291 comments

[ 3.4 ms ] story [ 314 ms ] thread
Google Chrome is required. Nope.
I wish Google would lower search engine rankings for sites that require some kind of browser or plugin.
Obviously they don't want to hurt themselves.
Well, their crawler uses Chrome anyway, so it wouldn't notice, I guess.
Why the hard requirement on Google Chrome?
Because it uses PNaCl
AKA Google ActiveX.

This was a bad idea when Microsoft did it and it's still a horrible idea now.

Bad idea for whom? Developers, users, shareholders?
users. it's terribly hard to secure.
(comment deleted)
It's ActiveX, but completely secure and open source. Sounds like a good idea.

Regardless, it's pretty much already been killed. Too bad it's taken 8 years for a web standard (WebAssembly) to match it.

> completely secure

Yeah, there has never been a case of software breaking out of a sandbox or VM.

Fact is it allows running native x86 code downloaded from the internet, it's orders of magnitude more dangerous than e.g. Javascript running inside a sandboxed runtime.

Why do you think that? There's been way more bugs in V8 over the years than the NaCl core sandbox.
I'm not sure why you're trying to say ? Two wrongs make a right ?

For the record: I think that JS is a horrible idea too. What happened to the old rule of thumb that once you let someone else run their own code on your machine you should consider it compromised ?

> What happened to the old rule of thumb

People realized it's a stupid and impractical rule and ignored it.

>> People realized it's a stupid and impractical rule and ignored it.

If by "people" you mean developers, and by "stupid and impractical" you mean inconvenient.

Given that your rule of thumb is one that only developers would know or care about, yes...exactly.

Developers realized that their rule was stupid and impractical (not just inconvenient, but actively impractical: async interaction is faster and requires js), and so ignored it.

People stopped following that rule sometime around the time the computer was invented.
Depending on your requirements, it can be really easy to sandbox code. I could write a simple virtual machine in an hour or two that's more secure than your typical jpeg decoder. So it's really worth looking at exactly what kind of sandboxing and complexity is involved.
And now we have browsers with both sets of bugs instead of only one. Lucky us.
Was wondering why it seemed so much faster. Looks like they're serving gzipped dds (gpu compressed textures) for the tiles as well. Would be interesting to see a WebAssembly version of this.
Maybe they made it so it can run on ChromeOS also ?

It's not so different than downloading a Windows application that requires JVM or some .NET distributable, or a Linux package with additional dependencies, IMHO.

I respectfully disagree. [P]NaCl is bad for the web just like ActiveX was.

They have a Google Earth desktop client if you want a desktop application. As this is sitting on the open web it should use open standards.

I also disagree.

There is nothing even sorta objectionable about Google software on Google servers requesting you use a Google browser to access them because it uses Google-developed browser features that aren't even part of web standards, and wouldn't be for years even if that was the goal.

The alternative would be a moral stance that "no non-RFC stuff on port 80", and that makes zero sense to me. What, they're not allowed to innovate unless they run through the bureaucracy first?

I'd love it if you clarified what you're asking for here, because as stated, it seems alien to me.

What I want is for a web site to use standard web technologies so I can use it with my browser of choice.

I have no issue with Google putting things over the wire via HTTP but if it only works with a specific client it isn't, to me, a web site but an internet-based application akin to Microsoft ActiveX-based sites which were awful for the web.

The whole point of the web is that it was developed as an open platform and we had to fight with Microsoft to stop them turning it into an Internet Explorer only platform. I don't want the same thing to happen with Google/Chrome.

I want an open web where I don't have to use one particular browser for certain services.

I should also say I do think Google do a lot of good though. They do work with the web standards committee and they do a pretty good job of making their innovations open so I am not quite as worried as I was about Microsoft but, to me, it is better to stop it before it happens than fight to get it back.

In which case you're in luck - Native Client is completely open, both the source code and the executable format[1]. However, it's new and hasn't been submitted to the standardization bodies yet.

With that in mind, I have a real hard time getting worked up over Google putting a draft feature they're working on in their own browser. I suppose if Mozilla and Microsoft really want to implement a moving target, there's nothing stopping them from doing so.

If Chrome doesn't restrict it, then they start getting negative press about how this new thing they created doesn't work.

[1]: https://developer.chrome.com/native-client/faq

Very impressive, but too bad zooming in a zone isn't smooth: it keeps accelerating and slowing down on my computer, which feels dizziness-inducing.
Looks like the Google Earth Flight Simulator[0] didn't make the cut. Many features are missing from the original desktop app and it's not immediately clear what this offers over using "Earth" within Google Maps. The "auto-rotate around the landmark" feature is also a pain for people who are used to maps being north-facing by default.

[0] https://googlesystem.blogspot.co.uk/2007/08/google-earth-eas...

It seems this is the Google way. They kill one product (especially a desktop product) and it takes years till the new one is on every feature on the same level as the old one... look at Picasa vs. Google Photos. Even today Picasa can do many more things that Google Photos cannot do today. I'm sure the native/old Google Earth has a ton of stuff which is not ported to the web version.

Good they cannot convert Chrome itself as a web app :D

It makes some amount of sense because there's less to maintain and more cultural knowledge of the new codebase. Instead of saying "We're removing features from Picasa because nobody here knows how they work and everyone's scared to change things", they can say "We have a shiny new project here that will eventually be better than Picasa". It's a nice PR stunt.
I can tell you for a fact though that at least some of the senior Engineers who developed Picasa are still at Google.
None of which are on the photos team or have the time to rewrite it for the newer stacks.
I find it interesting that they recently introduced location sharing to Google Maps as a new feature... which it had back in 2009, until they got rid of it.
They had tons of handy features in Google Maps before it was redesigned using Material Design. We used the route elevation data often to plan hiking trips with our parents to determine if they would be able to make a trek. The map scale was absent from the first couple releases and when they reintroduced it, it would only appear on zoom and then vanish. It took 3 or 4 more releases for them to finally add an option buried in the settings to make it always on.

I'd like to say they did this to present a clean minimalist app but the recent incarnations of it are more cluttered than the pre Material Design version. Sure you can tap on the map to minimize the clutter and then tap again to make it go away completely but tapping also drops a pin if you hold it for a sixteenth of a second longer. And I hope your touchscreen isn't super sensitive because it'll detect that flutter as a double tap and then you're just zooming.

The ability to show photo hotspots on the map and look at the photos was really handy for planning trips to unfamiliar places (or even familiar ones, really), but AFAI can tell it's gone now. Seems it was part of some service they bought, integrated, then killed.
I believe you can get the photos by using the street view person. (That is, drag the street view person over the map and the photos show up as pins)
If you'd like to plan trips that have elevation data, I highly recommend komoot.com. It uses OpenStreetMap, and is a great tool for planning hiking or cycling trips.
I'll check it out, thanks. We were using Maps in a more adhoc manner when touring Scotland. It was great, now not so much.
For any serious trip abroad I use OSMand (but only on Android) ... you can use hiking routes (GPX), all offline (great if you have no phone signal). Also helped me more than once for parking or finding the next public toilet (haha). You can even using for skiing, biking or navigation on the water.

And regarding elevation it seems it is also there inside: https://github.com/osmandapp/Osmand/issues/1795

OSMand has the worst UX I have ever experienced. Hands down.

Personally I use Gaia GPS and Pocket Earth for hiking/nature.

When was the last time you used it? 2 years ago I would agree.
It was part of Google+ until very recently. So my guess is, that they just moved it.
also I noticed that at least in Switzerland, the imagery is significantly older than the one of the "Earth" layer in google maps. If I had to guess, I would say the images on Google Earth are from 2011, whereas the ones in Google Maps seem very current: last summer.

edit: Actually, the "earth" layer of Google Maps is identical to Google Earth. What's more up-to-date is the "satellite" layer which I only see offered in the mobile apps.

There was a point where you could choose from which time to view the data. Very useful feature. Might even still be in there somewhere if you find the right chicken to sacrifice to make it appear.
Google sure spends an incredible amount of money making their UIs worse (and usually increasing memory use by a large multiplier at the same time).
>Google sure spends an incredible amount of money making their UIs worse (and usually increasing memory use by a large multiplier at the same time).

Isn't everyone doing this these days? All the UIs on PC software have been getting worse since the mid-2000s.

> The "auto-rotate around the landmark" feature

I think of this as "Drone View". I keep waiting for the JDAM impact.

Doesn't do anything on my ubuntu / chromium system.

    cb=gapi.loaded_0:121 GET https://clients6.google.com/oauth2/v3/userinfo?key=[redacted] 401 ()
    cb=gapi.loaded_0:45 Uncaught [object Object](anonymous function) @ cb=gapi.loaded_0:45
    clients5.google.com/log?authuser=0:1 POST  https://clients5.google.com/log?authuser=0 401 ()
    POST https://clients5.google.com/log?authuser=0 401 ()
How great would it be if they did the same thing for Google Finance as well? It hasn't been touched in years.
Just like Google A, B, C and then some.
Oh man this is amazing! It's fast too. I can only imagine this getting better still with more research in deep learning.
In my opinion it is not that fast on initial loading on a 3Mbit/s line... I had the feeling that native Google Earth was faster (but maybe only because of caching).
Can you please tell me how deep learning is used in this app?
Sorry for any confusion. I don't mean to say deep learning is used in this app (though it might be). What I wanted to say was that with the advent of generative networks, generating 3D models on the fly from a few image data will hopefully keep getting better.
Got a bit of vertigo there
I have Chrome but my computer doesnt support WebGL apparently.
go to chrome://flags and enable overriding the gpu black list also there is an option to force Hardware acceleration.

I tried it and works fine.

"Google Chrome is required to run the new Google Earth. Please try this link in Chrome."

Get fucked.

Embrace, extend, and extinguish. It's ActiveX all over again.

PS: I have a 4 Ghz, quad core, i7-6700K CPU, 64GB RAM, 250Mbps internet and this is both poorly working to the point of being impossible to use, and bringing my computer to a halt.

Amen: "Aw snap! The new Google Earth isn't supported by your browser yet".

No. _You_ didn't support my browser. The web is not "Just Chrome" any more than the web was "Just Internet Explorer". Same game, different player.

Well, to be fair to them, they take fair amount of time but eventually they support multiple browsers. Same was the case with Google Photos editing not supported on Firefox when it was launched. Turned out Firefox had to up their game to support it and now it works in both.
Indeed, really getting echoes of MS back in the day when IE was king.
What graphics card? This uses your graphics card.
AMD Radeon R9 M395X 4096 MB.
Off-topic from the update, but sometime in the past 2 years the quality of their universal photogrammetry became jaw-dropping.
Yes, I was blown away when I first noticed this. And I was thinking how much this data, which is free for everybody, would be worth in case of a war...
It's not really free.

It's free in the sense, that you can browse around, but if you wanted to do some analysis on it, you would need access to the raw bulk data, and this is very expensive.

There is free aerial imagery data available (e.g. ESA's Sentinel-2), but it has lower resolution (10m/Pixel). If you want to do the 3D building thing, I estimate you need data with something like 1m/Pixel and from different perspectives. This data volume is rather big (Exabyte-scale) and it is non-trivial to host and access it, so currently only the big players have it.

San Francisco definitely has resolutions better than 1m/pixel it feels like (for buildings downtown) At least, you can clearly identify segments that would be < 1m across.
Yeah, there is high-resolution data available, but it is scattered around small-scale web portals mostly built by the government in this local area, and aggregating and polishing this data is a non-trivial huge effort (they come in different scales, projections and quality).

Furthermore most data is orthographic only, so you can built a nice 2D map out of it, but to do 3D I estimate you need images from specified angles (e.g. 4 views per place), and I think this data is still behind closed doors and quite expensive.

Absolutely! I recently tried the Google Earth VR on Vive where the photogrammetry really comes into play, and I was awed just like back when I was a kid and hooked up my computer the first time to a color CRT.

Except getting motion sick 5 minutes later.

The new Apple HQ building site looks impressive in 3D: https://earth.google.com/web/@37.33478572,-122.00939683,47.8...

They didn't close the circle yet, to let trucks enter the midsection.

What's more interesting to me is the cranes are 3d modeled, that was not expected
Yes, I noticed that too. How is this done? There has to be some machine learning involved, accumulating all the aerial imagery and street view data and somehow generate a 3D model. This can't be all LIDAR data, can it?
still an annoying aspect of Google Earth -

I can see a long range of mountains from my 2nd floor windows. I want to drop down to approximately the same position and elevation (from my 2nd floor window) and take in the same view on Google Earth. The intent is to identify mountain peaks that I can not determine in real life.

Google earth limits the angle one can tilt while the POV is close to the ground.

Have you tried this site?

http://www.udeuschle.selfhost.pro/panoramas/makepanoramas_en...

Edit: It generates panoramas with mountains and hills labelled .

thanks for the link...

Yes this is precisely the perspective I want to view on Google Map, with the added benefit of being able to double click on a distant peak and fly right to it, or perhaps also a feature to return to previous position so one can easily toggle between points of interests

The panoramas generated do have links to the relevant locations in Google Maps - though no "flying" :-)
There are some pretty good apps that will tell you the mountains that your phone is pointing at. PeakFinder, for example.
You can ctrl + drag/pan to look around your current point in space.
Try to get to experience Earth VR in the Vive. It's a fantastic thing. Very useful to get a better grasp of mountain ranges. It's my goto app to show non gamers people that want to try VR.
404. That’s an error.

The requested URL /static/9.0.31.6/balloon/balloon.html was not found on this server. That’s all we know.

Great...

Edit: after some F5-ing got it working. The 3D view is great! And I wonder why maps should be so slow. This feels much faster.

Edit2: I keep getting the 404 error every time. Looks like something is still wrong.

Bug: the demo only shows beautiful aspect of the Earth, not the polluted industrial areas, trash yards, plastic gyres, ship graveyards, oil spills, et cetera.
each time I look at google's 3d representations for things, I really wonder why they don't do manual touchups for certain places

I mean, if you are including the Eiffel Tower in the intro pages, could both make it look cleaner, and make it use less net resources

We need something like OpenStreetMap but with 3D content like this. And allow people to touch up the models.
Just why did they re-use the K8s logo on the landing page?
It is a bit goofy using another products logo but Kubernetes means Helmsman and that product is called Voyager so I see where they were going with using a ships wheel.
What is the difference to the 3D view in Google Maps?

edit: (honest question, dear downvoters...)

I would like an answer too - I genuinely don't understand what the difference is.
With Earth you can load KML or KMZ files, and you don't have directions.
Honestly, I think this is (or will soon be)the new "3D view in Google Maps". Seems to be an improved version of that, with a (very) few of the features from the desktop Google Earth added.
I don't​ know why Google Earth exists. The satellite mode in maps is very good, doesn't depend on PNaCl, and is part of Google Maps, where it makes sense.
There is no satellite view anymore, it's called "Earth". Not to be confused with Google Earth. And to disable the crappy 3D you have to go into the menu.

What's funny is that if you disable WebGL then you get a much more usable version of Maps IMO.

I love that 3d experience you call "crappy". To each their own, I suppose.
Javascript has big performance limits which don't allow Maps to do everything that Earth does. The 3D engine in Earth does a whole lot of magic to get the visual quality that it has. In Javascript, you have to trade off everything to minimize your function call overhead into WebGL and try to pre-generate as much static content as possible. Earth has a much quicker access path to GL, which allows it to render more and higher quality visuals.

Satellite mode without terrain can look fine in JS/WebGL. Once you start tilting the view and seeing terrain and buildings, the JS performance will be horrible pretty fast, and adaptive LOD streaming is hard due to the function call overhead.

I worked on the Google Earth desktop app for a few years, so this isn't idle speculation.

This new NaCL version seems to be nowhere near feature parity with the old desktop client, sadly. I hope it catches up, since I loved the historical imagery, for instance. The good news is that the desktop app still works if you want to see that.

Thanks for the reply. I played with this new version and with Maps's Earth mode for a bit today, and to me they feel identical 3d-eise (except that the new thing only works in chrome, and I like Maps's UI better).
I'm really impressed by the amount of detail in the 3D views (hold middle mouse button to look around).

Even the fair at the field near my house [0] can be viewed in 3D!. Coincidentally they are currently building that same fair today, so for a moment I thought the images were live...

[0] https://earth.google.com/web/@52.08511654,4.31805086,0.37952...

How are they building the 3D views? Mapping cars driving around with sensors? I was under the impression that the shapes were inferred from satellite images, but the moment I step outside the major cities, it's back to flat maps. Even the couple hand-modelled buildings previously present in maps seem to be missing here.
This is exactly what I want to know. There's several towns in the UK that have all the houses in really good 3D (mostly the coastal towns) - I struggle to believe that there's a team of humans making these...
There's a (yuuuge) public Lidar dataset for the UK:

https://environmentagency.blog.gov.uk/2015/09/18/laser-surve...

One of the motivations for the collection was flood analysis, which overlaps with your observation about coastal towns.

I'm not sure that's it's only LIDAR data. There are e.g. some cranes and scaffolding that also are 3D modeled. I assume it's some magical machine learning pipeline that accumulates the aerial and street view imagery, LIDAR where available, and generates a 3D model.
It's done by plane, though I'm not sure what sensors. More than just photography.
It is possible to use satellite imagery for photogrammetry [1], but Google is using imagery from airplanes, possibly combining it with ground level data from their street view cars.

[1] http://www.vricon.com/

SLAM techniques allow reconstruction of a scene from video+position data. As the aircraft takes video of the terrain, it is logging its position. Large-scale batch processing can recover the 3D positions for each image feature that will minimize some error metric.

Rural areas probably have lower-quality image data (or noisier aircraft position data).

How can you look around on a Mac trackpad?
Hold shift/ctrl then click+drag.
Maybe I'm just in an area where buildings are a tad more complicated than simple boxes, but around here the results vary between acceptable and catastrophic. As if buildings were haphazardly thrown together in all kinds of weird angles and shapes. I know there are limits to what technology can do, but overall I liked the combination of flat texturing and hand-modelled buildings better.

They also seem to have lots of trouble with pieces floating in the air, which imho could just be pruned (e.g. lots of floating train tracks here, presumably where there also has been a train at the point of geometry capture: https://earth.google.com/web/@48.51659545,9.06299404,316.491...)

The detail is great but the coverage seems limited. There's no 3D in Moscow, Kiev, Beijing or Singapore as far as I can tell, for example.
Embrace and Extend the web with Chrome only apps like this. Google are basically old Microsoft at this point.
I really liked having a separate application to open for Google Earth instead of a browser. I take it this won't be an option any more?
Chrome allows you to turn any website into it's own app, but saving a website such as Google Earth to your bookmarks then drag that to chrome://apps and right click on it and choose Open Full Screen and create short cut. So it will you will be able to open it like an app and find it in spotlight/alfred and you can still use chrome for websites separately.
An "app". Just a glorified browser page without chrome. You can achieve the same thing with a shortcut.
Somebody will probably create an Electron app that wraps the web app, just like with Google Play Music
> Embrace and Extend the web with Chrome only apps like this.

Depends on the reason for "Chrome only" - if they use (upcoming) standards not supported so far by other browsers I see no harm in it. Other browsers will catch up in time.

It uses Native Client, but seems that at some time development of NaCl had been stopped[1] so it's strange that new Google product still uses this technology. Maybe they'll port it to Webassembly later. Or maybe plans had changed and they're betting on NaCl, Talk plugin and other activexey stuff.

[1] https://bugs.chromium.org/p/chromium/issues/detail?id=239656...

It indeed doesn't work on Firefox (when switching user agents).

Doesn't seem like it's doing much in Chromium neither, just showing a static page: http://imgur.com/a/Q00OI

It looks like a very good shinny little toy to convince those who have not Chrome installed yet that Chrome is a good browser.

It surely is, but for me, it seems weird that a huge company like Google will give you these marvelous toys for free and they don't take anything in exchange.

What I mean is that putting people in a g-bubble like this is bad. I also wonder what the shinny 3 dimensional thing can add as a value to a standard map. In term of education for example: people don't even know how to read a map without a GPS and a map app now. I don't say that map apps are bad, but I don't know...

* They didn't know how to read maps before

* They still can't read maps and gps today, they just follow instructions

Seems to fail for me on Chrome on all my OSX machines. Colleague on windows/chrome it works fine..
chrome is free, fast to download, runs on most platforms, and in no way forcing you to use it exclusively. It's not like they're asking you to download an .exe with direct3d drivers. you think they need to support microsofts 30 browsers or shouldn't release at all?
It's not bitching, I just think as a professional I'd be required to launch products with requisite browser support and I don't have the resources of Google. The whole point of Web Standards and the Open Web have been made numerous times and if this was Microsoft requiring Internet Explorer in 2017 I'm sure they'd get pilloried.

I'm not seeing how this is different than an Intranet site requiring Internet Explorer or a video site requiring Flash. Unless you believe that Internet Explorer and Flash are "Bad" and Chrome is "Good"?

not a bad point, but I think web standards and open web apply to most content. if you're pushing the envelope in what a browser can do for a flagship product, that becomes exponentially hard to support.
To what, if any, extent is Chrome itself collecting user browsing data? That is part of Brave's business model -- there are many things about your behavior only the browser can know. Is Chrome doing the same?
Chrome-only and that's if there's no bug. I have the latest version of Chrome on Ubuntu but my GPU (GTX 960m, latest drivers) is blocked and the --ignore-gpu-blacklist flag doesn't work.
Google has already said other browser support is coming soon.

So I guess your little analogy doesn't work anymore.

>Get the new Google Earth now on the web in Chrome; on Android as it rolls out this week; and on iOS and other browsers in the near future. (Of course, you can still access and download Google Earth 7 for desktop.)

Mobile view using F12 dev-tools shows a not-so-well-done design. That's a first for Google but I guess it will be improved even on various devices without having to install/try the apps.

But it's a good first shot, keep going guys!

Nice, I hope someone makes a GTA 1 clone using the models and data :)