#JPEG vs. GIF
14 messages in this thread
I just wanted to comment how much I like JPEG files, viewed with the JFIF as
opposed to GIF files viewed with the ZGIF datatype. I don't know if this is a
reflection of the superiority of JPEG as compared to GIF and a graphics format,
or if the JFIF datatype is superior to the ZGIF one. My main problem it that
because I am using an A2000, I am limited to 16 colors in hires-interlaced
mode. JFIF does an excellent job in remapping images to 16 colors. I recently
go a picture of Stonhenge of the USENET, and athough the colors of the sky and
clouds where off, the part of the image including the ground and Stonhenge
itself are truly remarkable. I suspect the if I used the Full size histogram
and Slow IDCT options of the JFIF datatype, I might have even gotten the clouds
right.
Of course in the image in question, the ground and the stones are are a shade
of brown, so the image was well suited for conversion to an small number of
colors, but I still am very impressed with JPEG/JFIF.
GIF images of anything with a large number of colors that are in any ditered,
on the other hand, always seem graining and the colors seem off. GIF seems ok
for simple graphics, like logos, but for complicated images GIF seems
inadequate.
Comments???
._ _ _ INTERNET: mshea@math.arizona.edu
| | |ichael _>hea COMPUSERVE: 76665,3360
JPEG is a more advanced graphics format than GIF. GIF was designed in the
days when 256 colors was all that "anybody needed".
JPEG is 24 bit, vs the 8 of GIF, and was designed by an industry
consortium to provide a tool for work.
GIF was designed by one outfit to provide different platforms the ability
to view the same graphic.
Different purposes, with different development dollars behind them, and at
different times in the evolution of the technology.
Once Ham was truely amazing. Now 24 bit is the standard.
So, in a very verbose way, "Yes JPEG is a superior format to GIF."
Yes, but I think Michael's issue was a little different, since he was
commenting on 16color renditions of images in both formats. That's a little
different than saying Yes, a 24bit image is better than an 8bit one, especially
if the 24bit file is smaller.
Mr. Zombie – now 76375,1776
True, a 16 color rendition is a far cry from either one, however the
technology in trhe JPEG format is superior. As far as the viewing of
either goes, that would be dependent on the software used.
The handling of the color mapping/dithering and the handling of the
resolution differences between the creating platform, and the viewing
platform must also be addressed in rating the various software.
I've seen some really bad GIF viewers over the years on different
platforms. By the time JPEG got going, there was a lot of previous work
to build upon, so it seems there was a head start on creating JPEG
viewers. However were a JPEG viewer to be of the same quality that the
early GIF viewers were, and the GIF viewer to be of the quality that the
current GIF viewers are, it would seem that the GIF was the better format.
Either way, if you have a 24 bit rendition of a scene, to start with, you
have a more accurate base to create whatever level you want to view at, be
it 16/32/64/256/4096 or more colors.
Just the same as 400 dpi scans look better at 182 dpi on a dot matrix,
than do 200 dpi scans. The 200 dpi still exceeds the capabilities of the
printer, but when you have more info to start with, you can put out a
better result in the end.
JPEG has quite a few nice things going for it and I use it all the time
with ImageFX. For one thing, JPEG is a 24-bit (16 million color)
compressed image format while GIF is only 8-bit (256 colors), so you have
a much larger palette to work with in the color map conversions. As far as
the JFIF datatype, it might do a much better job of dithering than ZGIF.
So as far as I'm concerned, it wouldn't bother me a bit if GIF dies
because of the Unisys/CompuServe royalty fiasco. JPEG might not be the
best if you need 24-bit true color accuracy because of it's degrading of
the image depending on compression, but for most purposes, it does a good
job.
// John (Warp 4040 & Emplant Quadra 840)
\X/ Mercury@ins.infonet.net
"JPEG might not be the best if you need 24-bit true color accuracy because of
it's degrading of the image depending on compression, but for most purposes, it
does a good job."
Actually, it does a good job for virtually all purposes, _including_ loss-less
compression. Lest we forget, JPEG has a loss-less compression mode that
behaves like a 24-bit GIF format (very good compression, but no image
compromise). Just set your comp. ratio to 100% when saving 24-bit images.
-Chris
How about this new-fangled fractal compression technique? I've read about this
in one of last year's Byte issues. Reportedly, this gives you superiour
compression rates with virtually no noticable degradation in quality as it is
observed in JPEG format pictures. This seems to be just _the_ thing for the
"Databahn" and yet I've only seen a few small(ish) ads for it since.
Does anyone have any experience with this?
Will it be possible to use such files on the Amiga (using Datatypes perhaps)?
Cheers,
// Christian "ELP" Busse
\X/ near Frankfurt, Germany
I read the article about Fractal Compression in Byte also Christian <g>.
It was an interesting article! I haven't seen too much on it since then. I
believe the article was written by two people who were part of a company
using the technology, and they haven't released much information. I would
imagine that they want to keep a close watch on it to keep it proprietary.
From what I understand, MicroSoft's "Encarta" CD uses Fractal compression
for the images, at least I read that somewhere. I do have some source code
of fractal compression routines that was posted in the Graphics Support
forum, but it only handles Greyscale images. One of these days I'll dig it
out and do some playing with it.
// John (Warp 4040 & Emplant Quadra 840)
\X/ Mercury@ins.infonet.net
The fractal compression schemes were at least in part developed by Michael
Barnsley of Georgia Tech. He wrote an introductory fractal geometry textbook
called "Fractals Everywhere" that is very good. He has since left the
University and formed a company to develope fractal compression commercially.
His company was envolved in the MS Encarta CD developement. I think we will
here a lot about this in the future.
He has also published a technical book on fractal compression theory. If you
have a good tech book store or library close, you might be able to find it.
._ _ _ INTERNET: mshea@math.arizona.edu
| | |ichael _>hea COMPUSERVE: 76665,3360
Thanks for the detailed information about Fractal compression Michael. It
seems though as long as it's been developed we would see more use, but
that doesn't seem to be the case. Either Michael Barnsley's company is
keeping it too proprietary so there is no advancement of the technology,
or it's not as good as it was cracked up to be.
// John (Warp 4040 & Emplant Quadra 840)
\X/ Mercury@ins.infonet.net
I think the major stumbling block to Fractal Image Compression is the time
it takes to _encode_ an image. While the results are truly stunning, I
seem to remember it taking hours to compress a single image.
I've seen ads for the Fractal compression package from Iterated Systems in
magazines like NewMedia – it seems they're targeting the developer market.
(Where they can make more money on fewer units sold.)
P.S. Turn your 4000 on its side, and you won't need that second fan
inside…….
Erik Flom
>I've seen ads for the Fractal compression package from Iterated Systems…
Iterated Systems is Michael Barnsley's company as I recall.
._ _ _ INTERNET: mshea@math.arizona.edu
| | |ichael _>hea COMPUSERVE: 76665,3360