jpeg is worse than gif regarding generation loss in a couple of ways
1: normal image editing software can dependablly be used to perform operations on gif losslessly. with jpeg it can't (both because of the rounding issues and because the transform matrix from the original image isn't generally stored and reused afaict). 2: on gif you can freely rotate crop and perform pixel level moves without loss, on jpeg you are limited to operations that preserve block boundries (rotation and flipping of blocks can be done losslessly but this can only be used to rotate or flip an entire image if the image size is a multiple of the block size).
btw, can you provide evidence that jpeg->8 bit rgb->jpeg is always possible without introducing any rounding errors if the transform matrix from the decode is used in the encode.
Couldn't agree less. The UK has adopted the metric system in pretty much all things, the only notable exception being the reference to miles rather than kilometres in signage. it did EVENTUALLY but really only because the EU forced it down thier throats.
don't confuse the tactical (short term during battle) with the strategic (long term)
strategically the milatary especially during wartime NEEDS civilians to serve it in order to keep functioning. and civilian industry is often very reliant on computers some of them MS based and (indirectly) connected to public networks.
Also, let's not forget, that PNG is in some respects is inferior to GIF. Transparency isn't supported by all displaying apps. (Sure that's largely due to Microsoft, but let's be adults here. You can't demand that users switch their apps for your whims. That's arrogance, which sadly is all too prevalent in the FOSS Community.) gif like (paletized with a single transparent color) pngs are supported just fine and are generally smaller than the equivilent gif.
And still, LZW compressed GIFs still needed to be supported by everything because there were, and still are, a significant number of them deployed iirc while unisys claims thier patent covers decompression those who have analysed it don't belive this would stand up in court whereas compression would.
Also, PNG doesn't support animations, which along is the other reason people use GIFs. (Yes, MNG exists, but it's supported by no one, so it's not an option.) unfortunately true but animations are only a very small proportion of web images.
In the end GIF stuck around because of a combination of legacy support, and the fact that most people simply didn't care. yeah, the fact that most people simply didn't care (just like most people don't care about the risk of being sued for sharing pirate content online until they are sent a personal threat by the ??AA) is almost certainly the main reason.
btw i stopped using gif because the version of paint shop pro i use (a magazine cover CD special release) doesn't support gif due to the patent issue.
No. The FUD spread by the patent the patent cost money and time. Total people sued by Unisys over the GIF patent? Zero. That is nothing. does it really make much difference whether you are actually sued or if you (or your legal department) just think the threat (and unisys certainly sent out threats) is credible enough that you'd better pay up? Things only go to court if one player is an asshole or if there is true dispute, if you know you are going to lose why go to court in the first place?
Conversely, if you look at the successful FOSS projects, you see that the common thread among them is that they aren't clones, they were first-movers. Apache, Perl, Python, Firefox, the Linux kernel (Yeah yeah, BSD386, but it didn't survive). I don't believe this is a coincidence. I believe it's simple economics (in the decision theoretical sense, not the monetary sense). When the niche is already filled, only those motivated by politics will switch away from a stable feature-rich app to a buggy feature-poor app. However when the niche is empty, people will come to you, and you can define the niche. btw linux (clone of minix) apache (a collection of patches to a university project) and firefox (essentially a stripped mozilla) were hardly first movers. and the linux kernel would have had hardly any success at all without the vast array of underapricated clone software built on top of it.
most software development is minor improvements over existing ideas, when the original is barred from you then you have to clone it before you can get up to the inovation stage and the FOSS community doesn't exactly have much option but to avoid anything which has patent threats hanging over it.
False. GIF is lossy, as it is limited to 255 colors per frame. Conversion from a 24-bit image to GIF involves an operation called dithering, which adds noise to the image. In fact, the amount of loss in GIF (called "palette size") can be dialed up and down just as easily as JPEG's quantization scale factor. by the same reasoning you could say every image format is lossy as it can't precisely represent the raw output from your cameras CCD.
yes an image beyond a formats limitations will incur loss during conversion but that doesn't imo make the format lossy in the same way that say JPEG is (otherwise you'd have to consider every image format lossy).
Sure, I could find some variant of PNG that would work, but could I be sure that it would work on every browser? gif like (8 bit or less pallette based with a single transparent color) work just fine accross every major browser back quite a number of versions.
if you are trying to match colors in the png to other colors on the page you should write the png WITHOUT color correction information to make it work right in most browsers (if your editor insists on writing this info use something like pngout to strip it).
Some programs (e.g. photoshop) also have notoriously bad png output support and third party optimisation software (of which i belive pngout is the best arround) is always worthwhile if filesize bothers you.
you might wan't to take a look at png tips for cartoonists which covers a lot of the issues with using png for this type of artwork.
1: it means free software can support gif writing properly (e.g. not using hacks that avoid the patent but also make the files bigger than plain bitmaps) without legal issues. This is a biggie for packages like gimp and imagemagick.
2: likewise for small commercial vendors who unisys fucked arround until they decided that gif support was more trouble than its worth.
3: unisys has in the past gone after some website owners forcing them to either somehow prove thier gifs were produced with licensed software (basically impossible) or pay an anual license fee themselves. It means theese websites are now safe.
Does this mean that Stallman has contemplated a future for the software industry in which patents are gone but (less restrictive) copyrights remain? software patents are FAR worse than software copyrights. While (i belive i can't seem to find the source right now) that stallman is anti copyright like most others in the FOSS community he sees patents as the far larger and more immediate threat that they are.
with copyrighted software i can use it (looking at the code if availible is more risky because i may inadvertantly remember and copy it or be accused of doing so but i can always pay someone else to look at it). then based on my experiances using it safely create a clone.
with patented software (at least under the US system) if i implement the methods in the patent (which may be nessacery for decoding a popular format) then there is no choice but to license from the patent holder at whatever price they set. The general lack of competance of the patent office (granting many obvious, already implemented elsewhere or even duplicate patents) doesn't help either (fighting a patent in court is outside most OSS projects financial means}.
iirc with oem windows there is NO support included at all, you are just told to contact the manufacturer of the PC (admittedly the OEM may get some support but i suspect its much less hours per copy than retail gets)
The ones that need support, or want to support the company for whatever reason, are the ones that pay. and all but the tiniest buisnesses in the western world who just can't risk it with the BSA rent a cops arround.
Sorry the support offered for free for OSS products is pretty hopeless for businesses. If a company has lost 100's of hours of work due a bug or problem, they have the option of complaining the makers of commercial software. and expect say microsoft to take them seriously, unless they have bought a seperate support contract or are a very large customer i doubt ms will.
Our startup honestly wanted to use OSS products. We do not want to spend time for any OSS bug fixing so our main requirement was -official support for all OSS products-. fine but you should understand that support is extra with most propietry software too.
We thought were prepared to pay the price for OSS products, but then we got a price sticker shock. Now behold: QT is $3300 per seat. yeah trolltech (strangely fitting name don't you think) have set themselves up in a neat position to rape commercial software developers for linux, use a freeer toolkit like gtk. IIRC that $3300 does include the distribution though (unlike with MS where you will have to pay for a copy for every device you sell).
We have dropped the development and rewrote everything to C# (MSVS 2005 is ~$700). and what if any support do you actually get at that price?
Embedded Linux from a reputable RT vendor is $25,000 per 5 seats per year.
A Cygwin commercial license will cost tens of thousands of dollars and is only available for large shops. if you are shipping software based on cygwin you are a f*cking idiot anyway. Cygwin is barely tolerable in the controlled environment of your own boxes, once its on machines you don't control expect crashes caused by different apps shipping different versions of cygwin1.dll which don't play nice when loaded at the same time.
ultimately with any software if you wan't good support you will have to pay through the nose for it. The software itself (whether free or propietry) tends to be dirt cheap in comparison.
I remember US Robotics X2 modems could be upgraded to V.90 with a firmware update. Most if not all KFlex users also had the same option from their manufacture. indeed, i belive V90 was specifically designed to be implementable on both X2 and K56Flex hardware (presumablly this is what the grandparent meant by "merged").
or in the case of theese asus devices at asus's risk
imo the fact that asus have made this offer means that they are pretty confident that thier kit will be firmware upgradeable to the final standard (and they almost certainly have insider information from sitting on the comitees).
1: an unhacked copy of OSX will not install on non-apple hardware. 2: i belive some versions of OSX shipped with machines won't install on older machines (so you can't buy a new mac and use its install CD to upgrade older macs)
i don't belive they do anything to stop you buying one upgrade and using it on multiple macs.
please stop spreading this myth, a gif like (paletized with a single transparent color) transparent png will display just fine in IE and will be smaller than the equivilent gif.
yes IE has problems with some transparent pngs but not with gif like ones.
the patent on lzw (which affected gif and some but not all variants of tiff) is gone, some tiff variants almost certainly suffer from other patent issues though.
the REAL problem with tiff is there are so many variants. full tiff support requires an insanely huge library (that will probablly have patent issues somewhere) and partial tiff support is a recipie for confusion.
afaict skype control the master identity servers so they can mitm any connection they wan't. and almost certainly the cops could force them to do so (and if they can't legally force it they could always bribe them instead).
the only secure system is one where you give people your public key and they use that to authenticate you, then assuming whatever means you use to give them the public key is not tampered with and the public key algorithm isn't cracked you are safe.
but public and private keys are generally too big for most people to memorize/easilly handle and with a system as described above you can't easilly change the private key so most systems (including skype) rely on a master server or root certificate to identify that "person holding the private key to public key "8932075623478901254734745678457845304" really is "joeblogs999". if that master server or root certificate cannot be trusted then neither can the system as a whole.
The fact that I know that every security patch will hit my machine fast, usually within a day, is why I use Debian. i beg to differ(the version of gaim in sarge-security is still the same version that was current when that bug was reported)
The official logos have been a documented issue for ages and indeed are not included in a firefox build unless explicitly turned on. Whats new is they don't seem to wan't use of the firefox name either (in which case there of course comes the question of why does the public source and the default binaries built from it use that name).
i see two possibilities for why this has happened.
1: they think they have the ability to push debian into a position of violating its core principles and shipping a mozilla approved firefox. 2: they have got a new legal team who are paraniod about protecting the trademark on the name.
either way i don't see how the likely outcome of this (debian shipping firefox but not calling it firefox) is going to be good for either mozilla or debian.
debian is (and always has been) rather slow on the stable releases and lacking a bit in newbie friendlyness.
but make no mistake, ubuntu (everyones favorite distro recently it seems) relies on debian to do most of thier work for them. From debian ubuntu gets a huge library of software that is kept up to date for them (yes they could of course keep the current version of it if debian disapeared but i highly doubt they would have the manpower to keep the bulk of it up to date). The fact that there are very few distros that both cover a wide range of software and do thier own packaging work (fedora/redhat,debian,gentoo,not sure about mandrake) just serves to highlight how much work it is.
debian are not going to include the trademarked firefox logo, that would just be a HUGE violation of thier founding principles (and would also raise legal issues for customisations based on debian) nor are they going to let mozilla decide what patches they include. worst case they are forced to rename or remove firefox but i suspect mozilla will blink first.
"RAW" (which btw doesn't describe any one format but a multitude of vendor specific formats) is used to get the pics out of a digicam in a form that loses as little information from the CCD as possible and doesn't introduce redundancy (because of the way color CCDs work converting thier output to a traditional raster format is basically gauranteed to introduce redundancy)
but while "RAW" is the best format for storing digicam originals it is next to useless for anything else. So anyone who edits images or obtains them from sources other than digicams is going to need something else. JPEG is lossy, PNG has a poor choice of color depth and no exif support so that really just leaves TIFF.
png has 1 2 4 and 8 bit paletted modes as well as 1 2 4 8 and 16 bit greyscale and 24 and 48 bit RGB.
image editor support can be poor though, one good example of this is windows paint (which isn't very good for actually editing images but is commonly used for things like dumping screenshots to files)
i took a screenshot of an empty gaim conversation window, pasting it into paint (the standard windows app for very basic image manipulation and getting clipboard images into files) and saving as a png and then a gif and then a png again yeiled.
first png 15,429 bytes truecolor gif 14,726 bytes noticable dithering, 256 color. second png 9,994 bytes looks same as gif
already the color reduced png produced by saving as a gif first is winning against the gif and the truecolor png is only slightly bigger than the 256 color gif.
after pngout the first png is down to 10,784 bytes and the second to 6,170 bytes.
i'd really like to know what software you are using that produces pngs larger than gifs of the same color depth.
I'd really like to know what tools you are using to end up with a png 4 times the size of the gif.
It clearly allows local palletes for the images (frames in animated gif speak) rendered to the logical screen.
and section "8. The Decoder." clearly states that a decoder should render without delays other than those specified (unfortunately most browsers seem to ignore this part of the spec and insert a minimum delay).
so loophole or not such images ARE compliant with the spec, thier existance is more a curiosity than a practical feature though.
It's already there: it's called MNG. It's the animated version of PNG. The only drawback is that it isn't widely adopted, but that's not the FOSS community's fault. We can't force people to adopt stuff, we can only put it out there. thats true but making it a seperate format without first frame backwards compatibility, handled by a seperate library and far more complex than animated gif they stronly discouraged its implementation making animated gif still the standard animation format.
i still don't understand why those trying to make a gif replacement didn't include animation support in the main spec from the start. It seems very odd to me to set out to replace a non-free format with a free one and yet leave out a major feature.
BMP is still still the default format used by windows paint so not going away any time soon.
PNG has nothing between 24bpp (normal desktop truecolor) and 48bpp (insane overkill) and doesn't support EXIF (though it does have its own text chunk system which could be used for the same functionality) so its not a suitable TIFF replacement by far.
Slashdot Mirror
jpeg is worse than gif regarding generation loss in a couple of ways
1: normal image editing software can dependablly be used to perform operations on gif losslessly. with jpeg it can't (both because of the rounding issues and because the transform matrix from the original image isn't generally stored and reused afaict).
2: on gif you can freely rotate crop and perform pixel level moves without loss, on jpeg you are limited to operations that preserve block boundries (rotation and flipping of blocks can be done losslessly but this can only be used to rotate or flip an entire image if the image size is a multiple of the block size).
btw, can you provide evidence that jpeg->8 bit rgb->jpeg is always possible without introducing any rounding errors if the transform matrix from the decode is used in the encode.
Couldn't agree less. The UK has adopted the metric system in pretty much all things, the only notable exception being the reference to miles rather than kilometres in signage.
it did EVENTUALLY but really only because the EU forced it down thier throats.
don't confuse the tactical (short term during battle) with the strategic (long term)
strategically the milatary especially during wartime NEEDS civilians to serve it in order to keep functioning. and civilian industry is often very reliant on computers some of them MS based and (indirectly) connected to public networks.
Also, let's not forget, that PNG is in some respects is inferior to GIF. Transparency isn't supported by all displaying apps. (Sure that's largely due to Microsoft, but let's be adults here. You can't demand that users switch their apps for your whims. That's arrogance, which sadly is all too prevalent in the FOSS Community.)
gif like (paletized with a single transparent color) pngs are supported just fine and are generally smaller than the equivilent gif.
And still, LZW compressed GIFs still needed to be supported by everything because there were, and still are, a significant number of them deployed
iirc while unisys claims thier patent covers decompression those who have analysed it don't belive this would stand up in court whereas compression would.
Also, PNG doesn't support animations, which along is the other reason people use GIFs. (Yes, MNG exists, but it's supported by no one, so it's not an option.)
unfortunately true but animations are only a very small proportion of web images.
In the end GIF stuck around because of a combination of legacy support, and the fact that most people simply didn't care.
yeah, the fact that most people simply didn't care (just like most people don't care about the risk of being sued for sharing pirate content online until they are sent a personal threat by the ??AA) is almost certainly the main reason.
btw i stopped using gif because the version of paint shop pro i use (a magazine cover CD special release) doesn't support gif due to the patent issue.
No. The FUD spread by the patent the patent cost money and time. Total people sued by Unisys over the GIF patent? Zero. That is nothing.
does it really make much difference whether you are actually sued or if you (or your legal department) just think the threat (and unisys certainly sent out threats) is credible enough that you'd better pay up? Things only go to court if one player is an asshole or if there is true dispute, if you know you are going to lose why go to court in the first place?
Conversely, if you look at the successful FOSS projects, you see that the common thread among them is that they aren't clones, they were first-movers. Apache, Perl, Python, Firefox, the Linux kernel (Yeah yeah, BSD386, but it didn't survive). I don't believe this is a coincidence. I believe it's simple economics (in the decision theoretical sense, not the monetary sense). When the niche is already filled, only those motivated by politics will switch away from a stable feature-rich app to a buggy feature-poor app. However when the niche is empty, people will come to you, and you can define the niche.
btw linux (clone of minix) apache (a collection of patches to a university project) and firefox (essentially a stripped mozilla) were hardly first movers. and the linux kernel would have had hardly any success at all without the vast array of underapricated clone software built on top of it.
most software development is minor improvements over existing ideas, when the original is barred from you then you have to clone it before you can get up to the inovation stage and the FOSS community doesn't exactly have much option but to avoid anything which has patent threats hanging over it.
False. GIF is lossy, as it is limited to 255 colors per frame. Conversion from a 24-bit image to GIF involves an operation called dithering, which adds noise to the image. In fact, the amount of loss in GIF (called "palette size") can be dialed up and down just as easily as JPEG's quantization scale factor.
by the same reasoning you could say every image format is lossy as it can't precisely represent the raw output from your cameras CCD.
yes an image beyond a formats limitations will incur loss during conversion but that doesn't imo make the format lossy in the same way that say JPEG is (otherwise you'd have to consider every image format lossy).
Sure, I could find some variant of PNG that would work, but could I be sure that it would work on every browser?
gif like (8 bit or less pallette based with a single transparent color) work just fine accross every major browser back quite a number of versions.
if you are trying to match colors in the png to other colors on the page you should write the png WITHOUT color correction information to make it work right in most browsers (if your editor insists on writing this info use something like pngout to strip it).
Some programs (e.g. photoshop) also have notoriously bad png output support and third party optimisation software (of which i belive pngout is the best arround) is always worthwhile if filesize bothers you.
you might wan't to take a look at png tips for cartoonists which covers a lot of the issues with using png for this type of artwork.
1: it means free software can support gif writing properly (e.g. not using hacks that avoid the patent but also make the files bigger than plain bitmaps) without legal issues. This is a biggie for packages like gimp and imagemagick.
2: likewise for small commercial vendors who unisys fucked arround until they decided that gif support was more trouble than its worth.
3: unisys has in the past gone after some website owners forcing them to either somehow prove thier gifs were produced with licensed software (basically impossible) or pay an anual license fee themselves. It means theese websites are now safe.
Does this mean that Stallman has contemplated a future for the software industry in which patents are gone but (less restrictive) copyrights remain?
software patents are FAR worse than software copyrights. While (i belive i can't seem to find the source right now) that stallman is anti copyright like most others in the FOSS community he sees patents as the far larger and more immediate threat that they are.
with copyrighted software i can use it (looking at the code if availible is more risky because i may inadvertantly remember and copy it or be accused of doing so but i can always pay someone else to look at it). then based on my experiances using it safely create a clone.
with patented software (at least under the US system) if i implement the methods in the patent (which may be nessacery for decoding a popular format) then there is no choice but to license from the patent holder at whatever price they set. The general lack of competance of the patent office (granting many obvious, already implemented elsewhere or even duplicate patents) doesn't help either (fighting a patent in court is outside most OSS projects financial means}.
iirc with oem windows there is NO support included at all, you are just told to contact the manufacturer of the PC (admittedly the OEM may get some support but i suspect its much less hours per copy than retail gets)
The ones that need support, or want to support the company for whatever reason, are the ones that pay.
and all but the tiniest buisnesses in the western world who just can't risk it with the BSA rent a cops arround.
Sorry the support offered for free for OSS products is pretty hopeless for businesses. If a company has lost 100's of hours of work due a bug or problem, they have the option of complaining the makers of commercial software.
and expect say microsoft to take them seriously, unless they have bought a seperate support contract or are a very large customer i doubt ms will.
Our startup honestly wanted to use OSS products. We do not want to spend time for any OSS bug fixing so our main requirement was -official support for all OSS products-.
fine but you should understand that support is extra with most propietry software too.
We thought were prepared to pay the price for OSS products, but then we got a price sticker shock. Now behold: QT is $3300 per seat.
yeah trolltech (strangely fitting name don't you think) have set themselves up in a neat position to rape commercial software developers for linux, use a freeer toolkit like gtk. IIRC that $3300 does include the distribution though (unlike with MS where you will have to pay for a copy for every device you sell).
We have dropped the development and rewrote everything to C# (MSVS 2005 is ~$700).
and what if any support do you actually get at that price?
Embedded Linux from a reputable RT vendor is $25,000 per 5 seats per year.
A Cygwin commercial license will cost tens of thousands of dollars and is only available for large shops.
if you are shipping software based on cygwin you are a f*cking idiot anyway. Cygwin is barely tolerable in the controlled environment of your own boxes, once its on machines you don't control expect crashes caused by different apps shipping different versions of cygwin1.dll which don't play nice when loaded at the same time.
ultimately with any software if you wan't good support you will have to pay through the nose for it. The software itself (whether free or propietry) tends to be dirt cheap in comparison.
I remember US Robotics X2 modems could be upgraded to V.90 with a firmware update. Most if not all KFlex users also had the same option from their manufacture.
indeed, i belive V90 was specifically designed to be implementable on both X2 and K56Flex hardware (presumablly this is what the grandparent meant by "merged").
or in the case of theese asus devices at asus's risk
imo the fact that asus have made this offer means that they are pretty confident that thier kit will be firmware upgradeable to the final standard (and they almost certainly have insider information from sitting on the comitees).
1: an unhacked copy of OSX will not install on non-apple hardware.
2: i belive some versions of OSX shipped with machines won't install on older machines (so you can't buy a new mac and use its install CD to upgrade older macs)
i don't belive they do anything to stop you buying one upgrade and using it on multiple macs.
please stop spreading this myth, a gif like (paletized with a single transparent color) transparent png will display just fine in IE and will be smaller than the equivilent gif.
yes IE has problems with some transparent pngs but not with gif like ones.
the patent on lzw (which affected gif and some but not all variants of tiff) is gone, some tiff variants almost certainly suffer from other patent issues though.
the REAL problem with tiff is there are so many variants. full tiff support requires an insanely huge library (that will probablly have patent issues somewhere) and partial tiff support is a recipie for confusion.
afaict skype control the master identity servers so they can mitm any connection they wan't. and almost certainly the cops could force them to do so (and if they can't legally force it they could always bribe them instead).
the only secure system is one where you give people your public key and they use that to authenticate you, then assuming whatever means you use to give them the public key is not tampered with and the public key algorithm isn't cracked you are safe.
but public and private keys are generally too big for most people to memorize/easilly handle and with a system as described above you can't easilly change the private key so most systems (including skype) rely on a master server or root certificate to identify that "person holding the private key to public key "8932075623478901254734745678457845304" really is "joeblogs999". if that master server or root certificate cannot be trusted then neither can the system as a whole.
The fact that I know that every security patch will hit my machine fast, usually within a day, is why I use Debian.
i beg to differ(the version of gaim in sarge-security is still the same version that was current when that bug was reported)
The official logos have been a documented issue for ages and indeed are not included in a firefox build unless explicitly turned on. Whats new is they don't seem to wan't use of the firefox name either (in which case there of course comes the question of why does the public source and the default binaries built from it use that name).
i see two possibilities for why this has happened.
1: they think they have the ability to push debian into a position of violating its core principles and shipping a mozilla approved firefox.
2: they have got a new legal team who are paraniod about protecting the trademark on the name.
either way i don't see how the likely outcome of this (debian shipping firefox but not calling it firefox) is going to be good for either mozilla or debian.
debian is (and always has been) rather slow on the stable releases and lacking a bit in newbie friendlyness.
but make no mistake, ubuntu (everyones favorite distro recently it seems) relies on debian to do most of thier work for them. From debian ubuntu gets a huge library of software that is kept up to date for them (yes they could of course keep the current version of it if debian disapeared but i highly doubt they would have the manpower to keep the bulk of it up to date). The fact that there are very few distros that both cover a wide range of software and do thier own packaging work (fedora/redhat,debian,gentoo,not sure about mandrake) just serves to highlight how much work it is.
debian are not going to include the trademarked firefox logo, that would just be a HUGE violation of thier founding principles (and would also raise legal issues for customisations based on debian) nor are they going to let mozilla decide what patches they include. worst case they are forced to rename or remove firefox but i suspect mozilla will blink first.
png forces you to choose between consumer truecolor and insane overkill with nothing in between.
"RAW" (which btw doesn't describe any one format but a multitude of vendor specific formats) is used to get the pics out of a digicam in a form that loses as little information from the CCD as possible and doesn't introduce redundancy (because of the way color CCDs work converting thier output to a traditional raster format is basically gauranteed to introduce redundancy)
but while "RAW" is the best format for storing digicam originals it is next to useless for anything else. So anyone who edits images or obtains them from sources other than digicams is going to need something else. JPEG is lossy, PNG has a poor choice of color depth and no exif support so that really just leaves TIFF.
png has 1 2 4 and 8 bit paletted modes as well as 1 2 4 8 and 16 bit greyscale and 24 and 48 bit RGB.
image editor support can be poor though, one good example of this is windows paint (which isn't very good for actually editing images but is commonly used for things like dumping screenshots to files)
i took a screenshot of an empty gaim conversation window, pasting it into paint (the standard windows app for very basic image manipulation and getting clipboard images into files) and saving as a png and then a gif and then a png again yeiled.
first png 15,429 bytes truecolor
gif 14,726 bytes noticable dithering, 256 color.
second png 9,994 bytes looks same as gif
already the color reduced png produced by saving as a gif first is winning against the gif and the truecolor png is only slightly bigger than the 256 color gif.
after pngout the first png is down to 10,784 bytes and the second to 6,170 bytes.
i'd really like to know what software you are using that produces pngs larger than gifs of the same color depth.
I'd really like to know what tools you are using to end up with a png 4 times the size of the gif.
see the section "11. About Color Tables." in http://www.w3.org/Graphics/GIF/spec-gif89a.txt
It clearly allows local palletes for the images (frames in animated gif speak) rendered to the logical screen.
and section "8. The Decoder." clearly states that a decoder should render without delays other than those specified (unfortunately most browsers seem to ignore this part of the spec and insert a minimum delay).
so loophole or not such images ARE compliant with the spec, thier existance is more a curiosity than a practical feature though.
It's already there: it's called MNG. It's the animated version of PNG. The only drawback is that it isn't widely adopted, but that's not the FOSS community's fault. We can't force people to adopt stuff, we can only put it out there.
thats true but making it a seperate format without first frame backwards compatibility, handled by a seperate library and far more complex than animated gif they stronly discouraged its implementation making animated gif still the standard animation format.
i still don't understand why those trying to make a gif replacement didn't include animation support in the main spec from the start. It seems very odd to me to set out to replace a non-free format with a free one and yet leave out a major feature.
BMP is still still the default format used by windows paint so not going away any time soon.
PNG has nothing between 24bpp (normal desktop truecolor) and 48bpp (insane overkill) and doesn't support EXIF (though it does have its own text chunk system which could be used for the same functionality) so its not a suitable TIFF replacement by far.