
                                ColorModel, and
                                restructuring
                                the tile grid.
                                Inverts the
           invert               pixel values of  X  X   X
                                an image.
                                Computes the
                                natural
           log                  logarithm of the X
                                pixel values of
                                an image.

                                Performs general           Only if table
           lookup               table lookup on  X  X   X  has less than or
                                an image.                  equal to 4
                                                           bands.
                                Performs a
                                piecewise linear
                                remapping of
           matchcdf             pixel values to
                                match a given
                                cumulative
                                distribution
                                function.
                                Chooses the
           max                  maximum pixel    X  X   X
                                values between
                                two images.
                                Chooses the
           min                  minimum pixel    X  X   X
                                values between
                                two images.
                                Multiplies the
           multiply             pixel values of  X  X
                                two source
                                images.
                                Multiplies the
                                pixel values of
           multiplyconst        a source image   X  X
                                by a set of
                                constants.
                                Inverts the
           not                  pixel values of  X  X   X
                                a source image.
                                Or's the pixel
           or                   values of two    X  X   X
                                source images.
                                Or's the pixel
                                values of a
           orconst              source image     X  X   X
                                with a set of
                                constants.
                                Overlays one
           overlay              image on top of
                                another image.
                                Performs
                                piecewise linear
           piecewise            remapping of the
                                pixel values of
                                an image.
                                Performs a
                                linear remapping
           rescale              of the pixel     X  X
                                values of an
                                image.
                                Subtracts the
                                pixel values of
           subtract             one image from   X  X   X
                                those of
                                another.
                                Subtracts a set
                                of constant
           subtractconst        values from the  X  X   X
                                pixel values of
                                an image.
                                Subtracts a set
                                of constant
           subtractfromconst    values from the  X  X   X
                                pixel values of
                                an image.
                                Maps the pixel
                                values that fall
           threshold            between a low    X  X   X
                                and high value
                                to a set of
                                constants.
                                Xor's the pixel
           xor                  values of two    X  X   X
                                source images.
                                Xor's a source
           xorconst             image with a set X  X   X
                                of constants.

       2. Area and Geometric Operators

                                                      Native Acceleration
               Operator Name        Description
                                                  C VIS MMX       Notes

                                 Performs first             InterpolationTable
           affine                order geometric  X  X   X  is not MMX
                                 image warping.             accelerated for
                                                            even byte images.

           border                Adds a border