#IFF Chunky Pixel
7 messages in this thread
I recently registered as a developer with CATS. I use the Amiga to develop
image processing and matching software, mainly for automatic fingerprint
recognition. At the moment I have a very ad hoc collection of program
harnesses used for developing compression algorithms. I want to
rationalise these, at the moment I use my own chunky pixel (1 byte/pixel,
255 level) image format. Are there any IFF standards for chunky pixel as
opposed to bitplane images in the works?
Another question, is anyone working on a JPEG compression algorithm
implementation for the Amiga?
Any and all comments welcomed.
Kevin
Kevin,
I'm interested in compressions schemes also. Have you been in touch
with Michael Barnsley of Iterated Systems? Unfortunately they want several
thousands of dollars upfront. Also unfortunately, they don't seem
interested in having their stuff ported to the Amiga, but they _do_ get
impressive compression (fractal compression)..claiming 100-1 and such.
As for the chunky pixel, I haven't heard of any such proposal for IFF.
Of course, you must keep images (which are to be displayed) in bitplane
format in the Amiga…but those you're working on certainly don't have to
be. BTW, how does one go about getting a copy of the JPEG compression
algorithm?
Victor A. Wagner, Jr.
I've read the general publications on Barnsleys methods. I'm not
impressed. The encoding computation costs are horrendous (and fingerprint
archives are huge, small systems are >300,000 images). Second he seems to
require human intervention to encode. I can't accept that.
As for JPEG the most recent suggestion I saw (in the ACM Comms multimedia
article) was to contact ANSI, 1430 Broadway, NY 10018 and ask for the draft
standard. I'll try it next week and let you know.
Kevin
Kevin,
The latest I have from Barnsley indicated completely 'automatic'
compression.
I agree that compression seems to be slow (they have a hardware assist
board for the PC), but his stuff is clearly, IMHO, for publishers (where
the data will be read a lot and written seldom).
Victor A. Wagner, Jr.
Kevin:
We're looking at the JPEG algorithm for something we're doing with image
archival on the Amiga. Unfortunately, at the data rates we are running it
would be impossible to compress in software. Luckily for us there is a
company called C-Cubed that is developing a JPEG chipset for the broadcast
industry. They claim that their chips can compress in milliseconds what
would take a 25mhz 68030 over an hour (!). C-Cubed intends to sample right
about now.
Ariel
Ari,
Please keep us informed. Has anyone else been contacted by those guys
in Atlanta (name USED to be "Iterated Systems") who do fractal
compression/decompression? I personally think their price is too
high…unless, e.g. C= picked up the license for ALL Amigas. The thought
of having reasonable de-compress times (a second) and 100-1 image
compression has a certain appeal.
Victor A. Wagner, Jr.
I know of the chipset. For my application I can do compression much faster
than JPEG with probably comparable image quality at the same compression.
My client wants software, the incremental hardware cost hurts his marketing
(he says). But I'm very specialised. The new NeXT machine graphics board
includes the C-Cube chip, but at 3,995 for the board !! Kevin