Google Guetzli is an open source JPEG encoder that creates 35% smaller files

The problem with JPEG 2000 is that it was never widely supported by web browsers. The advantage of Google's new Guetzli algorithm is that it produces standard JPEG files so it's compatible with any browser or software that can open JPEGs.

@ Samaistuin - LOL.
@Redzone - My understanding is that it was deployed, but it was the wrong answer to the question at the time it appeared - it took too much memory, too much processing power (for the average computer in 2000), for gains that were too subtle. Consumers waited for camera companies and programmers to adopt it, camera companies and programmers waited for consumers to adopt it. There are worries about patent trolling potential, many patents were "contributed" to the core codec royalty free, but there is no guarantee they'll stay that way.

I suspect there's going to be similar concerns and problems with Google's new codec, but at least they're positioned to get it into a consumer browser (Chrome).

Thanks for reply.
That's a shame as I remember that the compression was good as the preservation of details, avoiding the ugly artefacts the JPG produces sometimes.

[[know the more you open and close a jpg the worse it gets.
Dont loose the tiff.]]

1. Opening and closing a JPG does nothing to the file. Repeatedly saving a JPG file, depending on the application being used, can cause a loss of quality.

2. Loose rhymes with goose.

I tried to convert an image from TIFF to JPEG using Guetzli and it doesn't appear to be supported. But exporting to PNG and then converting to JPEG using Guetzli works fine.

Open JPEG as many times and nothing happens to quality.
Open JPEG to editor and save it as many times and nothing happens to quality.

Open > Save > Open > Save the JPEG many times and quality degrades.

When the image editor has the image in its memory, it doesn't get replaced by the exported JPEG version, unless the program is really built to reload the data from the exported file. And that is one of the stupidest thing you can do for a image editor/viewer as you already have the data in the program in the first place.

Tommi, no-one (who understands lossy compression) would ever consider saving JPG, then reloading it to edit further, save again, reload, etc. But then again, ANY image that is on your hard disk, you MIGHT one day want to open for further editing, and if your entire repository of images is stored in JPG format, then gradually, over time, they will lose quality if you want to make occasional tweaks.

And even within a single Photoshop session, remember there's a "Revert" (to last saved) option to use when a heap of edits have gone wrong. If you do occasional saving during a session (always a good idea), and then use "Revert", you will be reloading your most recently saved file - and if that is a JPG, then !!BANG!! - one more generation of lossiness and artifacts.

Stick to PNG, apart from disposable images for posting on Facegram or Instabook.

Brian

As long the image stays in the storage untouched it doesn't lose the quality.
As long the image stays in the memory for the editor, it doesn't lose the quality.

What the "Revert" really does is to flush all the data from memory except when the file was opened by reopening the file from the storage.
If you happen to save over the file (stupid thing) then it reloads it.
If you happen to save the file with another application (stupid thing) then it reloads it.

This is why the JPEG even if original, is always left alone as master and then re-save the edits to another.
Just like you can't go to save a raw file but you need to generate the copy that you use.

This is as well nice when you have a editors that allows you to save the history with the file, so you get a original + edited version + whole history. So you have a single file that has everything in it in lossless form.

PNG is great for master-copy from what to make multiple JPEG in different size.

Agreed, "The Squire", non-destructive editing is an option. So far, though, I have not got into LightRoom. I shoot RAW+JPEG on my Olympus OM-D E-M1, at the highest resolution and native ISO, relying on image stabilisation, or physical support (tripods, etc) for low light - and for all important images, work from the RAW files. I generate JPGs for one-off use, and PNGs at an appropriate resolution for any images I might want to tweak again quickly at some future point, without going back to the archived RAWs and starting over. Photoshop's own RAW import filter is great in remembering what tweaks I applied the previous time I opened the given RAW file anyway!

Brian

This is why the Adobe DNG is great idea.

You get a DNG file that contains a original raw, three image versions from it (can have, usually has just two, a full resolution JPEG and then thumbnail JPEG, but can have a TIFF) and all the metadata in the raw, but as well every metadata of the modifications to that raw file to produce that full size JPEG.

The file size is huge as it contains raw + full JPEG (can be set medium if just wanted for viewing purposes, it is up to person doing the saving) + maybe max megabyte for the editing metadata.

Now. You could take the files from the camera, convert them to DNG and embed the raw in DNG.

Now you store just this DNG file. And at any given point you can open it, and you see the edited JPEG file. Choose to do more editing and all the editing is done to the TIFF and then converted back to JPEG for output. And if you at any given time want to get a new raw conversion software a change, you can re-process the raw to new TIFF and JPEG).

@Biowizard, isnt PNG limited to 8bits per channel?

16bit TIFF is often used for sharing non-destructive, non-RAW images between systems for further processing. It's gonna be bigger than a PNG but leaves in a bit more colour depth which might be valuable in future edits...

Maybe you should have paid more attention to math classes?

The compression could be improved. Make it grayscale, put Google logo on it, everyone will adore it as the biggest IT innovation :-/

Most of web traffic is wasted by tons of JavaScript "frameworks" and such garbage.

?

found it but when i double click on the Files guetzli_windows_x86.exe or guetzli_windows_x86-64.exe nothing happens...

It's a command line tool, you can't just start it by double clicking on it

I totally agree, like every Google product I've ever used, it's written by kids with no instructions and hard as heck to figure out. Not everyone's a nerd programmer, you'd think if they wanted people to use this they would have made it at least somewhat obvious on how to do it. And as long as I'm on a Google rant, if they'd fix the stupid ads in AdSense, we webmasters wouldn't have to worry about trying to shave a few milliseconds by crushing jpegs or using the totally non-standard AMP HTML...

thank you i have no clue about command line tools :-)

It is pretty straight-forward as command-line tools go:
1. Press WINDOWS_KEY+R
2. Enter CMD in the new window and press ENTER
3. Drag and drop the .exe into the black window
4. Press space
5. Drag and drop the photo into the black window
6. Press space
7. Drag and drop it again and add a zero at the end of the file name
8. Press ENTER
9. Wait

Your command should look something like this:

"C:\guetzli_windows_x86-64.exe" C:\Users\you\image.jpg C:\Users\you\image0.jpg

The program seems to be very slow at the moment, so it may take ten or twenty minutes to complete depending on the image size. I would test it with something small first.

Schwäbisch is the dialect that makes "guacemole" sound really funny, right? ;-)

It's the new "native" Means cookie. Swabian is an abomination ;)

@brumd, Why not have ALL photos, videos, etc in portrait orientation? Much, if not most, web viewing is done on phones. So now the BBC has a vertical-video news feed http://www.newsshooter.com/2016/11/25/bbc-news-app-launches-vertical-video-is-the-joke-on-us/

The majority of my photos, including ads, were shot as portrait—it would really screw-up the composition to crop it as landscape ...ain't gonna happen.

Flash may indeed be awful, but how many times have we heard that it's on its way out, or that it's dead, or about to be deprecated...but yet it's still out there. Why do we think Flash has survived so many challengers?

Just try it on my Mac, Hard to find different vs JPEG Mini. The major different is it's 10-100 times slow.

Additional compression really benefits use cases where internet service speed are not the greatest (like mobile phones and tablets, 3G, 4G, LTE, etc.) or developing countries. It would potentially save people a little bit of money if they're on fixed data rate plans.

Reducing client web page load time & bandwidth costs are very handy also. Considering reducing web page load time by even 10% could earn/save a company thousands to millions of dollars.

It's not a simple math here. You need to estimate the cost of storage, the cost of bandwidth, the cost of extra processing in terms of hardware, power, cooling, cost of labor, maintenance, etc. And also the trends in all those areas.

If all associated storage costs go down faster than the volume of data then any optimization is a waste of time and money. If the opposite is true -- it's also a waste of time and money because you are screwed anyway and only delaying the moment the faulty business model fails.

I see they are near Bahnhof Enge, short walk from where I lived on SternenStrasse. Nice neighborhood.

@mick I explained why it is named as it is. That's relevant if the article itself is. The Zürich guys name all their compression stuff according to Swiss bakery products (Zopfli, Brotli). What Guetzli does mean exactly doesn't matter though so I left that out. A single word Google search would turn it up anyway.

@ Alfred: That's funny. We use "Brödli" for cookie and "Gützeli" for Bonbon exclusively. Thanks for the info. :-)

Yes. "Modern media" is not a single thing. Different technologies for different needs; MPEG sequence compression has an entirely different use case than static JPEG images. You can't study a frame in a running video. You can certainly study a static image, and this site is chock full of people who do exactly that, which makes my point pretty well, I'd say.

Personally, I use non-lossy PNG for static images meant for the web. Works fine. And when I go to reload the image for editing or composition, all of the pixels are exactly as they were when I last saved the image. With regard to image editing and processing, it's really not a good thing when each successive load/save cycle loses image information.

Further, RAW is just the natural extension of this idea: Save all the information you have about the image. Then, non-destructive editing is yet another extension of the idea: don't screw up the image.

JPEG is *literally* "now let's screw up the image."

Post size is too small for more

"With a little more coordination and a little less pride, they could work together and come up with something totally lossless."

A little less pride? How about just paying those people, then?

I'm sure that not all of those PhDs are holding out for tenure and avoiding doing anything else. I also doubt that all of those "laid-off over-40 computer programmers" are holding out for a comfortable pre-retirement position, either, or whatever it is that you're insinuating.

Indeed, quite a few of those people might like a challenge like this, but they shouldn't be expected to work for free.

If implemented properly, multiple re-saves of a jpeg file after non-global editing shouldn't reduce quality but in any case jpeg isn't an archival / editor format it is a publishing format. You should save in the native format of your image editor or if paranoid in a lossless editor neutral format like tiff (at the expense of losing features such as making and layers).

Sorry, I myself am both laid off and underemployed but my word was poorly chosen; what I was trying to say is that sometimes my own pride has gotten in the way of me diving into creative opportunities that were what I thought was too outside my field—but perhaps scientists and engineers are not burdened with such stiff necks as I.

Hope you find something new and rewarding to work with, MrBrightSide!

For some scientists, looking for things outside academia actually does become a priority, especially if they get disillusioned with the politics and the money is better elsewhere.

I think we'd all like to work on something meaningful and get paid for it. But I agree that sometimes we need to be shown those meaningful things because we otherwise wouldn't discover them ourselves.

But it still makes one uncomfortable that that much data about oneself is gathered and stored somewhere.

I wasn't referring to Google employees looking at my private pictures, but to the massive data collection Google is doing - for advertisement, right, but they might be as well used for other purposes. Think of e.g. predicting your health risk by your internet usage (totally possible) and selling it to insurance companies. They don't do this atm, but the sole amount of data gathered, analysed and merged is reason to be concerned. Company policies can and do change from time to time. BTW my comment I was actually *defending* them from accusations that they use Guetzli as a tracking tool.

You're naive if you don't think there's a Google Skunk Works somewhere with a room full of ultra-smart people crunching individual, identifiable personal data to figure out how to make even more money from you.
If it's not Google itself, it's some vaguely allied company in a country with lax privacy laws. Not to do so would be leaving cash on the table which is a mortal sin in contemporary capitalism.

Again, that sounds needlessly nefarious. Why would they care about you? This data is so easy to anonymize when doing projects on it, there is no need to expose any individual to needless risk. Their sucesss is more contingent on users trusting that others won't see their private data than some nonsensical skunk works project you can't even describe the purpose of or why anonymous data is not just as useful.
I can't stand ignorant conspiracy theories. If anyone is doing anything nefarious it is your elected representatives who can subpoena all of this data with very little effort and even force the companies to stay silent about it. Google does not care about individual users...

Tracking is not done via the JPEG algorithm, it's done from the web server, or injected javascripts into your browser. Compression algorithms do not have tracking capabilities.