Hacker News new | past | comments | ask | show | jobs | submit login
Microsoft Street Slide: Street view will never be the same (neowin.net)
155 points by iamelgringo on July 28, 2010 | hide | past | web | favorite | 62 comments

It's really amazing how decisively Microsoft is beating Google in world-mapping, despite being second or third to market. Here's some data points:

1. Microsoft's take on Google's idea of hybrid map rendering (labels overlayed on satellite/aerial images) has yielded the most beautiful hybrid I've ever seen on an online map.

2. Sometime ago a commenter of my blog needed to go from Moscow to Shanghai, and Microsoft's map was the only one that produced a usable route. (For this you need to road networks of all the countries along the way, and not choke on huge routes.)

3. Photosynth, 'nuff said. And the followup "finding paths through the world's photos" by the same research group. And that mind-blowing demo (given by the same person, Blaise Aguiera y Arcas) where they incorporated a live camera feed into panoramic views. Look them all up, you won't be disappointed.

4. Bird's eye view and the streets overlaid on top of that.

5. 3D cities in Bing Maps are created professionally, unlike the cities in Google Earth, and it shows.

6. Bing's street-side view has exceptional 3D cues of movement.

And now this. I love watching all this stuff slowly come together.

>"It's really amazing how decisively Microsoft is beating Google in world-mapping, despite being second or third to market."

Bing's mapping tools are built on/what were Multimap aren't they? AFAIK multimap preceded Google Maps and had aerial imagery - a mix between satellite and street view - before Google and before it was handed over to Bing? (This has been my perception, could well be wrong).

Doesn't this mean that Microsoft is beating Google in mapping because they bought the leader in mapping at the time and have [amazingly] built on that?? Slightly less glorious.

Um, citations? I don't think Bing Maps were built on Multimap in any way, and I don't remember Multimap having aerial images before Google. I could be wrong, though. AFAIK, the only map that had aerial data before Google maps (and the only draggable tiled map before Google maps) was map.search.ch, built by Endoxon which was later bought by Google.

I did take pains to stress this was my perception, note that GoogleMaps is not the same everywhere, neither I suppose is MM or BingMaps.

>I don't remember Multimap having aerial images before Google

Here's a good review from 2005 of the situation in the USA, http://www.webpronews.com/insiderreports/2005/05/24/msn-vs-g... ; I'm not in the USA.

From 2003:

"The public web site http://www.multimap.com provides a range of free, useful services to assist with everyday life.Key features include street-level maps of the United Kingdom, Europe and the US; road maps of the world; door-to-door travel directions; aerial photographs; weather forecasts; the London Underground map; links to location information; and services such as hotel, restaurant and entertainment booking.Multimap.com, which was recently voted Best Internet Service in the World Communication Awards 2002, attracts over 6.3 million unique users every month and is one of the top 10 most visited websites in the UK." (http://www.directionsmag.com/mobile/news/index.php?duty=Show...)

From 1999: "Aerial Images, Inc. and UK Perspectives announced today that high-resolution color aerial photography of all of the United Kingdom will be made available for viewing and purchase on the TerraServer(TM) at www.terraserver.com ." (http://www.highbeam.com/doc/1G1-54597413.html)

Not sure what it was built on, but don't forget Microsoft had TerraServer http://msrmaps.com/default.aspx long before Google bought Keyhole for their mapping technology which was limited to a desktop application at the time.

Now they "only" need to get the mindshare of the users - which is, granted, much easier if your product is better than the competition - word of mouth should take a bit of the burden off that.

Does anyone know if Bing Maps has started sending out teams to collect street view images? AFAIK they only have parts of Seattle and San Francisco and that is several years old. I have not installed Silverlight so perhaps it's already there and I just can't see it.

StreetSlide IS very neat but they're going to need Google-scale street image coverage to make it useful.


They have a lot of street-side data. Install Silverlight and try it, I just tried some major American cities randomly and they were there.

Bird's eye view is really good. I will use Bing Maps just for that. I'm surprised they don't have anything like that in Google Maps.

They do: http://goo.gl/maps/H4IZ

(admittedly, it's only available as a Labs feature and in a handful of cities).

Heh, I guess you can't direct-link to them either. (Need to click the aerial button once you follow the link, which I assume only shows up if you have the Labs feature turned on.)

Microsoft has really good coverage for theirs, and I find it very useful to actually find places, since it's not as low-level as street view.

In the 1970s, Stanley Kubrick sent a photographer out to take thousands of photographs step by step along London's Commercial Road. The result was somewhat like Street Slide: http://entertainment.timesonline.co.uk/tol/arts_and_entertai...

Kubrick's boxes - hundreds upon hundreds of them - were receptacles for millions of details about films he mostly never made. One contained photos of Commercial Road in London, taken over the course of a year by his nephew, Manuel Harlan. Harlan stood on a 12ft ladder to photograph every building on the road and laid out the results for the great man to examine in his lair near St Albans. “Sure beats going there,” said Kubrick, delighted.

Kubrick would have had a field day if he were still alive.

I was excited when I followed the link, not so much when I saw "Microsoft researchers have developed" as the lead in sentence.

CNET confirms: "MSFT has not announced if and when it will be making it to the Bing Maps, or any other map-embedded Microsoft products or services"


Hope they prove me wrong and just ship.

MSR is one of the top computer science academic-research labs in the world -- their output, especially in my field of computer vision/graphics, rivals any university at all of the top conferences. As other comments have stated, they have an unbelievable amount of freedom in tackling projects and have managed to attract lots of very good people. This freedom, however, comes at the price of research not necessarily making it to commercial product (at least not right away).

But that's not the goal. This kind of research solves the "hard" problems, hopefully making it easier for engineers at MS (or at other companies) to make a working system. This is not to downplay the latters' roles -- as any engineer can tell you, making a working system for consumers is a very time-consuming process that requires a lot of manpower, engineering, graphic and usability design, etc.

The difference is simply in division of labor. Researchers push the boundary of what's possible, engineers fill in the newly excavated space with working and polished products.

Incidentally, while Google manages to hire all the best engineers/hackers/programmers, their research groups aren't at the level of MSR's yet. In part, I think this is because their research actually makes it to product fairly fast -- because the researchers have to spend a fair amount of time integrating new research results into products. I wonder if this culture will change...

It seems mistaken to assert that Google's R&D enterprise is not "pushing the boundary of what's possible". Stopping with published papers and conference prominence, when you have the reach that MSFT does, seems to be missing an opportunity.

And given this, which culture is under more pressure to change, MSR's or Google's?

I think you're confusing MSR with MS corporate. The corporate culture at MS seems to be very different from the MSR culture, and it's the corporate culture that people pick on.

Comparing MSR with Google doesn't make sense because MSR is very independent from MS corporate, culture wise at least.

Just to give an example: MSR is developing a Haskell compiler, and they run Linux on some of their boxes.

Nope, not confusing them. I'm well acquainted with MSR. Just saying that the focus on papers and conference presence/sponsorship has not translated into impact on products. Seems a pity.

This has got to be the greatest thing MSFT has done for the web since Ajax. I don't know; maybe I'm missing something. Assuming they aren't going to patent it or something similarly insane. And assuming it works in a normal browser. There is hope though the video shows a demo on an iphone.

I've known a number of really smart people at Microsoft Research. And I've always wondered why they seem to produce so little. A secondary curiosity - how do they manage to hire such great researchers with their tarnished reputation over the last decade.

I've known a number of really smart people at Microsoft Research. And I've always wondered why they seem to produce so little.

it might be because the primary 'product' that MSR produces is academic research papers, not production software (or even prototypes). they are an academic research lab devoted to doing basic research, not an R&D incubator

> A secondary curiosity - how do they manage to hire such great researchers

From what I can tell, it's in large part because, post-Bell-Labs, nobody else offers a similarly nice job for senior researchers, where you have wide intellectual freedom, a nice salary, and few restrictions on publishing. I assume junior researchers can't just start their own projects, but it seems they let the famous researchers do more or less whatever they want.

Fortunately, that assumption is false: junior researchers can start their own projects. I joined Microsoft Research a year ago. In that time, I've been encouraged to come up with new projects where I can lead, as well as make contributions to existing projects. As you point out, this degree of independence is rare in industry, so it makes MSR a competitive place to work.

What are the typical languages you guys use? I know F# came out of MSR.

Whatever you want to use for the project. For a recent project, one of my coworkers created a domain specific language and the rest of us coded in that language. For another project, I use C++, because I'm leveraging an existing code base. Colleagues of mine use F# for program analysis tools, which they love. As you say, F# came out of the MSR Cambridge languages group, that's a great tech transfer story.

That being said, there are a lot of internal libraries or projects written in C/C++ or C#. So I see a lot of those languages. I've also been learning ASP.NET, in part so I can better understand cross site scripting attacks on applications written on that platform.

"I've known a number of really smart people at Microsoft Research. And I've always wondered why they seem to produce so little."

Taking something from research into a full-fledged product at a company like Microsoft is like turning an aircraft carrier - it has a lot of expensive hardware on deck that has to be kept safe and has to be down slowly given that the seas are shifting constantly.

If the aircraft carrier suddenly notices a large island right in front of it then it might should consider a sharp turn anyway, even if a few F-14s get dinged up a bit.

The aircraft carrier has advanced radar and sonar to avoid that situation :p

All the radar and sonar in the world won't do the aircraft carrier any good if Captain Balmer ignores them and steers based on where he thinks the island ought to be instead of where it actually is.

But you can't put an aircraft carrier on an island unless it has a deepwater dock!

This kind of thing interests me as a geek, its clearly pretty cool, but I dont think it adds that much to the standard street view. As a user, the standard street view is 80% of what I want, they have added another 5% on top of that which is nice, but oddly enough I dont feel (at the moment) as if it is actually that compelling. Its a weird perspective, I love the tech, but I dont think I will ever care enough to go out of my way to use it.

Fair enough,but from my POV, street view has a hardly-usable UI. The way you have to click, render, click, render, click, render, Street View works the way MapQuest worked before Google Maps was built.

Street Slide is to Street View what Google Maps was to MapQuest.

That is absolutely true, of course.

The thing is that I very rarely actually want to scroll right along a street, I usually only want to get a visual of the place that I am going.

Now, there have been times where I was interested in going exploring, but not as a matter of utility - just because the street view ui begged me to and I was interested in how it worked.

Having said that, I believe that there is a definite place for something that lets me zoom along in 3d land to explore the entire street - because I would love to be able to explore Paris for example - but that really requires a free form of movement that this solution still doesn't speak to.

Maybe that is precisely why you didn't want to use it -- google's street view interface was not right for you.

yeah, it could be. I do use street view, but just because I find it useful to have a visual of the entrance and location of the place I am going.

But I agree, aside from the playful inclination to go and explore, its never been a big thing for me.

"I very rarely actually want to scroll right along a street"

I do that all the time for country roads to plan cycle routes - the current Google interface leaves a lot to be desired.

I have to disagree, I think google street view has some really big usability issues, and that anything that helps you navigate the "bubble" data set more easily is a good thing. It adds a lot.

I feel like I'm constantly trying to use streetview in exactly the way that this thing is doing it; for example if Im looking for a particular location I was at, and I know the street, but I forget the cross street, I want to be able to scan down block after block to find what I'm l ooking for.... esp. since I don't tend to think in terms of address- more in terms of relations of streets and landmarks.

interesting, I can see how that would be frustrating.

I tend to use the map view for that kind of thing, with the right level of zoom seeing the names of cross roads and that kind of detail is easy and quick.

To each their own however, clearly this will be loved by many.

Call me when they have something other than a demo. If this demo was from Google, I could see it happening. But not from MSFT, specially if it more expensive (we don't know) to implement this and if there is no near term profit making potential. Thats, in my opinion, how MSFT works.

On the other hand, if there is something cool developed by Google, they will implement it and think about making money later (if it gains traction). Not necessarily because they are altruistic, but because thats how they work.

You will never see something like Google Wave coming out of MSFT; where Google obviously invested a lot of money, opened it up (mostly) for everyone to do what they want and after more than one year in no money is being made and unlikely to be profitable any time soon, if ever.

There is profit making potential - they could put advertisements on the shops above the actual shops in the overview mode

Agreed, but I think the parent is saying MS won't execute.

Looks neat. Have only skimmed the paper but I'm surprised they didn't cite Zheng et al.'s Route Panorama work: http://www.cs.iupui.edu/~jzheng/RP/

I did my PhD in Multiperspective imaging (if that can be considered a field :) and I remember some guys from Stanford did some work specifically on these kind of street views. Here's a link to an overview at least http://graphics.stanford.edu/~ggaurav/research.html

I think that guy in particular might have ended up at Google, it must be a bit painful to see Microsoft release a demo of this stuff.

He has ended up at Google:

"Bio Google::Software Engineer"


Microsoft researchers didn't let corporate politics stand in their way -- they wrote the mobile version for the iPhone, not Windows Phone 7 (Street Slide is still unreleased, mind you).

Now, will Google Maps engineers be able to perform a similar feat and implement a technology invented at Microsoft?

I'm having a hard time unpacking this. Microsoft wrote a demo on Apple's platform so you want Google to write something on a Microsoft platform? Seems like a non sequitur. BTW, Google's stuff has been on Apple's platform since day one.

Ugh. Here's some unpacking, then:

I never said Google should write something on a Microsoft platform. I implied that Google should implement this Street Slide method in Google Maps and mused as to whether the Google brass would allow them to copy a Microsoft innovation.

You see, within Big Companies there is a common problem where engineers aren't allowed to use competitors' technologies because it hurts the corporate image. No one wants to see "Google copies Microsoft with new Maps feature" headlines, just like no one wants to see "Not even Microsoft engineers are stupid enough to develop for Windows Phone 7" headlines. Some people call this Not Invented Here Syndrome. This is a problem because it's counterproductive. Street Slide should be implemented on the iPhone first. Google Maps should add Street Slide. Hopefully their corporate overlords will have the vision to allow the engineers to do so.

> "whether the Google brass would allow them to copy a Microsoft innovation."

Have you seen Google's search results lately?

The web results and the image search updates have no lack of Bing-ish-ness.

Well, you can actually buy an iPhone on the highstreet. If they wanted to do a demo on a Windows Phone 7 phone, they would need to hunt down a pre-production device. So far, WP7 is itself just a software demo.

I believe the message with the iphone is interoperability.

Love the idea otherwise. Very nice experience.

I like how they are listing businesses in the letterbox. Although, in that part of downtown Seattle there are 10 times more Starbucks than what they're showing in the video . . . and I'm not exaggerating.

On 1st or on 4th? There's only one SBUX on 1st between showbox and cherry st, on Columbia I think.

I was thinking more generally about downtown. I counted 10 between Belltown and Pioneer Square.

But to your point, there are 2 Starbucks' directly on 1st; one in Pike Place and one just up from the ferry terminal.

The one in Pike's Place is to the west of Post Alley; it's not really on 1st. There's only one on 1st between Showbox and Cherry, which is the one on Columbia just up from the ferry terminal.

I agree, though, that there are an inordinate number of SBUX-owned coffee shops downtown, especially once you factor in the SBCs.

Brilliant. I find I spend an awful lot of my time in street view trying to do exactly this.

This is very innovative. Great development in user interface and visualization. Something good came out of MS Research.

The general idea can be adopted to different use cases. Hope they don't patent it to dead.

While this is a neat way to navigate panoramas, in practice it is very difficult to get a smooth transition between perspectives without a lot of source images (you can tell when they switch to the unstitched view that they were using a much higher density of images than with street view panoramas, especially at intersections).

Interestingly, multi-perspective push-broom images is essentially how the street-view team got started.

This works fine for long, straight streets, typical of some cities (particularly US cities). Not so much elsewhere...

I thought they were going to allow seamless zooming down the street. That would be cool, and the data is available to do it.

The as-perspective-as-possible transitions could be a lot smoother, but I guess they wanted to minimize the processing power needed, so it can run in-browser etc.

A bit OT I know, but just visited the MicroSoft Street Slide site. They posted a very nice example video.

Let me just say thank you to everyone involved in producing that well-written, well-spoken, well-recorded, straight-forward intelligent piece of A/V work.

Well done.

Cool, but "will never be the same" is a bit much when descibing a relatively minor increment of a basically useless feature.

Neither Street View itself nor the ability to move within it are useless. I use Street View to locate the entrances to unfamiliar stores, find parking, identify landmarks, etc., saving myself a considerable amount of time spent driving (and CO2 emissions).


Registration is open for Startup School 2019. Classes start July 22nd.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact