I will be starting a contract with a film production company soon, and
will be writing code to process large images representing frames of
movies (unknown file format at this time). I'd also like to build the
interface in Java, and do the number crunching in Java too, if Java is
powerful enough to handle it.
I'd like to know what the best ways are to accomplish this. Should I
treat image data as arrays of bytes and use the old producer/consumer
model to construct images, or is it best to use BufferedImages and
Rasters? Should I use javax.ImageIO to load images, the Toolkit and
MediaTracker, or implement my own loader that takes advantage of NIO?
Top priority tasks would include loading images quickly, writing
images quickly, caching images and/or thumbnails for realtime
animation, and applying filters and tweaking individual pixles of
images. I'm very familiar with the Java2D API, but feel I don't know
much about the more data centered aspects of image manipulation.
Is Java up to this task? Would it be best to write the heavy duty
stuff in C++ and use the JNI to call it?
Thanks
Mark McKay
--
http://www.kitfox.com