Slashdot Mirror
GIF Slips Away From Unisys; Your Move, IBM
Twenty years ago, Terry Welch's improvement on Lempel-Ziv compression appeared in IEEE Computer magazine. The authors of unix 'compress' and the GIF standard incorporated that algorithm without realizing it was patent-pending. When the submarine patent surfaced ten years later, its new owner Unisys intimidated developers and web authors into moving away from GIFs, inspiring the creation of a better standard, though sadly still a less popular one. Today, July 7, 2004, Unisys's last LZW patent (in Canada) expires, leaving GIF once again free... almost. See, there's the small matter of IBM's patent, granted on the same algorithm, which is valid for another two years. That still has a chilling effect on GIF development, though the consensus seems to be that IBM would lose any court action it tried to bring. So how about it, IBM? You've got nothing to lose! Want to make a lot of geeks happy and release that final patent into the public domain?
I'm not sure on the merits of the GIF format after all these years, the only thing it brings to the web expierence is flashing adverts, PNG provides full alpha-transparency which is really required for the future of web design.
Because GIF is used MUCH more, so people writing software that make use of images in general (browsers, image editors, etc.) have to deal with this patented algorithm, or risk losing users because their software doesn't support one of the most widely used formats.
"The best laid plans of mice and men gang oft agley..." - ROBERT BURNS
It's hard to do away with GIF because GIF's are animated. PNG's are not. There's the MNG standard, which is basically an animated PNG, but it isn't widely supported yet.
Real link is here
Save your wrists today - switch to Dvorak
Quick source view of the main slashdot page shows that "gif" is found about 50 times.
"png" is found twice -- both of which are related to the original post.
Now you know why we care. The web community uses gif more than png. For better or worse...
Davak
You're talking about an obsolete technology [GIF] that nobody cares about.
I'd question that. Check Google images and see how many web sites still exclusively use .gifs. Not to mention a certain main-stream browser whose support for .pngs is still patchy.
I guess you and I have different definitions of "obsolete".
This is where the serious fun begins.
You can't make animations with PNG files....
Sure you can, only the result is called MNG.
Belief is the currency of delusion.
Indeed, the web would be much more beautiful if IE supported alpha transparency in PNGs.
Sticking feathers up your butt does not make you a chicken - Tyler Durden
Yeah, I know there are deflated TIFFs, but they can be like "wha...?" in the prepress world.
The original Welch paper is pretty readable:
Terry A. Welch, "A Technique for High Performance Data Compression", IEEE Computer, Vol. 17, No. 6, 1984, pp. 8-19.
If you don't want to go to a library and look that up, then Google will find you about 12000 hits on "Welch LZW", and the first few all seem to be exactly what you want.
But GIF is the worst format for animations ever!
Alpha channels are useful for antialiased drawings that are not dependant on a specific background.
But Slashdot's current HTML isn't standard either. I've had modern browsers completely barf on some of the slashcode output, splitting the screen in half. They are still using the CENTER and FONT tags, and ANY properly-written older browser will ignore newer HTML or CSS stuff. If you are running a low end machine, stick with a properly written low end browser. Don't blame new standards for odd rendering in old, crappy browsers.
Because of the LZW compression algorhythm.
It's still superior to PNG's compression and I hazard a guess that PNG can be modified to use LZW.
the whole ruckus was because they did publish the algorithm widely and it got used widely - and only then did they reveal their submarine.
the algo is/was very widely known.
world was created 5 seconds before this post as it is.
AFAIK, lossless means that no matter how many times you RE-save an image, it doesnt lose image data; in which case, yes, a GIF IS lossless.
Actually that patent is being used in IBM's (second amended) counterclaims in the SCO v IBM case.
JeR
bloating their sites with transparent effects where it is not needed.
Every once in a while somebody seems to open their mouth without realizing they have no clue what they are talking about.
How exactly is a transparent image bloat? I did a test. As a gif a logo I have is 3.32K without alpha and 3.33K with alpha. A PNG (both regular and alpha) it was 3.45K. That should dispel both the claims that PNG is bigger and that transparency adds bloat.
And what do you do everytime you change a websites background color? Change the image?
Ever tried Google? Try searching "GIF file format".
.pdf at The Scientist and Engineer's Guide to Digital Signal Processing website.
Even the first link has info on the format, except that particular info is hard to read. I've seen better documets about the subject.
For just the plain LZW algorithm, try this
- Jeppe Jääkarhu
Except the browsers all support PNG (minus alpha, but GIF and JPG don't have that anyway), where as most MP3 players can't play Vorbis..
...MNG code is not very good to begin with, I believe MNG support was ditched from Mozilla as well, which makes it supported in approximately 0% of the web browsers out there.
I use PNG quite a bit, but mainly as a competitor to TIF files, but I do prefer to use PNG over GIF in websites too. However, I'm only using non-transparent, plain PNGs for maximum compatibility.
Animated GIFs? Oh, right. I turned those off, along with pop-ups. If I wanted that, I'd actually use flash or something like that. I figure either you don't block stuff (which means GIF + flash) or you block stuff, in which case you don't see either. Either way, I don't see much room for GIF files...
Kjella
Live today, because you never know what tomorrow brings
Just use Sleight to make PNG transparency work with IE on your site.
To reiterate, GIF is essentially lossless, but only in two conditions: The image is less than 256 colors raw, or the image was a gif originally (basically the same condition.) Now, PNG is also lossless, and while GIF is still the most common standard, Mozilla supports PNG well, and IE pretends to support it resulting in an okay method, meaning it doesn't have to remain standard. After all, [intentionally ridiculous analogy]the horse and buggy was once the standard for transportation.[/intentionally ridiculous analogy]
I'm always right and I can prove it, because to the best of my knowledge, I've never been wrong.
GIF is pronounced JIF.
.GIF files are pronounced JIF
Bob Berry, the developer of CompuShow for CompuServe (remember them - the people that invented the GIF format?) included with it an animated GIF89 format file that had a picture of him. It had a speach bubble with him saying:
Oh, incidentally, it's pronounced "JIF"
A quick google later and I've found a web site which has this, and other evidence that
http://www.olsenhome.com/gif/
Thanks for playing
Exactly. I'm not a full-time web developer, so I don't even think on these issues too hard most of the time. Recently I had to write a little network-monitoring app. I coded the output in standards-compliant XHTML 1.1, standards-compliant CSS for styling stuff, and I used PNGs with transparent backgrounds for certain little icons. I only tested my app in Firefox (yeah, my mistake). Later someone who actually uses IE tried to use my little web app, and found gray background squares around all my supposedly transparent-backgrounded images. Sucked. Now I know what all you web developers already knew - I have to put a background color on these pngs which matches the background they're placed on that I specified in my stylesheet. How redundant and stupid.
11*43+456^2
PNG transparency works just fine in Internet Explorer. It's just a pain in the ass.
This website will tell you how to turn it on. You can see it working on my website.
No idea why it's not on by default, but if it works...
Just to set the record straight:
When I led the process of drafting the PNG specification, GIF animation did not yet exist. Animation was not part of the original GIF specification. The GIF89a specification *did* offer a mechanism for including multiple images in a single file, and a very basic (but, in retrospect, effective) mechanism for replacing only a specified part of the preceding image. But whether this was supposed to be animation with a time component was never defined, and there was in fact no way to specify how long each frame was supposed to appear, probably because the real intent was to be able to compose a single final still image from many sections. Multiple image GIFs were a footnote to the GIF specification which hardly anybody used until Netscape stepped in.
Netscape's animated GIF format was a clever hack on top of this: they defined a new GIF chunk to specify the pause between frames.
Here's the kicker: Netscape was repeatedly invited to participate in the PNG design process. They had someone reading the list, I gather, but they never offered any suggestions or contributions. If they had, they would likely have been considered very seriously.
But instead, the first we heard of GIF animation was its public release in Netscape (2.0 beta, if I recall correctly). They could have contributed to the design of a PNG or MNG that did include animation and, by way of that compelling feature, would have been more likely to quickly replace GIF. But they didn't.
We (the PNG designers) did consider retrofitting animation into PNG when Netscape's animated GIF appeared. In fact, I lobbied for that at one point. Unfortunately we had already finalized the functional specification and there was no hope of reaching agreement on how to "jam in" the animation feature at the last minute on top of an otherwise pretty elegant image format.
Instead, the MNG group was formed to create a specification for a powerful lossless animation format. And they succeeded -- but MNG has yet to really catch fire, and animated vector formats like SWF and SVG are gradually replacing animated GIF anyway for most purposes. At the end of the day, lossless bitmap animation is a pretty bandwidth-intensive proposition.
Check out the Apostrophe open-source CMS: http://www.apostrophenow.com/
Because Unisys only ever threatened users of the GIF format with legal action; they considered other uses of LZW as acceptable.
See http://www.daltonlp.com/daltonlp.cgi?item_type=1&i tem_id=217 for how to get pngs to display transparency on IE 5.5 and IE 6. This is now a well-known technique that works pretty well universally. Combine with CSS or javascript and you should be able to use pngs entirely.
No. JPEG at max quality looks perfect to the human eye, but it still has differences from the original image. Lossy compression should be avoided in situations where images are going to be decoded and recoded many times, as these errors build up to the point where they can become noticeable.
Also: make sure your PNG encoder is configured correctly. In most cases you want to be using the 'adaptive' filter.
``Basically it's just a much nicer version of GIF's transparency.''
Or, actually, what GIFs have is transparency, whereas PNGs have an alpha channel, which allows for the specification of translucency - or opacity which is the opposite quality.
Please correct me if I got my facts wrong.
In some limited circumstances it's possible to use alpha-transparency while gracefully degrading in IE.
For example, if you are using a solid-colour or almost-solid-colour partially-transparent image to achieve some kind of shading or tinting of the underlying background, you can do this and let IE display it as solid rather than transparent. People who only use IE will never know it was meant to be transparent and thus won't care.
The major trip-up here is that IE renders alpha-transparent PNG onto an unpredictable background colour. However, you can bypass this by adding a background colour chunk (bKGD, or something like that; it's been a while) specifying which solid colour you wish IE to render to. It will then render to that color and create the image with that color "showing through".
The limitations of IE's rendering are due to how IE was originally built to handle images. The image loaders hand the rendering component some kind of bitmap and a 1-bit transparency mask. This was a good choice at the time, but then alpha-transparent PNG came along, and since at the time GDI didn't have any mechanism to support alpha-transparency they just bodged it with the background color. At the time it didn't matter because no-one was using PNG anyway.
The new version of IE will hopefully support alpha-transparency since as of Windows 2000 GDI supports 32-bit images with (alpha,r,g,b) components, and there's already a PNG loader in the gdiplus library, so supporting it will be pretty trivial.
I'll bite.
...) changes an image in such a way that humans don't percieve the difference, but it can be stored more efficiently. At lower qualities, you will begin to notice some artifacts. It can go all the way down to a completely useless collection of pixels. It's a common misconception that 100% quality JPEG images are not distorted in any way. I don't know what 100% means, other than the lowest compression your encoder supports.
JPEG, like MPEG (and Vorbis, Theora,
PNG images, on the other hand, encode the image exactly as it looks. Basically, a PNG image is a collection of pixels, some metadata, optionally compressed with deflate (same algorithm used by gzip).
JPEG images are the better choice for photograpic images (which is what they are intended for), where the exact pixel colors don't matter that much. PNG is better for line drawings, text, high contrast images; basically anything that doesn't bear slight changes to the colors. For large images, JPEG can be significantly smaller, making the case for using JPEG for screen dumps and such.
Please correct me if I got my facts wrong.
If you would like to get MNG back into Mozilla, then you can follow/vote/contribute to Bugzilla bug 18574
http://bugzilla.mozilla.org/show_bug.cgi?id=1857 4
(Please don't post useless comments on that bug)
MOD THE CHILD UP!
Hello? People don't complain that LZW is obvious. They complain that Unisys sat on the patent for ten years while it found its way into standard data formats like gif and compress. Hence the term "submarine patent."
I suspect that it was a combination of several things:
:-(
1) Until recently, the US didn't publish the applications, only the granted patents, and the lifetime of the patent was from date or grant, not date of filing. There was the tendancy for the applicants not to bother rushing things through. (dare I say submarining).
Note that elsewhere, patent applications became public after, IIRC, 18 months and their lifetime started from the date of filing - there was no point in being tardy.
2) I suspect that, back then, searching existing patents was probably a lot more difficult. It wouldn't surprise me if it was all done manually which'd be utterly tedious. It's bad enough with online electronic systems.
3)It is the responsibility of the patenter to inform the patent office if new information/documents comes to light and to update their patents accordingly (I've done this with one of mine) but I guess researchers can move on to other fields, change employers, or simply lose interest. (shrug)
Somehow, your post reminded me of this.
Gamingmuseum.com: Give your 3D accelerator a rest.
For those who love irony, here it is in PNG format.
> Flash? bah, talk about bloated.
The plugin is 200K. Flash movies are typically smaller than a single image. You can rant all you want about flash, but I don't think you can call it bloated.
Actually, you cannot (legally) use a patented technology without permission from the patent holder, even if you don't make money from it.
i'm not sure this web community would be pleased that slashdot's being used as an example of currently-accepted design choices. at least on the front end, slashdot's code is dated and inefficient. alistapart.com has a fun article on how it could be made better.
bingo! mod parent up. everyone should be using PNG now. IE supports png binary transparency just fine. so you can do a straight conversion of all your gif->png and you will have no problem.
also, png's will almost always be smaller if optipng is used to optimize the compression
Marge, get me your address book, 4 beers, and my conversation hat.
Main #1 problem with that: it relies on DirectX, which, if you have any brains, is turned OFF on IE.
Oh wait.
I forgot all of the morons who leave it on.
My Mistake. Guess it will work after all.
Item 2 has been shot down because the majority doesn't rule on matters of punctuation. (pronounciation?)
Actually, the majority do rule on matters of pronounciation when it comes to English. The major linguistic project of English (the Oxford English Dictionary) is a descriptive not a prescriptive document. That means that once a significant minority of English users use or pronounce a word in a certain way, it'll get recorded in the dictionary.
All this is just to say that both "jif" and "gif" are acceptable pronounciations of GIF.
I have a lot of opinions about Cyborgs and Architects
You can indeed get IE 5.5 and above on Windows to display PNG images with alpha transparency, by using IE's built-in DirectX filters against themselves. For example:
r (src='images/next.png', sizingMethod='scale', enabled='true'); display:inline-block"><img src="images/next.png" alt="IE-compatibility link" width="36" height="52" style="filter: progid:DXImageTransform.Microsoft.Alpha(opacity=0) ;"></div>
<div style =" width: 36px; height: 52px; filter: progid:DXImageTransform.Microsoft.AlphaImageLoade
This doesn't interfere with other browsers which support PNG natively because they just see the standard IMG tag and ignore the filter stuff. Whereas on IE, the filter in the IMG tag prevents the (wrong) image from being displayed, and the one in the DIV tag actually does display it properly. Goodness knows why they make you jump through the hoops though; given the IE on MacOS just works, it's obvious that Microsoft as a company don't have a problem with understanding or implementing the specs. Do they just not share code between platforms in Redmond, or what?
And here is how you force IE into properly supporting PNG transparency.
Works like a charm, doesn't introduce any MS "extensions" into your documents, and doesn't do anything if the user is smart enough to be using a web browser that actually supports standards.
- chrish
This is actually doable, it's just convoluted and requires a browser-detect. I did this at a client last year. Google "alphaimageloader png internet explorer" for info.
If you're smart enough to come up with the next great compression algoritm, I encourage you to do so!
Best Buy can have you arrested
IANAL, but this is one of the key ways in which patents differ from, say, trademarks. The patent owner owns the patent and does not lose their intellectual property if they do not enforce it. (Else submarine patents would be impossible.)
Or better yet, http://dean.edwards.name/IE7/ and fix more than just PNG alpha.
The browser could never know if ASP or JSP or neither was used.
Hollow words will burn and hollow men will burn.
GIFs can be used properly to good effect. Just don't convert a high-color picture to GIF automatically. *shrug* Much the same way you could compress an AVI in 16 colors then say AVI sucks.
The question is, why not use JPEG if it results in smaller files and no visible artifacts? I think the answer to that question is, there IS no reason not to. If the compression fits your application, use it. Don't make decisions based on generalizations and textbook answers.
Will there be artifacts? Perhaps. But what is relevant, of course, is whether they are perceptible, and beyond that, whether they are distracting.
Don't just blindly follow rules of thumb. Do a real experiment, like the OP's colleague did. If a certain type of compression works best in your application, then use it. Don't discard options based on theory, especially when direct experience contradicts it.
Oh, and a small nitpick: PNG is not optionally compressed. The compression is mandatory.
IANAL, but my understanding of patent law is that when the patent is granted you must prosecute any infringements you are aware of. No getting to pick your battles. And you must file suit within a reasonable time period.
Failing to do so voids your patent.
At least, I'm almost certain that is how it works here in Australia.
Glenn
The Smrt way to trade CFDs on the ASX