Google Maps Pushes Dagger Into Bing Maps and Pictometry
When people always asked me, what did I like about Bing Maps, it was always the Bird’s Eye View. It was something that Google didn’t have in their map API and it really was useful since a top down view doesn’t always tell the full story. Well today Google announced that they too are going to have what they call Aerial Imagery in their API. Yep, that is the same perspective view that Bing Maps has with their Bird’s Eye View.
Hard Rock Hotel - San Diego, CA
Now of course how could one really get that excited about Google adding oblique imagery. Well unlike Pictometry, you can zoom out much further on the data so you can see how one might get from the Hard Rock Hotel to the San Diego Convention Center (Bing Maps is of course out of date).
Gain better perspective on where things are
Isn’t that so much more valuable than the top down version? I just can’t see how any current Bird’s Eye View user can continue to use Bing Maps over this (Shall we count the days until Zillow changes?). What I’m interested in seeing which company can build out their “new service” first. Google with their oblique aerial view, or Microsoft with their street level view. I think we all know who will be throwing more resources at getting this done, don’t we?
The one weird part is this appears to be in the v2 API only. I suppose this is because launch customers Orbitz, Redfin and Trulia are all using that version of the API on their websites. v3 can’t be far along.
Dave Brubeck Quartet flying over Google’s Oblique Aerial Imagery
Any idea who is collecting this imagery for Google?
This is a great move on Google’s part, but hardly a dagger (yet) in Bing for two reasons:
1) Coverage for Google’s aerial imagery is extremely limited.
2) What’s stopping Bing from simply negotiating a new deal with Pictometry for additional zoom capabilities and/or incorporating the measuring tool?
Yes it is limited, but you can be assured that they will roll this out much faster than Microsoft will be updating their out of data Bird’s Eye.
As for MSFT and Pictometry, I’m not sure the Pictometry guys want that. That would diminish their “professional” sales.
I think Microsoft has the advantage in that they are leveraging Pictometry’s existing business and thus they don’t have to shell out all the cash on their own. Google on the other hand has to pay for everything themselves, but they do have BIG pockets.
I think it is important to understand that imagery currentness varies by location. In Burlington, VT the Bing imagery is quite new (Google’s is 5 years old) and there is oblique imagery in Bing. No such luck (yet) with Google.
The Pictometry guys have told me if Microsoft came to them today to do the deal, they’d decline it.
Pictometry wears the pants in this relationship and they don’t value Bing Maps.
Indeed, Microsoft has a huge lead because Pictometry is the leader in the oblique imagery business. Just like Google has the big lead in StreetView.
Limiting it to v2 is fine by me. While I like the direction of v3, it is a work in progress and is missing some built-in support I quite liked (easy KML loading via GGeoXml rather than the contortions of subclassing OverlayView).
I wonder when Google’s Goggles will be able to search the 3D warehouse. These aerials are a used by Building Maker to allow free crowdsourced labor to extend Google’s 3D warehouse.
If Goggles could search the library and georeference photos of buildings, Streetview could present them in a slick PhotoSynth-y style – adding yet another location based advertising revenue stream for Google. Streetview has a zoom tool, but there just isn’t much to zoom in on – yet.
I posted this on All Points as well, but the Building Maker data for Detroit listed “Sanborn” in the copyright area. I tried out the developer sample page on Google and panned around San Diego. Pretty sharp imagery. I wonder if Google will “suddenly” unveil in 2010 that they’ve flown most of the country?
Nope. Not even close or possible.
Hence the
yodel. Do you have some insider knowledge as to how this might not be possible?
Yea after working in the VE shop I see now how much it would take fly, store and process the imagery and I think the only 2 companies that actually COULD do it are MS and GOOG. The gov’t could probably do it if they wanted to but all they need to do is buy it from one of the other 2! But since the gov’t subsidized the commercial satellite imagery market and the DoD are Digital Globe and Geoeye’s biggest customer, they’re essentially the inhouse supplier anyway.
The funny thing is (nervous laugh funny not funny haha) is that’s sort of what MS is trying to do with the Clear30 program. I chatted with a compadre in the industry a couple of weeks ago who told me MS was basically using up a lot of the plane and sensor resources in the country for that project, something I had never thought of until he mentioned it.
If you are looking for daggers I think that Bing is the one holding the bloody shiv.
Last week they rolled out their street view imagery and it wasn’t just plain panoramas like google maps, the images are actually mapped onto rudimentary 3d models of the buildings. The effect is subtle but significant. Microsoft’s street view cars have been using lasers to actually create 3d maps of all the locations that they visit.
Combine that with the already robust photosynth technology and Microsoft is winning the technology war. Sadly though the one place where google kicks their butt is in frequency of updates and keeping images current.
Google seems to be a lot more willing to dedicate resources to maps where Microsoft seems to be putting their resources toward innovation. It will be interesting to see how it all pans out but certainly it seems like the end user is the winner.
We’ll see Hans. Past performance from Microsoft doesn’t mean innovation gets supported.
Actually they aren’t rudimentary models, they were compiled in 3d by subcontractors for MS. I haven’t heard any mention of it lately and I assume it must be on the bing maps site by now but the plan was to use the higher resolution Pictometry images for the building sides. The beta cities for that showed some pretty phenominal detail for the buildings.
A quick check of Vegas and I don’t see what we saw in the office but it’s probably still in the works I assume although there is no telling how priorities changed after the big blowout in May.
[...] it’s hard to tell whether this move by Google puts the “dagger into Bing Maps,” the recent moves by both companies certainly show that they are aggressively expanding [...]
Gentlemen, it will take many MANY years for Google to come close to what is available on Bing’s Birdseye today whether in America or overseas. Its a lot easier to go down some major roads with coverage in LINEAR miles than it is to fly millions of SQUARE miles. For example, Google is presently holding a contest for a google streetview bike to go to particular college campus that is voted “the winner”. Bing Maps already has every major college campus covered from every direction.
Actually MS is going to try the square mileage to:
http://media.digitalglobe.com/index.php?s=43&item=187
That actually is good news compared to google.
In fact in the new beta Bing Maps Silverlight control you can zoom out into a new generated oblique mode beyond the limit of the actual pictometry images.
I’d hold my final judgement for how they compare when they do a complex city with tall skyscapers, not all imagery is created equal.
For example go checkout http://nearmap.com who are really doing something amazing, processing and releasing high res imagery of Australian cities every month, about 2 week delay from capture to live on the web. They too have a generated oblique mode.
Bing now has, Satellite (DigitalGlobe), Aerial (Ultracam), Generated Oblique (textured 3d models), Oblique (Pictometry), streetside (Ultracam!!!) and Photosynth (user generated) all in one interface.
First off, Google has much better imagery than Bing over the whole world. Second when was the last time they updated their 3D models? Third Streetside is locked up behind Silverlight and Photosynth has no user generated data in it because no one uses Microsoft APIs.
Harsh yes, but to even think of comparing Google with Bing Maps is hilarious.
But he if I built a business on Microsoft rather than open APIs, maybe I’d feel like I had to defend it, eh “James”?
Well phooey I lost a long comment I was working on here. Anyway bing maps has many 10′s of thousands more 3d buildings (it wouldn’t surprise me if it was low 100 thousands) than google at least for now because the buildings were a major focus when the original Virtual Earth program started.
I’m certain there are some issues just by default because of new construction but Gmaps/Sanborn have a long way to go to match the number of buildings that are in bing maps.
Didn’t Sanborn make most of the 3d buildings in Bing Maps? And hold all the rights to those buildings?
Seems like it would not take Sanborn long at all to catch up Google if Google wanted to catch up.
Besides, 3d buildings are ridiculously easy now with the extensive lidar deployments going on right now (statewide Iowa anyone?). Remember, there is some $12M sitting out there for first time lidar collection.
Actually no the buildings for VE/bing were compiled offshore in India (of course). Sanborn may have been involved at a very early stage but if so they were out of the picture quick and the VE shop dealt directly with the contractors.
It would take longer than you think to catch up in the sheer number of buildings that bing has online. I processed 10′s of thousands myself and I was just 1 of the group doing it. It’s a LOT of man hours to compile and process (successfully).
Were you using lidar to do it though? I think that is going to be the big difference now (especially ground based lidar). If you have the data, you can do a 700,000 structure county like St Louis in a day.
[...] it’s hard to tell whether this move by Google puts the “dagger into Bing Maps,” the recent moves by both companies certainly show that they are aggressively expanding [...]
The person who finds a way to turn bird’s eye views into a revenue stream will be the first person to shove a dagger into the others back.
the revenue stream is in the hit counts and visit time for end-users.
If [let's say on average] every member of your audience spends an additional 2.35 seconds browsing the site and happens to see 1.27 additional items of interest that are posted in that view, the ad rate per user just tickled up the meter an incremental amount.
Let’s say I’m Bing and you’re Google – how much out of our annual marketing budget will it cost to move the needle on per-user stickiness? Is it worth spending two- or three- or even five-hundred million a year on content for my virtual view of Earth in return for, oh…an additional 2.35 seconds of per-user per-session? How many user views do Bing (second-raters that they are) and Google (big brother that they’re becoming) have per minute, worldwide? See the math???
The math for the revenue stream is very simple, and both Google and Bing are recalculating the metric every millisecond. If it didn’t pay off in eyeballs, the arms race for who has the kewlest view would have withered long ago.
Read up a little on what Google’s done to the newspaper business [this is pretty good http://www.brandrepublic.com/News/972492/Murdoch-the-ad-model-dead/ ], and you’ll get an idea of what this is all about. Geography is nothing, it’s all about views.
The web’s kinda different, ya know?
By expediting the crowdsourcing of 3D building creation (via Building Maker) I think Google is in a better position to monetize their aerials than Microsoft.
I don’t think streetview has reached its potential. It needs more photos. Higher resolution photos in streetview would bring in a lot of ad revenue.
For example, in this streetview scene the image is not something attractive to advertisers. However, if there were a higher resolution image of the business available via a photosynth style UI it would be more appealing.
Notice the wireframes as you move the mouse around. This is similar to the feel of photosynth, but when I click nothing happens. I suspect this will change.
If Google could leverage Goggles to search the 3D warehouse and then determine camera position, angle etc. many photos could be added to streetview.
The major difference between Bing and GM, other than coverage and image quality, is the “seamless” of tiles. Google’s Oblique is “rectified”, it has a consistent scale across the view port. Bing/Pictometry is sliced true perspective image, and the scale changes. When you pan to the edge of the original pictometry image, you “JUMP” to another image, mostly from a different view, that leave many user confused.
Google’s GObliqueMercator projection solved the seamless pan issue in a very smart way, way better than Bing. You can hold your arrow key and pan from south of a city to north end. It’s not possible with Bing.
Google’s method of stitching together obliques has some distortion problems. A quick look around my old haunts in San Diego shows more than a few noticeable seams and leaning walls, some fairly egregious.
Probably doesn’t matter to the average user, but let’s not pretend that Google’s method doesn’t have its drawbacks.
Personally, I can’t wait for Google to build out their oblique imagery, but I don’t think this will kill bing maps by any means.
In this context it doesn’t matter because this is visualization, not analysis.
For true analysis, I’d be using the raw Pictometry on its own with their tools which gives you that precision that one needs.
I agree, but scenes like this
http://dl.dropbox.com/u/52511/subbase.html
just look bad.
Agree. For analysis you use Pictometry (when you are a customer) to integrate with your GIS and use the various tools. For visualization, use whatever you prefer…
The imagery isn’t free either…we’re all paying for it one way or another.
The distortion in this case is not that different to the old good “rectified” ortho images. Anything taken from a camera, as long as it does not have infinite long lens, will have such issue. If you have enough shots and have enough control points, the distortion in theory can be minimized to sub pixel. Pictometry images have similar leaning walls at the edge of the photo.
You can zoom further out on Bing maps as well. See for instance this: http://www.bing.com/maps/explore/#/3tx4lbmsw9y700y4
Bing use their 3D model to generate a seemless rendered 3D Oblique view. Zoom further in and it switches over to Pictometry imagery. Zoom further out and it switches over to top-down aerial.
This even works with mountains as well so you get a better “feel” for their shape.
Proprietary extensions to view that Morton. What happened to your open SharpMap days?
It wasn’t 3d modeling, it was running actual AT on the oblique Picto data sets so they drop in place on the surface model. Pretty cool eh?
Could it be ImageAmerica?
http://news.cnet.com/8301-10784_3-9748227-7.html
@Brett, sorry I guess I can’t reply within that post anymore.
No LiDar was involved in the making of the buildings on bing maps. They were all compiled by hand in their exact locations with 3d softcopy software based on the rectified UltraCam imagery at outsource shops in India. Once in a while there would be a mild panic if a building set came back and they were off by 1-1.5 meters and the peeps had to figure to why they were off.
I have to disagree with the idea that LiDar is the magic wand that would make that job so easy just simply from an economic standpoint. The low 100′s of thousands (I’m pretty sure it’s low 100′s anyway) of buildings published in bing maps were compiled by hand with pretty cheap labor whereas a company using ground based LiDar would need multiple vehicles and LiDar units in the field just to cover 1 big city and that would probably take a few days of driving. If a person is reasonably competent with a softcopy system that 1 person could easily compile a simple building in less than 60 seconds into a very small CAD file. Frankly, I think the cost/benefit analysis of capturing buildings via ground based LiDar versus having them compiled would be no contest for keeping things simple and cheap. Even if it took a bit longer to do it by hand you still have vastly less data and processing to go through on the back end. After all that you still have to get the correct images on the sides of the correct buildings. Now multiply all that fun by several hundred cities!
I’m not sure what you meant by “you can do a 700,000 structure county like St Louis County in a day” but you maybe could clarify a bit for me? I’m assuming that’s the unincorporated area outside the city of St. Louis? I don’t think you could possibly drive all that in a day and I’m not sure why you’d want every single one of the 700,000 structures online but I could see it if you were a local government entity. Of course, I can also envision the torch bearing mobs coming down the street as people realize their houses are online in 3d…
So even if you capture those 700,000 buildings in 3d with air or ground based LiDar then process that gaggle of data how will you get the correct images in place on the sides of the buildings? Ponder that next step for a while then add it to the workflow
If someone wants to pay me to do that cost/benefit analysis I’m available with plenty of free time!
I meant specifically aerial lidar. Sorry about that confusion. Ground based lidar would only be for specific structures (bridges, the Arch, etc.)
And I was referring to the grey side extruded or simple shaped 3d buildings which make up the bulk of the 3d buildings; not the skinned buildings.
(But, to add to that, there are several photorealism processes coming out of UW, Oregon State, and UCSB that can skin 3d buildings awful fast. Focus has been on natural scenes, i.e. skinning topography and vegetation, but the processes should adapt to buildings pretty readily.)
LiDAR, schmidar.
Cost of collection at resolution sufficient to resolve rooftop details, then cost of collection of oblique aerials to paint the grey boxes with reality, compared to…
Aerial obliques, camera models, and dozens/hundreds of reasonably-efficient resources working on “the subcontinent” or wherever else you can pay folks $10 a day to do the work.
Which gets you a good enough product at less expense? the latter, my friend.
Or better yet, you assemble the databases of obliques and put the task out there for townspeople to do themselves…sorta like, what Google’s BuildingMaker app, isn’t it?
We’re living in the land of good enough for public viewing in these applications. It’s a different standard than the last 40 years of GIS. Sometimes ‘better’ and often ‘worse’ but always viewed under a different set of criteria.
Sounds like you have not looked at the cost of lidar lately. 1.4m point spacing with 15cm vertical RMSE (the new USGS standard), easily enough to give you rooftops, is cheaper than 6″ orthos now.
Btw, look at the new posting about CloverPoint to get an idea of where 3D skinning is going.
[...] Tables API. That’s a big deal too. And in the final big deal Google pulled off this month, Google Maps is getting oblique photography. Coverage is pretty limited right now, but that should grow over time. This is the one thing that [...]