Re: java 2D adv imaging api

• To: jim schmidt <j-m.schmidt@xxxxxxxxx>
• Subject: Re: java 2D adv imaging api
• From: Bill Hibbard <hibbard@xxxxxxxxxxxxxxxxx>
• Date: Tue, 06 Feb 2001 12:38:24 -0600

Hi Jim,

> ref:  http://java.sun.com/products/java-media/jai/collateral/datasheet.html
> 
> has this api been evaluated with respect to use within VisAD?

Curtis recently added visad.data.jai.JAIForm that uses JAI
(Java Advanced Imaging) to import BMP, GIF, FlashPix, JPEG,
PNG, PNM, and TIFF files into VisAD data objects.

Other than support for these image file formats, JAI is
designed to provide efficient manipulation (including
viewing) of large imaging via tiling (dividing images into
rectangular tiles) and deferred execution (only applying
operations to those pixels that actually affect the user's
display). These would be useful functions for VisAD with
large images.

It would be possible to implement a sub class of FlatField
(or a whole new class implementing Field if you wanted to
work hard) whose constructor constrained it to only generate
Fields with MathTypes in the form ((x, y) -> z) and possibly
with constraints on CoordinateSystems abd Units (again
depending on how hard you want to work), and that used JAI
for its implementation of the Field methods. To really get
the benefit, you'd probably implement a sub class of
DataRenderer to use JAI's efficiencies for display operations.

Some of the benefits of JAI exist or are in the works in
VisAD. VisAD's event logic with CellImpls etc provide ways
to build deferred execution logic in applications. And I am
currently working on a pretty general tiling capability in
the visad.cluster package (the idea is that tiles are
distributed across the nodes of a cluster, and may be 2-D
image tiles or 3-D grid tiles).

Cheers,
Bill
----------------------------------------------------------
Bill Hibbard, SSEC, 1225 W. Dayton St., Madison, WI  53706
hibbard@xxxxxxxxxxxxxxxxx  608-263-4427  fax: 608-263-6738
http://www.ssec.wisc.edu/~billh/vis.html

• References:
  • java 2D adv imaging api
    • From: jim schmidt