What is JPEG?
 Why use JPEG?
 When should I use JPEG, and when should I stick with GIF?
 How well does JPEG compress images?
 What are good "quality" settings for JPEG?
 Where can I get JPEG software?
 How do I view JPEG images posted on Usenet?
More advanced questions:
 What is color quantization?
 What are some rules of thumb for converting GIF images to JPEG?
 Does loss accumulate with repeated compression/decompression?
 What is progressive JPEG?
 Isn't there a lossless JPEG?
 Why all the argument about file formats?
 How do I recognize which file format I have, and what do I do about it?
 How does JPEG work?
 What about arithmetic coding?
 Could an FPU speed up JPEG? How about a DSP chip?
 Isn't there an M-JPEG standard for motion pictures?
 Where are FAQ lists archived?
Subject:  What is JPEG?
JPEG (pronounced "jay-peg") is a standardized image compression mechanism.
JPEG stands for Joint Photographic Experts Group, the original name of the
committee that wrote the standard.
JPEG is designed for compressing either full-color or gray-scale images
of natural, real-world scenes. It works well on photographs, naturalistic
artwork, and similar material; not so well on lettering, simple cartoons,
or line drawings. JPEG handles only still images, but there is a related
standard called MPEG for motion pictures.
JPEG is "lossy," meaning that the decompressed image isn't quite the same as
the one you started with. (There are lossless image compression algorithms,
but JPEG achieves much greater compression than is possible with lossless
methods.) JPEG is designed to exploit known limitations of the human eye,
notably the fact that small color changes are perceived less accurately than
small changes in brightness. Thus, JPEG is intended for compressing images
that will be looked at by humans. If you plan to machine-analyze your
images, the small errors introduced by JPEG may be a problem for you, even
if they are invisible to the eye.
A useful property of JPEG is that the degree of lossiness can be varied by
adjusting compression parameters. This means that the image maker can trade
off file size against output image quality. You can make *extremely* small
files if you don't mind poor quality; this is useful for applications such
as indexing image archives. Conversely, if you aren't happy with the output
quality at the default compression setting, you can jack up the quality
until you are satisfied, and accept lesser compression.
Another important aspect of JPEG is that decoders can trade off decoding
speed against image quality, by using fast but inaccurate approximations to
the required calculations. Some viewers obtain remarkable speedups in this
Subject:  Why use JPEG?
There are two good reasons: to make your image files smaller, and to store
24-bit-per-pixel color data instead of 8-bit-per-pixel data.
Making image files smaller is a win for transmitting files across networks
and for archiving libraries of images. Being able to compress a 2 Mbyte
full-color file down to, say, 100 Kbytes makes a big difference in disk
space and transmission time! And JPEG can easily provide 20:1 compression
of full-color data. If you are comparing GIF and JPEG, the size ratio is
usually more like 4:1 (see " How well does JPEG compress images?").
If your viewing software doesn't support JPEG directly, you'll have to
convert JPEG to some other format to view the image. Even with a
JPEG-capable viewer, it takes longer to decode and view a JPEG image than
to view an image of a simpler format such as GIF. Thus, using JPEG is
essentially a time/space tradeoff: you give up some time in order to store
or transmit an image more cheaply. But it's worth noting that when network
or telephone transmission is involved, the time savings from transferring a
shorter file can be greater than the time needed to decompress the file.
The second fundamental advantage of JPEG is that it stores full color
information: 24 bits/pixel (16 million colors). GIF, the other image format
widely used on Usenet, can only store 8 bits/pixel (256 or fewer colors).
GIF is reasonably well matched to inexpensive computer displays --- most
run-of-the-mill PCs can't display more than 256 distinct colors at once.
But full-color hardware is getting cheaper all the time, and JPEG images
look *much* better than GIFs on such hardware. Within a couple of years,
GIF will probably seem as obsolete as black-and-white MacPaint format does
today. Furthermore, JPEG is far more useful than GIF for exchanging images
among people with widely varying display hardware, because it avoids
prejudging how many colors to use (see " What is color quantization?").
Hence JPEG is considerably more appropriate than GIF for use as a Usenet
A lot of people are scared off by the term "lossy compression". But when
it comes to representing real-world scenes, *no* digital image format can
retain all the information that impinges on your eyeball. By comparison
with the real-world scene, JPEG loses far less information than GIF.
The real disadvantage of lossy compression is that if you repeatedly
compress and decompress an image, you lose a little quality each time
(see " Does loss accumulate with repeated compression/decompression?").
This is a serious objection for some applications but matters not at all
for many others.
Subject:  When should I use JPEG, and when should I stick with GIF?
JPEG is *not* going to displace GIF entirely; for some types of images,
GIF is superior in image quality, file size, or both. One of the first
things to learn about JPEG is which kinds of images to apply it to.
Generally speaking, JPEG is superior to GIF for storing full-color or
gray-scale images of "realistic" scenes; that means scanned photographs and
similar material. Any continuous variation in color, such as occurs in
highlighted or shaded areas, will be represented more faithfully and in less
space by JPEG than by GIF.
GIF does significantly better on images with only a few distinct colors,
such as line drawings and simple cartoons. Not only is GIF lossless for
such images, but it often compresses them more than JPEG can. For example,
large areas of pixels that are all *exactly* the same color are compressed
very efficiently indeed by GIF. JPEG can't squeeze such data as much as GIF
does without introducing visible defects. (One implication of this is that
large single-color borders are quite cheap in GIF files, while they are best
avoided in JPEG files.)
Computer-drawn images, such as ray-traced scenes, usually fall between
photographs and cartoons in terms of complexity. The more complex and
subtly rendered the image, the more likely that JPEG will do well on it.
The same goes for semi-realistic artwork (fantasy drawings and such).
But icons that use only a few colors are handled better by GIF.
JPEG has a hard time with very sharp edges: a row of pure-black pixels
adjacent to a row of pure-white pixels, for example. Sharp edges tend to
come out blurred unless you use a very high quality setting. Edges this
sharp are rare in scanned photographs, but are fairly common in GIF files:
consider borders, overlaid text, etc. The blurriness is particularly
objectionable with text that's only a few pixels high. If you have a GIF
with a lot of small-size overlaid text, don't JPEG it. (If you want to
attach descriptive text to a JPEG image, put it in as a comment rather than
trying to overlay it on the image. Most recent JPEG software can deal with
textual comments in a JPEG file, although older viewers may just ignore the
Plain black-and-white (two level) images should never be converted to JPEG;
they violate all of the conditions given above. You need at least about
16 gray levels before JPEG is useful for gray-scale images. It should also
be noted that GIF is lossless for gray-scale images of up to 256 levels,
while JPEG is not.
If you have a large library of GIF images, you may want to save space by
converting the GIFs to JPEG. This is trickier than it may seem --- even
when the GIFs contain photographic images, they are actually very poor
source material for JPEG, because the images have been color-reduced.
Non-photographic images should generally be left in GIF form. Good-quality
photographic GIFs can often be converted with no visible quality loss, but
only if you know what you are doing and you take the time to work on each
image individually. Otherwise you're likely to lose a lot of image quality
or waste a lot of disk space ... quite possibly both. Read sections 8 and 9
if you want to convert GIFs to JPEG.
Subject:  How well does JPEG compress images?
Very well indeed, when working with its intended type of image (photographs
and suchlike). For full-color images, the uncompressed data is normally 24
bits/pixel. The best known lossless compression methods can compress such
data about 2:1 on average. JPEG can typically achieve 10:1 to 20:1
compression without visible loss, bringing the effective storage requirement
down to 1 to 2 bits/pixel. 30:1 to 50:1 compression is possible with small
to moderate defects, while for very-low-quality purposes such as previews or
archive indexes, 100:1 compression is quite feasible. An image compressed
100:1 with JPEG takes up the same space as a full-color one-tenth-scale
thumbnail image, yet it retains much more detail than such a thumbnail.
For comparison, a GIF version of the same image would start out by
sacrificing most of the color information to reduce the image to 256 colors
(8 bits/pixel). This provides 3:1 compression. GIF has additional "LZW"
compression built in, but LZW doesn't work very well on typical photographic
data; at most you may get 5:1 compression overall, and it's not at all
uncommon for LZW to be a net loss (i.e., less than 3:1 overall compression).
LZW *does* work well on simpler images such as line drawings, which is why
GIF handles that sort of image so well. When a JPEG file is made from
full-color photographic data, using a quality setting just high enough to
prevent visible loss, the JPEG will typically be a factor of four or five
smaller than a GIF file made from the same data.
Gray-scale images do not compress by such large factors. Because the human
eye is much more sensitive to brightness variations than to hue variations,
JPEG can compress hue data more heavily than brightness (gray-scale) data.
A gray-scale JPEG file is generally only about 10%-25% smaller than a
full-color JPEG file of similar visual quality. But the uncompressed
gray-scale data is only 8 bits/pixel, or one-third the size of the color
data, so the calculated compression ratio is much lower. The threshold of
visible loss is often around 5:1 compression for gray-scale images.
The exact threshold at which errors become visible depends on your viewing
conditions. The smaller an individual pixel, the harder it is to see an
error; so errors are more visible on a computer screen (at maybe 70
dots/inch) than on a high-quality color printout (300 or more dots/inch).
Thus a higher-resolution image can tolerate more compression ... which is
fortunate considering it's much bigger to start with. The numbers quoted
above are typical for screen viewing. Also note that the threshold of
visible error varies considerably across images.
Subject:  What are good "quality" settings for JPEG?
Most JPEG compressors let you pick a file size vs. image quality tradeoff by
selecting a quality setting. There seems to be widespread confusion about
the meaning of these settings. "Quality 95" does NOT mean "keep 95% of the
information", as some have claimed. The quality scale is purely arbitrary;
it's not a percentage of anything.
In fact, quality scales aren't even standardized across JPEG programs.
The quality settings discussed in this article apply to the free IJG JPEG
software (described in part 2), and to many programs based on it. Other
JPEG implementations, notably Apple's and HSI's, use completely different
quality scales; for instance, Apple's scale runs from 0-4, not 0-100. Some
programs don't even provide a numeric scale, just "high"/"medium"/"low"
style choices. (Fortunately, this doesn't prevent different implementations
from exchanging compressed files.)
In most cases the user's goal is to pick the lowest quality setting, or
smallest file size, that decompresses into an image indistinguishable from
the original. This setting will vary from one image to another and from one
observer to another, but here are some rules of thumb.
For good-quality, full-color source images, the default IJG quality setting
(Q 75) is very often the best choice. This setting is about the lowest you
can go without expecting to see defects in a typical image. Try Q 75 first;
if you see defects, then go up.
If the image was less than perfect quality to begin with, you might be able
to drop down to Q 50 without objectionable degradation. On the other hand,
you might need to go to a *higher* quality setting to avoid further loss.
This is often necessary if the image contains dithering or moire patterns
(see " What are some rules of thumb for converting GIF images to JPEG?").
Except for experimental purposes, never go above about Q 95; using Q 100
will produce a file two or three times as large as Q 95, but of hardly any
better quality. If you see a file made with Q 100, it's a pretty sure sign
that the maker didn't know what he/she was doing.
If you want a very small file (say for preview or indexing purposes) and are
prepared to tolerate large defects, a Q setting in the range of 5 to 10 is
about right. Q 2 or so may be amusing as "op art". (It's worth mentioning
that the current IJG software is not optimized for such low quality factors.
Future versions may achieve better image quality for the same file size at
low quality settings.)
Subject:  Where can I get JPEG software?
See part 2 of this FAQ for recommendations about programs for particular
systems. Part 2 also tells where to find free C source code for
implementing JPEG, in case you want to write your own programs using JPEG.
The comp.graphics FAQs and the alt.binaries.pictures FAQ are more general
sources of information about graphics programs available on the Internet
(see " Where are FAQ lists archived?").
Subject:  How do I view JPEG images posted on Usenet?
Image files posted on the alt.binaries.pictures.* newsgroups are usually
"uuencoded". Uuencoding converts binary image data into text that can
safely be posted. Most posters also divide large posts into multiple parts,
since some news software can't cope with big articles. Before your viewer
will recognize the image, you must combine the parts into one file and run
the text through a uudecode program. (This is all true for GIF as well as
JPEG, by the way.) There are programs available to automate this process.
For more info see the alt.binaries.pictures FAQ, which is available from
rtfm.mit.edu:/pub/usenet/news.answers/pictures-faq/part[1-3], or on WWW at
(see also " Where are FAQ lists archived?").
Subject:  What is color quantization?
Most people don't have full-color (24 bit per pixel) display hardware.
Typical display hardware stores 8 or fewer bits per pixel, so it can display
256 or fewer distinct colors at a time. To display a full-color image, the
computer must choose an appropriate set of representative colors and map the
image into these colors. This process is called "color quantization".
(This is something of a misnomer; "color selection" or "color reduction"
would be a better term. But we're stuck with the standard usage.)
Clearly, color quantization is a lossy process. It turns out that for most
images, the details of the color quantization algorithm have *much* more
impact on the final image quality than do any errors introduced by JPEG
itself (except at the very lowest JPEG quality settings). Making a good
color quantization method is a black art, and no single algorithm is best
for all images.
Since JPEG is a full-color format, displaying a color JPEG image on
8-bit-or-less hardware requires color quantization. The speed and image
quality of a JPEG viewer running on such hardware are largely determined by
its quantization algorithm. Depending on whether a quick-and-dirty or
good-but-slow method is used, you'll see great variation in image quality
among viewers on 8-bit displays, much more than occurs on 24-bit displays.
On the other hand, a GIF image has already been quantized to 256 or fewer
colors. (A GIF always has a specific number of colors in its palette, and
the format doesn't allow more than 256 palette entries.) GIF has the
advantage that the image maker precomputes the color quantization, so
viewers don't have to; this is one of the things that make GIF viewers
faster than JPEG viewers. But this is also the *disadvantage* of GIF:
you're stuck with the image maker's quantization. If the maker quantized to
a different number of colors than what you can display, you'll either waste
display capability or else have to requantize to reduce the number of colors
(which usually results in much poorer image quality than quantizing once
from a full-color image). Furthermore, if the maker didn't use a
high-quality color quantization algorithm, you're out of luck --- the image
For this reason, JPEG promises significantly better image quality than GIF
for all users whose machines don't match the image maker's display hardware.
JPEG's full color image can be quantized to precisely match the viewer's
display hardware. Furthermore, you will be able to take advantage of future
improvements in quantization algorithms, or purchase better display
hardware, to get a better view of JPEG images you already have. With a GIF,
you're stuck forevermore with what was sent.
A growing number of people have better-than-8-bit display hardware already:
15-bit "hi-color" PC displays, true 24-bit displays on workstations and
Macintoshes, etc. For these people, GIF is already obsolete, as it cannot
represent an image to the full capabilities of their display. JPEG images
can drive these displays much more effectively.
In short, JPEG is an all-around better choice than GIF for representing
photographic images in a machine-independent fashion.
It's sometimes thought that a JPEG converted from a GIF shouldn't require
color quantization. This is false: even when you feed a 256-or-less-color
GIF into JPEG, what comes out of the decompressor is not 256 colors, but
thousands of colors. This happens because JPEG's lossiness affects each
pixel a little differently, so two pixels that started with identical colors
will usually come out with slightly different colors. Considering the whole
image, each original color gets "smeared" into a group of nearby colors.
Therefore quantization is always required to display a color JPEG on a
colormapped display, regardless of the image source.
The same effect makes it nearly meaningless to talk about the number of
colors used by a JPEG image. Even if you tried to count the number of
distinct pixel values, different JPEG decoders would give you different
results because of roundoff error differences. I occasionally see posted
images described as "256-color JPEG". This tells me that the poster
(a) hasn't read this FAQ and (b) probably converted the JPEG from a GIF.
JPEGs can be classified as color or gray-scale, but number of colors just
isn't a useful concept for JPEG, any more than it is for a real photograph.
Subject:  What are some rules of thumb for converting GIF images to JPEG?
Converting GIF files to JPEG is a tricky business --- you are piling one set
of limitations atop a quite different set, and the results can be awful.
Certainly a JPEG made from a GIF will never be as good as a JPEG made from
true 24-bit color data. But if what you've got is GIFs, and you need to
save space, here are some hints for getting the best results.
With care and a clean source image, it's often possible to make a JPEG of
quality equivalent to the GIF. This does not mean that the JPEG looks
pixel-for-pixel identical to the GIF --- it won't. Especially not on an
8-bit display, because the color quantization process used to display the
JPEG probably won't quite match the quantization process used to make the
GIF from the original data (see " What is color quantization?"). But
remember that the GIF itself is not all that faithful to the full-color
original, if you look at individual pixels. Looking at the overall image,
a converted JPEG can look as good as its GIF source. Some people claim that
on 24-bit displays, a carefully converted JPEG can actually look better than
the GIF source, because dither patterns have been eliminated. (More about
dithering in a moment.)
On the other hand, JPEG conversion absolutely *will* degrade an unsuitable
image or one that is converted carelessly. If you are not willing to take
the amount of trouble suggested below, you're much better off leaving your
GIF images alone. Simply cranking the JPEG quality setting up to a very
high value wastes space (which defeats the whole point of the exercise, no?)
and some images will be degraded anyway.
The first rule is never to convert an image that's not appropriate for JPEG
(see " When should I use JPEG, and when should I stick with GIF?").
Large, high-visual-quality photographic images are usually the best source
material. And they take up lots of space in GIF form, so they offer
significant potential space savings. (A good rule of thumb is not to bother
converting any GIF that's much under 100 Kbytes; the potential savings isn't
worth the hassle.)
The second rule is to look at each JPEG, to make sure you are happy with it,
before throwing away the corresponding GIF. This will give you a chance to
re-do the conversion with a higher quality setting if necessary. Also
compare the file sizes --- if the image isn't suitable JPEG material, a JPEG
file of reasonable quality may come out *larger* than the GIF.
The third rule is to get rid of the border. Many people have developed
an odd habit of putting a large single-color border around a GIF image.
While useless, this is nearly free in terms of storage cost in GIF files.
It is NOT free in JPEG files, either in storage space or in decoding time;
and the sharp border boundary can create visible artifacts ("ghost" edges).
Furthermore, when viewing a bordered JPEG on an 8-bit display, the quantizer
will think the border color is important because there's so much of it, and
hence will waste color palette entries on the border, thus actually reducing
the displayed quality of the main part of the image! So do yourself a favor
and crop off any border before JPEGing.
The fourth rule is to know where the image came from. Repeated GIF<=>JPEG
conversions are guaranteed to turn an image into mush, because you pay a
quality-loss price on each round trip. Don't reconvert images that have
been converted before.
Gray-scale images usually convert without much problem. When using cjpeg,
be sure to use the -gray switch. (Otherwise, cjpeg treats a GIF as color
data; this works, but it wastes space and time if the image is really only
gray-scale.) Quality settings around the default (75) are usually fine.
Color images are much trickier. Color GIFs of photographic images are
usually "dithered" to fool your eye into seeing more than the 256 colors
that GIF can actually store. If you enlarge the image, you will find that
adjacent pixels are often of significantly different colors; at normal size
the eye averages these pixels together to produce the illusion of an
intermediate color value. The trouble with dithering is that, to JPEG, it
looks like high-spatial-frequency color noise; and JPEG can't compress noise
very well. The resulting JPEG file is both larger and of lower image
quality than what you would have gotten from JPEGing the original full color
image (if you had it). To get around this, you need to "smooth" the GIF
image before compression. Smoothing averages together nearby pixels, thus
approximating the color that you thought you saw anyway, and in the process
getting rid of the rapid color changes that give JPEG trouble. Proper use
of smoothing will both reduce the size of the compressed file and give you a
better-looking output image than you'd get without smoothing.
With the IJG JPEG software (cjpeg or derived programs), a simple smoothing
capability is built in. Try "-smooth 10" or so when converting GIFs.
Values of 10 to 25 seem to work well for high-quality GIFs. GIFs with
heavy-handed dithering may require larger smoothing factors. (If you can
see regular fine-scale patterns on the GIF image even without enlargement,
then strong smoothing is definitely called for.) Too large a smoothing
factor will blur the output image, which you don't want. If you are an
image processing wizard, you can also do smoothing with a separate filtering
program, but appropriate use of such tools is beyond the scope of this FAQ.
Quality settings around 85 (a bit higher than default) usually work well
when converting color GIFs, assuming that you've picked a good smoothing
factor. You may need still higher quality settings if you can't hide the
dithering pattern with a reasonable smoothing factor. Really badly dithered
GIFs are best left as GIFs.
Don't expect JPEG files converted from GIFs to be as small as those created
directly from full-color originals. The dithering noise wastes space, but
you won't be able to smooth away all the noise without blurring the image.
Typically, a good-quality converted JPEG will be one-half to one-third the
size of the GIF file, not one-fourth as suggested in section 4. If the JPEG
comes out much more than half the size of the GIF, this is a good sign that
the image shouldn't be converted at all.
The upshot of all this is that "cjpeg -quality 85 -smooth 10" is probably a
good starting point for converting color GIFs. But if you care about the
image, you'll want to check the results and maybe try a few other settings.
Blindly converting a large GIF library at this or any other setting is a
recipe for disaster.
Subject:  Does loss accumulate with repeated compression/decompression?
It would be nice if, having compressed an image with JPEG, you could
decompress it, manipulate it (crop off a border, say), and recompress it
without any further image degradation beyond what you lost initially.
Unfortunately THIS IS NOT THE CASE. In general, recompressing an altered
image loses more information. Hence it's important to minimize the number
of generations of JPEG compression between initial and final versions of an
It turns out that if you decompress and recompress an image at the same
quality setting first used, little or no further degradation occurs. This
means that you can make local modifications to a JPEG image without material
degradation of other areas of the image. The areas you change will degrade,
however. Counterintuitively, this works better the lower the quality
setting. But you must use *exactly* the same setting, or all bets are off.
Also, the decompressed image must be saved in a full-color format; if you do
JPEG=>GIF=>JPEG, the color quantization step loses lots of information.
Unfortunately, cropping doesn't count as a local change! JPEG processes
the image in small blocks, and cropping usually moves the block boundaries,
so that the image looks completely different to JPEG. You can take
advantage of the low-degradation behavior if you are careful to crop the
top and left margins only by a multiple of the block size (typically 16
pixels), so that the remaining blocks start in the same places.
The bottom line is that JPEG is a useful format for archival storage and
transmission of images, but you don't want to use it as an intermediate
format for sequences of image manipulation steps. Use a lossless 24-bit
format (PPM, RLE, TIFF, etc) while working on the image, then JPEG it when
you are ready to file it away. Aside from avoiding degradation, you will
save a lot of compression/decompression time this way :-).
Subject:  What is progressive JPEG?
A simple or "baseline" JPEG file is stored as one top-to-bottom scan of the
image. Progressive JPEG allows the image to be stored as a series of scans
of increasing quality. The first scan shows the image at the equivalent of
a very low quality setting, and therefore it takes very little space.
Successive scans improve the quality. Each scan adds to the data already
provided, so that the total storage requirement is about the same as for a
baseline JPEG image of the same quality as the final scan. (Basically,
progressive JPEG is just a rearrangement of the same data into a more
The advantage of progressive JPEG is that if an image is being viewed
on-the-fly as it is transmitted, the viewer sees an approximation to the
whole image very quickly, with gradual improvement of quality as one waits
longer; this is much nicer than a slow top-to-bottom display of the image.
The disadvantage is that each scan takes about the same amount of
computation to display as a whole baseline JPEG file would. So progressive
JPEG only makes sense if one has a decoder that's much faster than the
Up until recently, there weren't many applications in which progressive JPEG
looked attractive, so hardly any implementations of it are available. But
with the popularity of WWW browsers running over rather slow modem links,
and with the ever-increasing horsepower of personal computers, progressive
JPEG is starting to look like a win for WWW. I expect that progressive JPEG
capability will start to appear in WWW browsers during 1995.
Subject:  Isn't there a lossless JPEG?
There's a great deal of confusion on this subject. The JPEG standard does
include a truly lossless compression algorithm, i.e., one that guarantees
its decompressed output is bit-for-bit identical to the original input.
However, this lossless mode has almost nothing in common with the regular
lossy JPEG algorithm, and it offers much less compression.
Lossless JPEG typically compresses full-color data by around 2:1. Lossless
JPEG works well only on continuous-tone images; it does not provide useful
compression of palette-color images or low-bit-depth images. (The JBIG
standard is considered superior to lossless JPEG for images of less than
about 6 bits/sample.)
At present, very few implementations of lossless JPEG exist. The PVRG code
mentioned in part 2 handles lossless JPEG. Another free implementation
is available from ftp.cs.cornell.edu:/pub/multimed/ljpg.tar.Z; this is a
smaller program that handles *only* lossless JPEG.
Cranking a regular JPEG implementation up to its maximum quality setting
*does not* get you lossless storage; lossless JPEG is a fundamentally
different method. Even at the maximum possible quality setting, regular
JPEG cannot be lossless because it is subject to roundoff errors in various
calculations. The roundoff errors are nearly always too small to be seen,
but they will accumulate if you put the image through multiple cycles of
Many implementations won't even let you get to the maximum possible setting,
because it's such an inefficient way to use regular JPEG. With the IJG JPEG
software, for example, you have to say not only "-quality 100" but also
"-sample 1x1" to eliminate all deliberate loss of information. The
resulting files are far larger and of only fractionally better quality than
files generated at more reasonable settings. And they're still slightly
lossy! If you really need lossless storage, don't try to approximate it
with regular JPEG.
Subject:  Why all the argument about file formats?
Strictly speaking, JPEG refers only to a family of compression algorithms;
it does *not* refer to a specific image file format. The JPEG committee was
prevented from defining a file format by turf wars within the international
Since we can't actually exchange images with anyone else unless we agree on
a common file format, this leaves us with a problem. In the absence of
official standards, a number of JPEG program writers have just gone off to
"do their own thing", and as a result their programs aren't compatible with
The closest thing we have to a standard JPEG format is some work that's been
coordinated by people at C-Cube Microsystems. They have defined two
JPEG-based file formats:
* JFIF (JPEG File Interchange Format), a "low-end" format that transports
pixels and not much else.
* TIFF/JPEG, aka TIFF 6.0, an extension of the Aldus TIFF format. TIFF is
a "high-end" format that will let you record just about everything you
ever wanted to know about an image, and a lot more besides :-). TIFF
is far more complex than JFIF, and is generally less transportable,
because different vendors have often implemented slightly different
and incompatible subsets of TIFF. It's not likely that adding JPEG to
the mix will do anything to improve this situation.
Both of these formats were developed with input from all the major vendors
of JPEG-related products; it's reasonably likely that future commercial
products will adhere to one or both standards.
JFIF has emerged as the de-facto standard on Usenet. JFIF is simpler than
TIFF and is available now; the TIFF 6.0 spec for incorporating JPEG is not
widely implemented, partly because it has some serious design flaws. It is
likely that the TIFF 6.0 JPEG section will be changed significantly before
widespread adoption occurs. Even when TIFF/JPEG is well defined, the JFIF
format is likely to be a widely supported "lowest common denominator";
TIFF/JPEG files may never be as transportable as JFIF.
A particular case of wide interest is Apple's Macintosh QuickTime software.
QuickTime uses a JFIF-compatible format wrapped inside the Mac-specific PICT
structure. Conversion between JFIF and PICT/JPEG is pretty straightforward,
and several Mac programs are available to do it (see Mac portion of part 2).
If you have an editor that handles binary files, you can even strip a
PICT/JPEG file down to JFIF by hand; see the next section for details.
Another particular case is Handmade Software's DOS programs (GIF2JPG/JPG2GIF
and Image Alchemy). These programs are capable of reading and writing JFIF
format. By default, though, they write a proprietary format developed by
HSI. This format is NOT readable by any non-HSI programs and should not be
used for Usenet postings. Use the -j switch to get JFIF output. (This
applies to old versions of these programs; the current releases emit JFIF
format by default. You still should be careful not to post HSI-format
files, unless you want to get flamed by people on non-PC platforms.)
News flash: the ISO JPEG committee seems to have won their turf wars. They
will define a complete file format spec called "SPIFF" in the next version
of the JPEG standard. It's pretty late in the game though, so whether this
will have much impact on real-world files remains to be seen.
Subject:  How do I recognize which file format I have,
and what do I do about it?
If you have an alleged JPEG file that your software won't read, it's likely
to be HSI format or some other proprietary JPEG-based format. You can tell
what you have by inspecting the first few bytes of the file:
1. A JFIF-standard file will start with the four bytes (hex) FF D8 FF E0,
followed by two variable bytes (often hex 00 10), followed by 'JFIF'.
2. If you see FF D8 at the start, but not the 'JFIF' marker, you may have a
"raw JPEG" file. This is probably decodable as-is by JFIF software ---
it's worth a try, anyway.
3. HSI files start with 'hsi1'. You're out of luck unless you have HSI
software. Portions of the file may look like plain JPEG data, but they
usually won't decompress properly with non-HSI programs.
4. A Macintosh PICT file, if JPEG-compressed, will have several hundred
bytes of header (often 726 bytes, but not always) followed by JPEG data.
Look for the 3-byte sequence (hex) FF D8 FF. The text 'Photo - JPEG'
will usually appear shortly before this header, and 'AppleMark' or
'JFIF' will usually appear shortly after it. Strip off everything
before the FF D8 FF and you will usually be able to decode the file.
(This will fail if the PICT image is divided into multiple "bands",
but that is uncommon.)
5. If the file came from a Macintosh, it could also be a standard JFIF
file with a MacBinary header attached. In this case, the JFIF header
will appear 512 bytes into the file. Get rid of the first 512 bytes
and you're set.
6. Anything else: it's a proprietary format, or not JPEG at all. If you
are lucky, the file may consist of a header and a raw JPEG data stream.
If you can identify the start of the JPEG data stream (look for FF D8),
try stripping off everything before that.
HSI files used to be rather common in alt.binaries.pictures.* postings,
although thankfully they have gotten less so. You can spot an HSI posting
by looking at the first few characters of the uuencoded data. The
characteristic HSI pattern is
whereas standard JFIF files begin with
If you learn to spot the HSI pattern, you can save yourself the trouble
of downloading unusable files.
Subject:  How does JPEG work?
You can find an introduction and references for further reading in the
comp.compression FAQ, which is available from the news.answers archive at
rtfm.mit.edu, in files /pub/usenet/news.answers/compression-faq/part[1-3]
(see also " Where are FAQ lists archived?").
The comp.compression FAQ is also a good starting point for information on
other state-of-the-art image compression methods, such as wavelets and
fractals. A quick comparison: wavelets are likely to be the basis of the
next generation of image-compression standards, but they are perhaps 10
years behind JPEG in the standardization pipeline. Fractals have been
terribly over-hyped by their chief commercial proponent, and it's difficult
to say what their true capabilities are.
Subject:  What about arithmetic coding?
The JPEG spec defines two different "back end" modules for the final output
of compressed data: either Huffman coding or arithmetic coding is allowed.
The choice has no impact on image quality, but arithmetic coding usually
produces a smaller compressed file. On typical images, arithmetic coding
produces a file 5 to 10 percent smaller than Huffman coding. (All the
file-size numbers previously cited are for Huffman coding.)
Unfortunately, the particular variant of arithmetic coding specified by the
JPEG standard is subject to patents owned by IBM, AT&T, and Mitsubishi.
Thus *you cannot legally use JPEG arithmetic coding* unless you obtain
licenses from these companies. (Patent law's "experimental use" exception
allows people to test a patented method in the context of scientific
research, but any commercial or routine personal use is infringement.)
I recommend that people not use JPEG arithmetic coding; the space savings
isn't great enough to justify the potential legal hassles. In particular,
arithmetic coding *should not* be used for any images to be exchanged on
Usenet. Even if you don't care about US patent law, other folks do.
Subject:  Could an FPU speed up JPEG? How about a DSP chip?
Since JPEG is so compute-intensive, many people suggest that using an
FPU chip (a math coprocessor) should speed it up. This is not so.
Most production-quality JPEG programs use only integer arithmetic and so
they are unaffected by the presence or absence of floating-point hardware.
It is possible to save a few operations by doing the DCT step in floating
point. On most PC-class machines, FP math is enough slower than integer
math that the overall speed is still much worse with FP. Some high-priced
workstations and supercomputers have fast enough FP hardware to make an FP
DCT method be a win.
DSP (digital signal processing) chips are ideally suited for fast repetitive
integer arithmetic, so programming a DSP to do JPEG can yield significant
speedups. DSPs are starting to be found as add-ons for PCs and small
workstations, so you can expect to see DSP-based JPEG programs popping up.
Subject:  Isn't there an M-JPEG standard for motion pictures?
As was stated in section 1, JPEG is only for still images. Nonetheless,
you will frequently see references to "motion JPEG" or "M-JPEG" for video.
*There is no such standard*. Various vendors have applied JPEG to
individual frames of a video sequence, and have called the result "M-JPEG".
Unfortunately, in the absence of any recognized standard, they've each done
it differently. The resulting files are usually not compatible across
MPEG is the recognized standard for motion picture compression. It uses
many of the same techniques as JPEG, but adds inter-frame compression to
exploit the similarities that usually exist between successive frames.
Because of this, MPEG typically compresses a video sequence by about a
factor of three more than "M-JPEG" methods can. The disadvantages of MPEG
are (1) it requires much more computation to generate the compressed
sequence (since detecting visual similarities is hard for a computer), and
(2) it's difficult to edit an MPEG sequence on a frame-by-frame basis
(since each frame is intimately tied to the ones around it). This latter
problem has made "M-JPEG" methods rather popular for video editing products.
It's a shame that there isn't a recognized M-JPEG standard. But there
isn't, so if you buy a product identified as "M-JPEG", be aware that you
are probably locking yourself into that one vendor.
See the MPEG FAQ for more information about MPEG.
Subject:  Where are FAQ lists archived?
Many FAQs are crossposted to news.answers. Well-run netnews sites will have
the latest versions available in that newsgroup. However, there are a *lot*
of postings in news.answers, and they can be hard to sort through.
The latest versions of news.answers postings are archived at rtfm.mit.edu.
You can retrieve this FAQ by FTP as /pub/usenet/news.answers/jpeg-faq/part1
and /pub/usenet/news.answers/jpeg-faq/part2. If you have no FTP access,
send e-mail to firstname.lastname@example.org containing the lines
(If you don't get a reply, the server may be misreading your return address;
add a line such as "path myname@mysite" to specify your correct e-mail
address to reply to.) For more info about the FAQ archive, retrieve the
The same FAQs are also available in the World Wide Web; see the index at
http://www.cis.ohio-state.edu/hypertext/faq/usenet/FAQ-List.html. This FAQ
organizer, Independent JPEG Group
email@example.com or firstname.lastname@example.org
Información Tecnica >