Slashdot Mirror
Can A New Open Photo File Format Replace JPEGs? (cnet.com)
Got lossless compression? An anonymous reader quotes CNET:
Google, Mozilla and others in a group called the Alliance for Open Media are working on a rival photo technology. In testing so far, the images are 15 percent smaller than Apple's HEIC photo format, said Tim Terriberry, a Mozilla principal research engineer working on the project. But smaller sizes are just the beginning... it's got a strong list of allies, an affinity for web publishing and modern features that could make it the best contender yet for overcoming JPEG's 1990s-era shortcomings... JPEG isn't just limited by needlessly large file sizes. It's also weak when it comes to supporting a wider range of bright and dark tones, a broader spectrum of colors, and graphic elements like text and logos...
The HEIC's new rival is from the Alliance for Open Media, a group whose top priority is a video compression technology called AV1 that's free of patent licensing requirements. It's got heavy hitters on board, including top browser makers Google, Microsoft, Mozilla and the most recent new member, Apple -- though Apple's plans haven't been made public. And it's got major streaming-video companies, too: Netflix, Amazon, Hulu, Facebook, videoconferencing powerhouse Intel and Google's YouTube. And with the support of chip designers Intel, Nvidia and Arm, AV1 should get the hardware acceleration that's crucial to making video easy on our laptop and phone batteries.
To use Apple's HEIC, "makers of software, processors and phones must jump through a lot of hoops to license patents," which CNET predicts "means HEIC will have trouble succeeding on the web: patent barriers are antithetical to the web's open nature."
The HEIC's new rival is from the Alliance for Open Media, a group whose top priority is a video compression technology called AV1 that's free of patent licensing requirements. It's got heavy hitters on board, including top browser makers Google, Microsoft, Mozilla and the most recent new member, Apple -- though Apple's plans haven't been made public. And it's got major streaming-video companies, too: Netflix, Amazon, Hulu, Facebook, videoconferencing powerhouse Intel and Google's YouTube. And with the support of chip designers Intel, Nvidia and Arm, AV1 should get the hardware acceleration that's crucial to making video easy on our laptop and phone batteries.
To use Apple's HEIC, "makers of software, processors and phones must jump through a lot of hoops to license patents," which CNET predicts "means HEIC will have trouble succeeding on the web: patent barriers are antithetical to the web's open nature."
HEIF is not an Apple format. Apple only got involved with it years after it was standardized by MPEG in 2015.
Can someone tell us what's wrong or deficient with JPEG?
Quoting the post (not even the linked text):
"JPEG isn't just limited by needlessly large file sizes. It's also weak when it comes to supporting a wider range of bright and dark tones, a broader spectrum of colors, and graphic elements like text and logos..."
Then use PNG for those types of images.
It's too universal. It's one of the few things in the technology world that we all assumed would be present and in use, unchanged, 20 years from now. Someone has decided that is unacceptable. We need to disrupt the image file format world, shake it up, introduce some competition.
Why isnâ(TM)t JPEG 2000 supported on all browsers? Wavelet compression was invented in the 1980s and itâ(TM)s still not supported on all browsers.
24-bit PNGs are lossless, so if they incorporate photographic elements, the compression ratio is very bad.
An image format that balances lossy compression for photographic content but which can also efficiently handle graphic elements efficiently without major artifacts would be a good thing.
Graphic artist for 13 years here: FUCK JPEG.
Wasn't there another one mentioned here last year that seemed to fade away in developer?
So no, it'll never get replaced.
Why is somehow every single codec from the 90s supposed to be the be all end all on that issue?
We've replaced video codecs and containers multiple times already because more efficient algorithms are created.
Regular file compression has changed, and also audio codecs.
Yet every single time people complain that this is change for changes sake.
No. Maybe, just maybe, we've developed better compression algorithms in the last 30 years, and we are using those images in ways we never thought of back then.
Heic is an image container. A container is a data structure for the actual still images, in this case encoded as jpegs. As a container its encumbrance is something that will preclude it from use, but jpegs themselves will be replaced by the AV1 codec which should have a bitstream freeze soon. With backing from every major tech company but Apple it'll replace jpegs and video codecs as time goes on, with HDR support and higher storage to quality efficiency it'll be a win for everyone. As for which image container will become standard to go with it, that I don't know.
AV1 outperforms JPEG. AV1 delivers a smaller file size at the same quality or better quality at the same file size. Try this comparison of JPEG and AV1 at the same file size.
It is called portable network graphic (PNG). Do people not read?
I generated SHA-256 hashes of all my precious family photos going back 20+ years then deleted the originals. I figure if we ever wanted to look at them, I can just reverse hash 'em!
Trolling is a art,
Why not JPEG-LS? It's way faster and lower complexity than png while also yielding much higher compression ratios..
Can we do that instead?
Nobody deserves to get money without getting work in return. That's stealing! No matter if you call it "license fees", "interest", or "profit".
If it took so much work to create that idea, then we gladly pay what it was worth (with the choice of paying somebody else or refusing to, if you suck). But not a cent more!
If that nonsense was OK, then I could do *precisely* the same act, by making a copy of every bit of money my employer gives me, call it my "pecunial property", put it on the photocopier, buy shit with it until 70 years after my death, have the law protect my exclusivity on doing that, and call everyone who doesn't play along "a thug, murderer and rapist (on the high seas)" .
It'd be funny throwing those criminal imaginary property pieces of shit into a world where that would be normal, but their shit wouldn't, and see them struggle and coil.
If you want to annoy the recipient of your pictures, send them in JPEG 2000 !
Holding this up against JPEG is the wrong standard. Does it outperform PNGs?
All the comparison images on that site are saved as png, and the original is smaller than any of the others.
Why? Because the big IT corporations want to spend less time and money on lawyers and licensing. Who knows, maybe us mere creatives and users might get some improvements out of it too?
We should only need one image format, that automatically identifies the type of image that it is looking at. There is no format today that can take an image of, say, a newspaper page with both text and image on it. Different parts need different image compression techniques. Some lossless, some lossy.
That is something that would be useful. Particularly to non-technical people that do not understand the difference between JPeg and PNG. 15% better compression is a waste of time and certainly not worth confusing the standards space for.
The number of image formats documented for computer systems is pretty big playing field. Do we really need another one? Surely one of the already existing formats will suit the needs of every possible use case, already?
If compression is the goal, I have to question that goal.. is that really necessary? Our storage is getting bigger exponentially, our bandwidth between devices is growing just as fast, is better compression really needed?
I keep all of my files in PiedPiper format. They are tiny and the compression is lossless. It's all about middle out people! What's your D2F ratio?
(For those that don't get it: Google "Silicon Valley TV Show")
http://flif.info/
I guess I knew Spectre and Meltdown were bad, but to completely give up on their chip business because of it is craaaazy!
Of course they are saved as PNG. Saving the output losslessly as PNG allows the comparison to be made without AV1 or HEVC still image support.
Beyond the stuff in the summary for a lossy format the quality is still shit relative the competition and it's lossy to begin with. Looks more like shit than necessary and totally suck for text. What's the fucking reason to not replace it for new content?
I've always hated it if I have a graphic arts hat on, and yet when as an amateur photographer I end up rendering to jpeg in the end.
We need a new format, then we need to wait 10 years for The Store Formerly Known as Kinkos to buy "copiers" (printers) that support the format, and we'll be free of choosing between TIFF (80M file size limit and lack of compression support on most public printers makes this unrealistic), PDF with embedded JPEGs, or JPEG.
For internet use of course the thrash rate is much higher. The need seems less though, monitors already display as much bit depth as they are capable of.
A better format is a much bigger deal for print than web.
Something like a restaurant menu has to be printable on standard digital printers at a local print shop.
FLIF
It uses a lossy compression method. That makes it unsuitable in any application where you can expect some uses will be copies of copies (of copies), such as long term archiving, especially when periodic winnowing by a human curator is a part of the process. Or broad distribution where reformatted copies are permitted or even encouraged.
An example of the last are illustrated instructions for use written in English with the intent that others will translate the work into Arabic, Chinese, and even Australian, in environments where the original images are not available such that the reformatting necessary to fit non-letter-size paper has to be done from the images received.
Um, it might be that written Australian is not so much a non-English language, the way that spoken Australian is. :-)
Have you seen what jpeg compression does to photos with lots of dark areas? Very noticeable colour gradients. Happens even if the entire image is on the dark side. You should be able to compress an image without making it look that bad.
(But then I have the same problem with video compression, while most people seem quite happy watching colour gradients flicker wildly across the screen during darker scenes, so maybe I just need to learn to tolerate compression artifacts).
This isn't BBS era computing, we don't use dial-up anymore. get with the times gramps, PNG is perfectly fine.
Why not remove half the useless bloat in your software, os, drivers...
Oh right you don't actually care. You just want people to use YOUR format.
Sure we have. The question is: at what point do the benefits of replacing an almost universally accepted codec outweigh the problems of transitioning, which typically involves a morass of competing "solutions", potential to end up with a patent-encumbered result, and the inevitable attempts to shoe-horn anti-features like drm into whatever gets proposed.
You can't just throw down a new format and say "look, it's 0.1% better than the one we all use" and expect people to say "well gosh then I'd best rush off and rewrite everything and risk pissing off my existing customers right now". You have to make your case, and it had better be a good one.
JPEG was made for photography, therefore you have no business using it for graphic design. If you do use it, it tells me you are incompetent.
http://xooyoozoo.github.io/yol...
Sadly, I think it's got some stuff caught up in patents or something - the demo is very good.
I must admit, image wise, we haven't gone far in a long time. I'd like to see a very high compression lossless replacement myself and now that I (occassionally) do some light graphics work, JPG NEEDS to die, as soon as humanly possible, it's awful.
JPEG was made for photography, therefore you have no business using it for graphic design. If you do use it, it tells me you are incompetent.
But you, in all your intellectual superiority, don't have the reading comprehension to grasp the (very valid) point he just made about the real world that handles your output for things like printing. NOBODY CARES what file format you work in, archive in, render from or anything else. What matters is what you can transport to the end user or print shop in real life. Which, if you were competent yourself, you'd know.
Don't disappoint your bird dog. Go to the range.
av1 is slow, laggy and glitchy as fuck on low power devices.
that have no problem with any other encoding.
I was responding to the part about text and logos not photos.
Blah blah evil corporate corporations with sinister corporate evil corporateness. No, the reason is entirely technical and quality-related. Which you know, but are pretending is just some side issue so you can rant about Teh Evil Corporate Corporations.
Don't disappoint your bird dog. Go to the range.
Large file sizes...
At one point, there was a push to make Slash support Unicode better. That ended when vandals figured out how to use bidirectional override code points to spoof moderation scores and otherwise wreck Slashdot's layout. Others used the new code points to post obscene "ASCII art".* That led to a code point whitelist and a halt on further development of Unicode support in Slash.
Rehash, a fork of Slash maintained by SoylentNews PBC, fully supports UTF-8. I don't know exactly what it does with current and future directionality control characters.
* I mean ASCII art in the broad sense: use of characters from other blocks for their glyphs rather than their meaning, in the same way that ASCII art in the strict sense uses Basic Latin.
Given we're still using the FAT format for data interchange (camera cards etc) I shall not be holding my breath whilst waiting for JPEG to be replaced.
No, no. Your'e suppose to say FUCK ADOBE.
Faster! Faster! Faster would be better!
Microsoft and other proponents of VC-1 (SMPTE 421M) thought it too was royalty-free until the patent holders came out of the woodwork, pulled allegedly essential patents out of their waste chutes, and formed a patent pool in MPEG LA.
An example of the last are illustrated instructions for use written in English with the intent that others will translate the work into Arabic, Chinese, and even Australian...
Get a dog up ya! Bloody nong.
Lossless compatible photo format supporting layers. Dickhead.
I seem to remember Google's 2013 license from MPEG LA only covering VP8 and one successor, not what amounts to VP10. From the WebM project's announcement of the license, with my emphasis:
PNG does lossless far better. TIFF does need to die. Aside from the poor compression, the format itself is a horrific mess.
I write image processing applications, and part of that is writing loaders - I know TIFF is a mess. Just go read the PNG spec, then the TIFF spec. That'll convince you if you are able to grasp the specs.
I've fallen off your lawn, and I can't get up.
Do we really need lossy compression for still images any longer?
The network is way faster, local memory, storage, and graphics card resources are all way less expensive, and data lost from an image is data lost forever.
What we need is fiber everywhere, or something of equivalent speed.
I've fallen off your lawn, and I can't get up.
Sorry, but Adobe is the one that does the fucking in that relationship. The world's artists collectively decided that nothing Adobe does can ever drive them from Photoshop, so every month, bend over, pants down...
Irony: Agile development has too much intertia to be abandoned now.
Then use PNG for those types of images.
PNG has its own set of deficiencies, and not every image fits cleanly into "better as a JPEG" or "better as a PNG". Many images are way smaller using wavlet compression, but also need transparency, which JPEG doesn't support. But if you convert to PNG, you get a file size 10 or even a 100 times larger, because PNG is lossless.
It would be much better to have one universal standard that covers everyone's use cases, a single format that allows a mixture of wavelets, rasters, and vector graphics, as well as alpha transparency.
Nowadays I am using SVG for my websites since they are much smaller in size and scale perfectly for any screen size.
We just had Apple introduce their new format, which isn't bad but is yet another format to deal with
Yes, JPG isn't as efficient, but storage is super-cheap these days and I'd gladly take universal support over slightly better compression.
Nothing other than iStuff knows what to do with it. It's fine, but right now it's inconvenient, so JPG it is.
Does HEIC support transparent backgrounds? That's the only real shortcoming of JPG that has fostered the growth of PNG.
It's a perfect time for being wasted.
A perfect time to watch the stars.
- Burden Brothers, "Beautiful Night"
In two words, probably not. Where do millennial posters even find this stuff? I understand that you are exploring, but damn it's annoying to share the table with you. Where do the grownups go?
Apart from a half dozen esoteric specialty cameras, ALL the digital cameras made in the last 20 years save JPG. The better (more expensive) ones can also save to the specific manufacturer's own proprietary RAW format. Their proprietary bundleware or 3rd party software might take that RAW to TIFF for working on, but the final export is to JPG 99.9999997% of the time. That will not be easy to shift out and replace. One of the biggest barriers will be people wanting to share images in a commonly accessible format.
Betteridge Law == NO
https://en.wikipedia.org/wiki/...
~ People that think they are better than anyone else for any reason are the cause of all the strife in the world.
Got rid of JavaScript and replaced it with a better language didn't we? Oh wait....
Because nothing else even comes close, except for people whose artistic skills are limited to stamping "Happy Birthday Grandma!" on a family photo, unaligned, of course. From them, it's time to bring out the gimp.
The need for "lossy compression for photographic content" belongs to the time when 640 K was more RAM than anyone would ever need and a modem faster than 2400 baud was too fast for even the best speed reader. Back when the big argument was how to best partition that 40 MB hard disk: 32/8 to get the biggest possible C: drive? 20/20 so C: and D: were equal size? Or the common compromise: 30 and 10?
PNG does not compress as well as JPG, but the difference between a display latency of 32 microseconds and one of 53 microseconds is meaningless when eyeball latency is measured in milliseconds. The number of users whose internet connection is so slow that they would notice the difference between JPG and PNG is a vanishing quantity, and those stuck in the slowest of the slow lanes are undoubtedly experiencing other problems with unwanted pop-up ads and gawdawefool redundancies in posts from people who don't know how to unnecessary repetitions of old news.
I'm sure there is at least one good argument for replacing JPG with a brand new format; I'm certain that there is some edge case out there where PNG would not serve well. But I cannot think of it.
oh wait: Mars to Earth true color 4 K images of alien sand. Maybe, that would be a use case. Maybe.
Once upon a time... it took my Amiga 3000 (68030@25Mhz) around 30 seconds to decode a 1024x768 JPEG at 15 Bits color depth. Resizing to screen size (usually 800x600), smoothing out artifacts, put in some dithering took another 30 seconds... I dare not to imagine how long it would have taken on a much slower Amiga 500 (68000@7Mhz), not to speak of the infamous JPEG decoder for the CPC6128 which a friend used to convert single pictures OVER NIGHT.
This was no fun at all. I always hated JPEG for being horribly slow. Even my first 486 usually took ten seconds to decode the same picture and only after aquiring the "quick picture viewer 386" from Oliver Fromme. It took Irfanview on early P3 and Athlons to display JPEG without feeling bored...
I hope the new AV1 format will not be a slow poke.
"Life is short and in most cases it ends with death." Sir Sinclair
I've been a bit of a fool. Even written Aussie, when used 'tween themselves, is a foreign tongue. Cheers!
Our storage is getting bigger exponentially, our bandwidth between devices is growing just as fast, is better compression really needed?
only a really profound idiot would base important technology decisions on the ridiculous assertion that such trends will continue into the future
maybe you would also expect the mobile phone market to continue to expand sales exponentially even after every person already has one
you are the yeast in the glass of sugar water who thinks that they have a long life ahead of them
By outperforms, do you mean "looks better"? Then it's a maybe. AV1 images may look better due to having a smoothed out look, but if it's details you are after (science) then it fails. Select the hurricane photo "Claudette", look at all the tiny cloud details that AV1 gets rid of.
Maybe using JPEGs for the comparison images wasn't the best choice (at least I can't see much difference by the unnamed and HEIC).
I use WebP for archival and JPEG for distribution. Nothing is going to supplant JPEG because it's far too widespread and file sizes are a concern of the past with modern hard drive and SSD capacities.
Aside from vectors, WebP covers all of that. It's also a free and open format that is supported by all browsers and image related programs.
If you really want to get enraged, note that we're still using 8-1/2" x 11" sheets of paper regularly. At least in the US. Other places use paper of a slightly different arbitrary size.
It makes no sense at all!!!!!
MATLAB puts all Adobe products to shame.
Why did nobody use JPEG2000? It is notably better than JPEG. Both had patent issues, but that hasn't held-up open adoption of the JPEG format.
Well, will they?
It's hard to move billions of people in one direction.
my personal viewpoint is we should be moving to lossless video, audio, and pictures on the net. no reason for lossy compression at all.
the limit here is companies like Comcast, Verizon, AT&T, T-Mobile have crapped on us with bad investment in broadband infrastructure while the Republicans controlled the FCC and Congress to promote it, and prevent government regulation to force the upgrades to modern technology like fiber optics and holography.
Yes in fact companies / Republicans are behind many of the shortcomings in American infrastructure including lack of bullet trains, soring rent rates, bad schools, etc. Rather than buy the billionaires rounds of drinks, prostitutes, and yachts, we need to tax that money and force it to be spent on infrastructure in order to get tax breaks like Roosevelt and dwight Eisenhower were aiming for..
https://www.trumpsweapon.com/
Most folks on this thread seem to be under the impression that 8bit per channel sRGB color space is the end all of reproduction. Sadly it is not and that is one of the biggest things addressed by going to modern formats. The fact that loss factor vs compression ratio is lower is just a bonus.
Bandwidth-capped internet user here: FUCK PNG.
(except for logos and vector/gradient stuff. Then sometimes it compresses better than jpg anyway)
8 bit is getting limited for todays RAW images.
Keep the quality and color.
Domestic spying is now "Benign Information Gathering"
I was just getting used to png.
Apparently people don't surf on their phones.
I guess you're not aware that there are many many people living in rural areas that struggle to use the internet with the connection speeds that they get.
Lossless would be a waste since it wouldn't be humanly distinguishable from lossless.
Also lossless video? Are you fucking crazy or do you just not have any idea how much space and bandwidth that would need? Lossless 1080p true colour video at 60 FPS would take 375MB per second of video. That would be 2.7 terabytes for a 2 hour film. Now imagine what 4K or 8K video would require.
List the things that you regularly do with PhotoShop that you can't do with other photo manipulation programs.
I'm guessing you won't even respond because you aren't an artist and you have no idea what you are talking about.
Nothing at all
Yes, because the standard 32GB to 64GB in most phones is totally not enough to store some JPEGs on and neither are the dirt cheap 128GB micro SD cards. FYI my nearly 4 year old phone has 32GB + 256GB of storage.
So did you have a point or where you just trying to one up posture?
There's a few things wrong with JPEG that are covered above and below, but the truth of the matter is this: practically speaking, it's plenty good enough and plenty ubiquitous enough to secure it's place for the foreseeable future.
Text and logos are already handled better by PNG and SVG. But for photos? A high quality JPEG will look perfect to nearly everyone, and just about nobody cares if they can save 15% on a still image when they're slinging tons more data watching HD videos. Especially when that file size saving comes at the expense of guaranteed support on every platform, application, and device made in the modern computing age.
At least as an end-user format, JPEG won't be replaced anytime in the near future regardless of it's fairly minor deficiencies.
Compression artifacts. Take a small (800x600) JPEG photo, and try scaling it up to say 2400x1028 and see what happens. It will "look" blurry and print even worse. I thought SVG (scaled vector graphics) or something similar was suppose to offer a solution to that "jaggie" problem with compression artifacts.
We are in an age of cheap bandwidth and storage. Why are we even bothering with lossy formats any longer? It's not like the dialup era where it took 10 minutes to load a jpg and you had gotten off by the time the image was done loading.
Anything I put on the web these days is PNG
PNG does not compress as well as JPG, but the difference between a display latency of 32 microseconds and one of 53 microseconds is meaningless when eyeball latency is measured in milliseconds
Microseconds? You're off by a factor of 1000.
I took a 1024x640 TIFF photograph I had laying around and converted it to a 50KB jpg with moderate compression and a 500KB PNG and uploaded them to a well connected web server (10Gbit interface to the internet)
Then loaded them both in my home computer's browser. The jpg file took 86msec (that's milliseconds, not microseconds) to load, and the PNG file took 796msec (average of 5 tries). My home internet connection is 100mbit, pretty decent by American standards, if not by world standards.
I don't know about you, but I can definitely notice the difference between 800 msec and 80 msec.
I think they're just trying to soften us up for this abomination:
http://www.nasdaq.com/article/...
You are welcome on my lawn.
Internet explorer had to be Firefoxed and Chromed before it supported transparent pngs and SVG. Now Microsoft abandoned it for Edge we will have a large amount of users who won’t suppport the new image format and will be forced to support old image formats in enterprises and people who think the internet is the blue e.
Both jpg and AV1 do an awful job of reproducing the details of the twigs. Why should we bother switching to a format that isn't much better?
I get why JPEG isn't perfect but the big issue here is that for the vast and overwhelming use cases, JPEG is "good enough".
It's going to be hard to come up with an alternative that's not ridiculously patent encumbered.
LK
"Hi. This is my friend, Jack Shit, and you don't know him." - Lord Kano
What matters is what you can transport to the end user or print shop in real life.
You mean like PDF/X?
Ezekiel 23:20
One thing that's clear from the comparison is that x265 and AV1 both have a marked reduction in sharpness around the fine details. They appear blurred and smeared. In the JPEG version, more sharpness is retained compared to the x265 and AV1, although with more noise in the form of macro blocks. I'd still take the JPEG version over the x265 or AV1 version for whatever settings they happen to be using. That said, the amount of macro blocks showing up indicates very high compression. I've played around with JPEG compression in the past and for anything over quality 70, they wouldn't be noticeable like they are here.
Actual real graphic designer here, not a poser. I use JPG constantly.
PDF doesn't have to compress JPG. You can choose the level of compression, or you choose lossless compression, or you can choose no compression at all. You can even select different levels and types of compression for different types of images. Works great, you should try it.
HAHAhahaHAHAhahaHAHAhahaHAHAhahaHAHAhahaHAHAhahaHAHAhahaHAHAhahaHAHAhaha So funny oh wait not.
Other places use paper that makes sense. One sheet of A0 is 1 square meter with an aspect ratio of sqrt(2). Keep cutting it in half to get A1, A2, A3, and so on, all with the same aspect ratio. A4 is close to 8.5x11, and is used for the same things. Metric paper really shines when you want to shrink/enlarge to the next paper size, thanks to the common aspect ratio.
That is precisely why it is so attractive.
Read the Wikipedia page. Every choice is elegant and makes sense and is perfect. Mmmbecause we Germans came up with it. <Clarkson's German cousin>
The graphics artists hate JPEG (commented on Slashdot).
The scientific imagers hate JPEG (personal experience).
The photographers hate JPEG (commented on Slashdot).
The users don't know any better and couldn't care less as long as they can see what they want to.
So who the fuck decided to put JPEG on every thing that produces anything visual ?
Entia non sunt multiplicanda praeter necessitatem.
TIFF does need to die. Aside from the poor compression,
Poor compression? Chances are that's your TIFF tools, not the format, the reason being that:
the format itself is a horrific mess.
Yep. It's mostly a container for image formats. Compared to PNG, well it's bilevel compression is usually better if you use the old Fax encoding (really!). It can also compres things as JPEG.
In PNGs domain, it's worse, but not much. It supports LZW with horizontal pixel differencing (one of PNGs modes). It dowsn't support the area or top/left differencing and has no option for different filtering on different rows.
From my experience with PNG, the latter can make a noticable difference but it's quite rare.
I write image processing applications, and part of that is writing loaders - I know TIFF is a mess.
Me too! I only support the first TIFF image at the moment.
SJW n. One who posts facts.
Photoshop saves artists time. They are familiar with it, there is a huge amount of support from the community, thousands of plug-ins etc. Changing would reduce their productivity and the quality of their work.
So Adobe can get away with a lot, because like Windows it has to get really bad before it makes financial sense for most people to switch.
const int one = 65536; (Silvermoon, Texture.cs)
SJW, n: "Someone I don't like, and by the way I'm a fuckwit" - AC
The JPEG artifacts are so severe because they compare the quality of approximately the same filesize. The JPEG, x265 and AV1 are all 48K +/- 1K. If you want JPEG without horrible artifacts then the smallest I've managed with that sample image is 250K.
My own opinion from looking at the comparison: Under very high compression (from 3.75M source payload down to 49K) the AV1 image looks as good as the original (colours are good, no visible artifacts, doesn't appear blurry) but fine details is smoothed over. The JPEG has even less detail, but it adds a lot of very visible noise. It seems sharper to me, but the sharpness is from squares and sparkles that weren't in the original image. x265 looks decent overall, but comes across as rather blurry.
I'd like to see a comparison at higher quality levels. For the same filesize as a JPEG of acceptable quality, how would AV1 look? And how low can you go without losing fine detail?
In PNGs domain, it's worse, but not much. It supports LZW with horizontal pixel differencing (one of PNGs modes). It dowsn't support the area or top/left differencing and has no option for different filtering on different rows.
You didn't read the spec. PNG filter type 2, 3, and 4 all take into account a pixels from the previous row. Filters are selected per line. There is also interlaced encoding that sometimes helps with the compression.
And it's deflate/inflate compression, not LZW. LZW was patent protected when PNG was conceived.
https://xkcd.com/927/
try all you want. it took *twenty fucking years* for PNG to get what usage it does see today.. and that was with silly patent trolls on the loose for other formats or compression schemes.
another format simply won't "stick". period. /endofdebate.
optimize your fucking images. nobody wants to download a 20 megabyte logo when they load up your web page or app.
if it's a logo. it should be optimized.
if it's a spot graphic. it should be optimized.
if it's a heading or article picture. optimize it; it's small enough, it won't matter if you use jpg.
if you're sending truly high quality images. you're sending tiff or raw anyway. for anything and everything else, jpg is good enough... if it's not, the recipient can ask for something 'better'.
and newsflash: nobody fucking cares if you have 'truer hues' or some other bullshit in your app or web site images. 99% won't even care, or even notice, if you use a higher compression on jpg.. digital tv (especially cable and satellite) is full of compression artifacts and those same people don't see it there either.
24-bit PNGs are lossless, so if they incorporate photographic elements, the compression ratio is very bad.
No.
It is not uncommon to use 16-bit floats to store each component when you have a HDR source.
A picture taken on a bright day with a HDR camer would be cut to completely black and completely white areas if converted to 24-bit PNG without doing color/histogram compression on it before conversion.
24-bit PNG is lossless in the same way 8-bit WAV is lossless. It is only true if the source doesn't contain higher precision.
What is true is however that it doesn't introduce artifacts beyond the normal quantization error and if you apply some white noise with an amplitude of one destination LSB on the source before conversion it will remove most of the banding artifacts that typically occur when you cut off lower bits.
The problem is expecting the transport and storage standards to catch up after all other standards have.
Cameras capture wider colour ranges at better bit depths than they used to. Monitors have gone not only to wider colour gamuts but also to HDR. The average consumer now has access to equipment that outperforms the standard storage mechanism.
The problem here is that the user is driving the adoption. It's why JPEG2000 hasn't replaced JPEG, or why any of the other better than JPEG filetypes have done so either. In the mean time the content producers are the ones driving adoption of other storage standards. You buy a UHD bluray with Dolby Vision encoded movie, or stream an online video in 12bit HEVC. Why would you be happy visiting a website and seeing an image encoded in something from the 90s that results in banding and poor dynamic range when attempting to convey the full spectrum of what your equipment is capable of?
JPEG2000 also outperforms the 1990s era JPEG funny enough. The problem is leaving users up to driving adoption means it won't happen.
We need to learn from the video industry that the content producer needs to drive the adoption. We'll see wide spread use of 12bit HEVC long before we'll see JPEG phased out for this reason.
TIFF is a container format, not an image storage format. There's big difference in scope between PNG and TIFF. The defining a container is what makes TIFF so messy, but it is also what makes it far more useful than PNG. It has far more flexibility and PNG can't replace it in all cases. Classic point: You can't even save a 32bpp file in PNG as it tops out at 16bpp.
the content producer needs to drive the adoption
So it's a good thing that content producers like Amazon, Hulu, and Netflix are members of AOMedia. Facebook in particular would benefit from an AV1 still image format.
We'll see wide spread use of 12bit HEVC
Not for web video. VP9, for example, has twice the installed base of HEVC. The royalty-free licensing of AV1 (and HEVC's patent licensing mess) will encourage quick adoption of AV1. The fact that AV1 also out-performs HEVC is a nice bonus.
HEVC has no future on the web. That future belongs to AV1, and it's not too far away. Netflix wants to start encoding to AV1 in the second quarter of this year.
... then programmers need to get a clue and write an interpreter that becomes widespread AND also write a format that allows it to be read by legacy JPEG codecs that are now in use - i.e. at first jpegs will not reduce in file size because you're implementing backwards compatibility so people can make the transition to the new format.
This is news?
We'll make great pets
Do we really need lossy compression for still images any longer?
Absolutely yes. Not even a question. A lot of things you take for granted would literally not be possible or would be notably worse without lossy compression of images. Not to mention that using lossless formats when unnecessary is both wasteful and pointless for many purposes.
The network is way faster, local memory, storage, and graphics card resources are all way less expensive, and data lost from an image is data lost forever.
That doesn't mean storage space and computing power and bandwidth are infinite. And losing data from an image is not necessary a bad thing or worth being concerned about. For example it REALLY does not matter if I lose a bit of data on my pictures from my last vacation. The point is to store a memory, not to archive a well crafted image for all eternity. It simply doesn't matter if I lose a bit of information and the resources necessary to preserve it would cost more than the data is worth. 99.99999% of images taken with a smartphone would not benefit one bit from using lossless compression and in fact many would actually lose utility from doing so.
What we need is fiber everywhere, or something of equivalent speed.
You seem to be under the misapprehension that the main obstacle to using lossless image formats is the speed of the network connection. It's a factor in some cases but not the biggest one in most. Nor would it make it possible/practical to use lossless compression everywhere. You could make every network connection everywhere the fastest fiber connection you can imagine and it lossy compression of images would still be a useful thing.
Thus proving that there are never a shortage of idiots who will insist on using the wrong tool for the job at hand.
More likely it's a case of the proverbial "only tool they have is a hammer so all problems look like nails". Same reasons people use spreadsheets to do things better done by databases. They know how to use a spreadsheet and don't know how to use a database so they make do with what they know how to do even when it isn't the best solution.
Here is a default email reply to customers.
Mail 1) Sorry, we are unable to open your format
Mail 2) Sorry, we are unable to download from an external site
Mail 3) Sorry, we are unable to open your format
Mail 4) We are now able to see the image.
Don't fight for your country, if your country does not fight for you.
https://xkcd.com/927/
Sent as ripples into the electromagnetic field. No single photon has been harmed in the process.
Because newer isn't always better. In fact, it's almost never better when it comes to software.
Formats (and software) that have been around for a while are robust, well supported and have undergone heavy optimisations. Compare an early MP3 to one compressed with a recent version of LAME and you can immediately tell the difference.
PNG is from the 90s. So is LZMA. Vorbis is from 2000. MP4 is from 2001. FLAC is from 2001. MKV is from 2002. H.264 is from 2003. Know what they have in common? They work well and work on many computers and devices.
On so many levels...
First, JPEG is not large, it's very small which is why it's so popular. BMP files are large and TIF files are VERY large, but there's no compression in that format so it's pure, and the one to use for the best quality.
JPEG is good enough quality and it's small enough for everybody to use. Anybody can use it without some special license or software.
Whenever Apple or the like is pushing another format, it's because they think they can seize control of that market and require you to use their software or drivers or to have a license.
Sorry about their luck but I will stick with JPEG, or I will create my own format and release it into the public domain.
I don't know about you, but I can definitely notice the difference between 800 msec and 80 msec.
And it's going to be much, much worse than that with something like a gallery page with (say) 50 images on it.
Classic point: You can't even save a 32bpp file in PNG as it tops out at 16bpp.
No idea what you're talking about. PNG supports 32 bits per pixel, and I was working with one a few minutes ago.
This is web transport, not archive. Use tiff for HDR.
https://www.w3.org/TR/2003/REC-PNG-20031110/#11IHDR
Bit depth and sample depth are synonyms, and "sample" is a synonym for channel in this context. So not only does it support 32bpp (8-bit samples; RBGA), it also supports deep colour. 48-bit RGB and 64-bit RGBA. These would easily encompass any other potential meaning of 32bpp supposing that you didn't mean "with the alpha channel."
Last week I sent a pdf poster rendered from a vector graph to be printed at a graphic shop.
Their answer was: "send it in a standard format"
That's the real world.
Other places use paper that makes sense. One sheet of A0 is 1 square meter with an aspect ratio of sqrt(2). Keep cutting it in half to get A1, A2, A3, and so on, all with the same aspect ratio. A4 is close to 8.5x11, and is used for the same things. Metric paper really shines when you want to shrink/enlarge to the next paper size, thanks to the common aspect ratio.
I agree it is a decent system but there are two factors working against it in the US-
1. Familiarity with the 8.5x11 / 11x17 / etc. system, & sunk costs
2. It arguably is a barrier to foreign firms supplying the US market. Changing how the paper is cut, at a minimum, is a little more work for foreign firms accustomed to metric paper. They need to design double the amount of packaging, make adjustments to paper size (probably not that hard depending on the machine) and if their process is optimized to get 20 sheets of A4 from a width of paper roll, they may have slightly more waste. It may be a very light "tarriff" but it has the advantage of not being counted as one.
Even those who arrange and design shrubberies are under considerable economic stress at this period in history.
"I write image processing applications" - you must be familiar with Phatch, I presume?
I think the parent poster wanted to say bits per channel not bits per pixel. 32bits per pixel is 8 bits per channel (R G B) with an 8 bit alpha channel, and 16bpc would give you 48bpp (RGB) or 64bpp(RGBA). Tiff will do 32 bits per channel or 96 bits per pixel (no alpha channel) or 128bits per pixel with an alpha channel.
Time to offend someone
Having lived in a metric country for a while, that vaunted "being able to seamlessly convert units" ability comes in handy a lot less often than you'd think. It just doesn't come up. Plus, A4 paper is too thin and tall. When I use Letter, it's fat and has nice margins, better for printing text.
Shutting down free speech with violence isn't fighting fascism. It IS fascism!
Great. I'm sure they won't have an upgrade to my expensive Canon so it'll create the new format.
Real geeks don't look at pictures.
WebP is great. Better lossless compression than PNG. Better lossy compression quality than jpg for comparable size. Support for lossy compression with alpha (and even lossy alpha). Unfortunately, it is not supported by IE, Edge, Safari or Firefox (although the last two are "experimenting with supporting WebP images"). See caniuse.
https://caniuse.com/#search=we...
Sure, unless the place doing the printing doesn't want anything but JPGs or TIFs. Which is very common. You need to get out more.
Don't disappoint your bird dog. Go to the range.
the FLIF format seems to do a nice job in compressing LOSSLESSLY to sizes much smaller than PNG. Why not use that?
On top of that, it's free, open source and patent free.
You are confusing download speeds with presentation speeds. Which is understandable, I could have been more specific.
Without knowing much more about your internet connection than I really want to know, there is no way to tell whether you are doing an apples to oranges comparison. My connection through Verizon is very fast--- until I do something large and trigger Verizon's throttling. Frontier is much better in that and many other respects; I use it when I'm in range.
Throttling isn't the only thing that will arbitrarily slow some data movement. The load balancing pretty much assures that some packets of a large download will take roundabout routes, and some will be cached along the way. If I'm transferring a couple of megabytes between my place in Gresham and my friend in Vancouver, a crow-flying distance of 35 miles, some of the packets will be routed through several different intermediate servers, possibly including servers in San Francisco, Seattle, Boston,, etc. The route, and the delays encountered, have little to do with your connection or the file server's connection; they are mostly affected by the amount of traffic at various key transit points.
Work perfectly for me, with thousands of pics, spread on dozens of years and various cameras. Properly handle Raw files along with jpegs too...
I remember I started to consider Darktable seriously, years ago, when I discovered one of their noise filter was far better than any paying thing on the market...
If you don't want to leave Adobe just say this, don't say there are no alternatives.
Herve S.
by "supported by all browsers" you mean "supported only by chrome and chrome derivitives (opera, samsung)". No other browsers support it.
too bad :-D
Herve S.
I get why JPEG isn't perfect but the big issue here is that for the vast and overwhelming use cases, JPEG is "good enough".
It's going to be hard to come up with an alternative that's not ridiculously patent encumbered.
I don't think patents are the most serious issue. It's a handy whipping boy, but the legacy installed base is the big one.
There have been numerous "replacements" to JPEG, as well as other image formats. It took decades for the vastly suprior PNG to overtake GIF (even with patent issues in GIF), but even then, GIF is very much alive and well.
The same can be said about older video codecs (Many Blu-rays are still encoded using MPEG-II, in spite of the availability of better codecs). Nobody complains about the 1992-era Dolby AC-3 audio on DVD's, and 1993's MP3 shows no few of slowing down, in spite of considerably better codecs being developed & deployed in the past two decades.
The bottom line is that while higher quality is possible, the "better" level isn't enough for most people to even care.
Glare on your device's screen can make a bigger difference.
-- Sometimes you have to turn the lights off in order to see.
320x240 images are 25 KB, and that is way too huge. At (insert website here), we use the new, 15% smaller format. Note that in order to view the static content at said website, your browser must support, download, and execute 15 MB of Javascript and CSS.
stream an online video in 12bit HEVC.
So... who uses HEVC? Certainly none of the big players: YouTube, Netflix, Hulu, Vimeo, etc, etc, etc... none use HEVC. H.264 is "good enough" with a few that offer content in VP9.
You buy a UHD bluray with Dolby Vision encoded movie, or stream an online video in 12bit HEVC.
We still can't get color management right with the newest UHD HDR displays. I'd have to pay a few hundred bucks to get a technician to come out with a colorimeter and calibrate the display properly to get the benefit of 12 bit HEVC or Dolby Vision.
Hell, Wayland is supposed to be the new shiny to replace X11, and it can't do color managementat all.
Why would you be happy visiting a website and seeing an image encoded in something from the 90s that results in banding and poor dynamic range when attempting to convey the full spectrum of what your equipment is capable of
Have you seen a billboard recently? Or a magazine, newspaper, or any other printed media? Just because the image is better, doesn't necessarily mean people will notice and/or care.
There is more than enough variance from one production run to the next within the same make/model that specialized equipment and training is needed to actually achieve that quality on an individual display.
-- Sometimes you have to turn the lights off in order to see.
Nobody uses Edge or Safari. Firefox has been garbage for years and is always late to implement functionality, but you can use a JavaScript shim if you really want to support the idiots who use it.
Chromium, Chrome, Android Browser, Opera, Opera Mobile, Vivaldi and Pale Moon all fully support WebP. That's like 99% of all internet users.
Real geeks don't look at pictures.
That's all I look at. Downstairs in my parents basement. In the dark. With a large bottle of hand lotion at my side.
-- ducks --
Using ExFAT now. This is an improvement.
I guess you mean 32 bits per channel not 32 bpp - bits per pixel.
Yes, I see mixed results.
AV1 is clearly better on curves and edges, doesn't have that blocky and too enhanced look to the edges that JPEG gets. Does better on smooth areas, doesn't introduce the nasty artifacts JPEG does. AV1's gradients are much smoother. The Clovisfest image with air balloons against the sky shows all this very well.
But JPEG does better with rough texture. AV1 blurs the heck out of textures. Look at Cecret Lake. The greenery just below and to the right of the highest peek is all blurred into vague smoothness in the AV1 image, while JPEG preserves the original grainy look much better. Clear sky looks better in AV1, but clouds look better in JPEG. AV1 blurs out the fine filaments. Same with tree branches. AV1 makes branch tips and twigs disappear, while JPEG mangles and shifts twigs, but keeps them. Hair and fur is another problem for AV1, just look at the Tiger picture. Clearly better in JPEG.
Intellectual Property is a monopolistic, selfish, and defective concept. It is "tyranny over the mind of man"
Because support for JPEG-LS is practically non-existent.
Why? Because there are better algorithms out there. What a silly question. Please reply on a wax cylinder please.
We still can't get color management right with the newest UHD HDR displays. I'd have to pay a few hundred bucks to get a technician to come out with a colorimeter and calibrate the display properly to get the benefit of 12 bit HEVC or Dolby Vision.
You have to do nothing of the sort. The display primaries don't tend to drift over time and are done in the factory. The only reason for calibrating a display is if you need to use a non-default curve, or want to be absolutely positive for perfect colour neutrality with respect to brightness, something which is not even remotely relevant unless you're reading an X-ray chart.
Hell, Wayland is supposed to be the new shiny to replace X11, and it can't do color managementat all.
And? Linux has always sucked even more so than Windows at colour management. Linux has been retarded in this regard in both the dictionary definition and the insulting sense of the word. That doesn't mean we aren't able to handle colour management just fine, e.g. Firefox has no problem displaying accurate colours even on Wayland, because not every problem needs to be handled by an OS.
Have you seen a billboard recently? Or a magazine, newspaper, or any other printed media?
Yeah I routinely print my photos at 150ppi and black and white because newspapers are awesome. Have you seen my desk recently? No reason for asking, other than it's just about as relevant as you examples. When you start looking at your TV from 35m away through the rain then let's talk about comparing it to a billboard.
There is more than enough variance from one production run to the next within the same make/model that specialized equipment and training is needed to actually achieve that quality on an individual display.
That just simply isn't true. Any wide gamut monitor will ship from the factor with an delta-E measurement across the entire range of between 2-3. This is a range where humans can only tell the difference through careful side by side comparison of colours. My current monitor is now 12 years old, when I do a factory reset the spectrometer measure an average delta-e error of 3.2, most of which is due to the yellowing of the plastic across the front of the display. From the factory it averaged 1.3, after calibration it came in at 0.2 and ... looked exactly the same as it did before. The only reason I calibrate at all is for white balance matching.
This is even less important for technologies like OLED or QLED which have such well defined primaries determined by their chemistry that factory calibration for colour isn't needed at all, just a brightness characterisation curve.
"programmers" should get a clue?
It sounds an awful lot like you couldn't code your way out of a wet paper bag.
Apple never met a standard it didn't either want to own, sell a license to or kill outright. They're remarkably similar to MS in that regard.
Organization? You must be joking..
gif was very well entrenched before Compuserve or its heir(I don't remember who offhand) started to push back on the licensing issue.
If they had tried from the beginning, it wouldn't have been nearly as widespread.
Still, I agree. Patent encumberment is the least of the issues. For the vast majority of uses and users, JPEG is still "good enough" and there's no pressing need for an improvement.
LK
"Hi. This is my friend, Jack Shit, and you don't know him." - Lord Kano
Then you are doing it wrong. JPEG is the last resort, usually for those who have no idea or some legacy technical issue. Lossless is beauty.
Ooh, 13 whole years! You must be a real pro, kid.
Then why are you responding with text when you could have posted a video?
You are a fucking idiot.
That premise is incorrect. Even good 4G data connections cannot instantaneously upload/download modern sized 5-16 MB photos. Wifi won't help if you live in an area with congested signal. And wired broadband won't even help if you're connecting to slow web sites. And when your photo library is several GB, it takes significant time to transfer this data, regardless of the method. Even writing it to a local disk is slow enough to be annoying. Writing to USB flash drives can be frustrating. If you store those images without compression, you're looking at increasing the library size by ten (depending on the compression your cameras and image editors originally chose). It would be a huge problem.
A cat can't teach a dog to bark.
Why would I read the article? I come to slashdot for the girlz
Anyhow fixing JPEG is not the answer. It has way too many legal issues.
Anyone who complains about it's state likes to send faxes and write checks.
Are you just a troll or a fanboy, because you've obviously never looked at internet usage statistics. On desktop, Firefox, IE/Edge, and Safari make up 35% of usage. On mobile, Safari makes up 30% of usage. Those are significant numbers.
True enough.
However the situation remains that you've got to choose between high accuracy and high compression; you can never have both. But that's okay.
Certainly the highest possible accuracy is what is needed in the archival image. But that image is not in frequent use ---or at least it should not be. So how long it takes to stuff it into the archive is not an issue. If you are using good practices, you will be selecting any archival image from thumbnails, and only pulling the actual image once, the first time you've decided to use it for something. So leave it in its native format.
When you do pull an image out of the archive, put it in a working library of all the images, and only the images, that you will be using. These should be in a compressed lossless format ---PNG is the best choice for nearly everybody. But each of these images needs to be edited before it is converted to PNG. Crop out anything you will never need. Look at regions that you will always want to photoshop out, such as that photobombing idiot in the background, and fill them with a neutral gray or make them transparent (you'll fix the properly later). Identify the areas of least interest and deliberately reduce their quality by simplifying them: desaturate a little, or maybe a lot; use bokeh and blur techniques to hide or smooth unnecessary (and probably distracting) detail. Run the result through Photoshop or GIMP and clip both ends of the histogram.
Here's the thing: every image out of a camera contains much more information than anyone can perceive, and a hell of a lot more data than what good art needs. Contrary to popular belief, a good photo artist is not like a painter, adding stuff to what comes out of the camera. A good photo artist is like a sculptor who makes art by removing what stuff that gets in the way of the art.
Be gentle in this process as you prepare your working library; you want a product to use as raw material in making multiple pieces of art. You want to be able to crop heads from that group portrait to work up individual portraits. That dog catching the frisbee: sometimes you want to make the dog the star of the photo, but sometimes the dog is less important than the frisbee (maybe you are sending it to your Ultimate Frisbee team). These working images should be stored as PNGs, since they will be light enough that PNG compression will not be too slow to be a bother, and you want to be lossless in every step of your tool chain (except possibly the final step).
So an image out of your working library might be 10% to 50% of the weight of the archived image, but it has all the detail you might ever want to work with in a lossless format. It will still be too heavy for most purposes. So a copy of it goes back into GIMP or Photoshop and you use all the tools available to increase its artistic value (by removing stuff that gets in the way) which further reduces the weight. Depending on the size and purpose of the final image, you might use either PNG or JPG.
Images that will go on web pages never need to be heavy; often they will weigh 10% or less of the original archived image (but be much more impressive because of that weight loss).
So I don't see where a replacement for JPG/PNG fits in this picture. It would seem to be something designed for people who take snapshots and throw them into a shoe box; I don't see the value for an artist. Or anyone who wants to share stuff with others that is not crappy.
That workflow would work for someone that wants to put a lot of time and thought into image management. It wouldn't work so well for someone that wanted a compromise between versatility (including speed of backing up or transferring from phone to computer) and effort. I fall into the latter group. I have my own strategies to mitigate the inherent problems of lossy compression, but at no point do they involve dealing with 50 MB lossless images, except as saves during photo editing. The value just isn't there.
Maybe true artists don't need a better lossy format. The rest of us could use one.
A cat can't teach a dog to bark.