The site is in German and a bit clunky, so follow along: Once the map loads, find a bit you like using the navigation icons (though be aware that you can’t zoom out more than you already are).

When you want to export your view as KML, click on the gray button with a downwards-facing arrow. Choose the default, dargestellter Kartenausschnitt, to export your entire view. (Select Abgrenzung in der Karte angeben to make a selection inside the map, or choose the last option to search for individual landmarks.)

Next, select KMZ as your export format. Voila:

Of course, getting Google to serve Berlin in 3D via the default 3D Buildings layer would be even cooler. As would be textures.

And while we’re on the topic of real estate, check out John Jason Fallow’s latest flight of fancy: In Arizona, property plot polygon heights as a function of their sale value, all linked to the timeline.

Here’s how to take the default railroad layer in Google Earth to the next level — by connecting individual stations to their timetables. Over on Google Earth Community, Chaver shows us how, using Switzerland as an example. Nice touch: Just like with Google Earth’s base layers, zoom in to get more stations.

Wouldn’t it be cool if, in some future version of KML, the API would let you select placemarks to send to your server to do calculations on? Then you’d be able to get timetables for specific itineraries. Find and select JFK and LHR airports, press go — and get back prices. Even better, use the timeline to find bargains! (Maybe this is something for LastMinute Labs?)

This network link again makes me yearn for more comprehensive support of HTML tags in Google Earth popup windows — I’m thinking IFRAME and EMBED, specifically. That way the timetables could be right in the popup window.

Remember the difficulty US National Weather Service employee Chuck Hodges from Tulsa had in getting approval for publishing his local weather data as KML? (Take one, two, three) That’s so 2005. Now the NWS has gone full bore with KML weather overlays of its radar imagery. These localized network links update every five minutes, contain the past 10 or so overlays, and are timeline enabled — what more do you need to get in touch with your inner weatherman? (Via Google Earth Community)

CNBC’s Maria Bartiromo interviews George Bush and asks him if he uses Google. Turns out he’s used Google Maps… or Earth… one of them, for sure:

I think the Google is going to have to give its products even more descriptive names. (Via ThinkProgress)

• Frank at Google Earth Blog has been on top of the latest iteration of the US election layers in Google Earth, and now also the Washington Post IT blog covers it, exploring it in detail. But why is this IT news to the Post (and to most others covering it)? It should be political news, primarily, aimed at non-techies who would never think of looking at Google Earth to find their electoral district and candidates.
• This article is so wrong on so many levels that we have to hope it’s because of a shoddy reporter, not a mendacious PR person. It begins “A local Canary [Islands] company is one of the first in the world to team up with Google Earth and create a property search engine, …” and it gets worse from there.

• Jabal al-Lughat is a linguist blog written by Lameen Souag, who has recently discovered Google Earth. He blogs how easy it is to creates collections of placemarks to create linguistic maps of places where obscure and minority languages are spoken. And indeed it is. Here’s one he’s made of where Tunisian Berber is spoken. Tip to Lameen: Use Tagzania‘s place tagging and commenting tools to create fully searchable linguistic indices. Or even just upload your KMZ file to Google Earth Community to have it show up automatically on other people’s Google Earths.
• Speaking of Tagzania, the team behind it has landed a commission to produce 3D models of Spain’s latest architectural wonders for a video installation at the Venice Architecture Biennale (!!). You can travel to Venice to see it (until Nov 19) or you can download the models for your computer now.
• Snowfall info is not just for skiing anymore — it’s an important leading indicator for drought. jfleck at inkstain links to the KML versions of data from the Natural Resources Conservation Service’s Snotel automated snowpack measurement stations in the western US.
• Geonames.org fixes its ontology after some feedback from the big names in web semantics.
• Freelancer turned Flickrer Rev Dan Catt sets the timeline straight as to who developed geotagging when.
• I’m a bit late to report it, but what a boost for GeoRSS: National Geographic gives it a big wet sloppy kiss on the lips. Way to go GeoRSS. (NG has a whole series of articles on the digital revolution in mapping — perhaps a bit late to the party?)
• If you want to see GéoPortail’s 2D imagery draped over a 3D Google Earth, Guilhem Vellut has created a Ruby application that will fetch the relevant tiles and present them as overlays in Google Earth 4, on demand. PC only at the moment. (Via Le Blogue d LFG)
• Newham High Street in London seems set to be coming to Google Earth sooner rather than later, courtesy of an enlightened city council and the Centre for Advanced Spatial Analysis (CASA). Comes with YouTube eyecandy.

Last week we saw the release of Paolo Mistrangelo’s Super Overlay Tiler, a free PC tool for creating region-based KML. Now we get Valery Hronusov’s Superoverlay beta 1.0.0, a application for PC that has the exact same aim. It costs $20, but it has more features.

How do they stack up? Both are beta; Paolo’s free tool is a one-size-fits all app for PC with a command line for a GUI, whereas Superoverlay beta is a stand-alone GUI app — you basically feed it a file and some coordinates, and you’re good to go. In addition, you get to choose how many levels of detail (LOD) you’d like, what file format the images should be, and whether the result should be compressed into a single KMZ file.

Valery gave me a copy of Superoverlay beta, and I have now thrown the same 100MB TIFF file of South Georgia at it. For comparison purposes, I chose the same level of detail that Paolo’s free tool used for the image &mfash; six levels of detail — and changed the file format to JPEG, leaving the compression ratio at the default value.

The result was the exact same number of files (2,371) but with a smaller footprint on the server, at 8.54MB (vs. 26.74MB for Paolo’s tool).

As both apps appear to produce the same quality imagery, the JPEG compression algorithm Valery uses is likely more efficient — his JPEG files are smaller, even as his KML files are larger in size, on average. Note that there are artefacts at the edges of the tiles made by both apps.

Superoverlay beta:

Super Overlay Tiler:

Google Earth base layer (from same original file):

A couple of observations: The option in Valery’s app of turning the output into one big KMZ file certainly simplifies things, but you do then lose the main advantage of regions-based code, which is that you only download those tiles you’re zooming in on, and not others.

Also, neither application at the moment churns out web-friendly code. The <href> tags in KML point to local file names, not URLs, and while this is fine if you have the files sitting on your hard drive, once you upload them to the server you get code that can’t find the imagery. I’m currently solving it with a deftly placed multi-file search & replace query in BBEdit before I upload the KML to my server. The other solution — turning the entire output into one KMZ file — defeats part of the purpose of using regions.

Finally, if you don’t know the exact coordinates of the boundary of your image, then Paolo’s solution remains the easiest ‐ letting you use Google Earth to help find two points that can pin the image to the base layer.

(Valery has also made an example file with his application, downloadable here.)