Check this out: Virtual Earth Streetlevel

Remember how A9 took a car with a few cameras on it down streets of major cities and let you see what the storefront looked like at a particular address? Well, we thought that was cool so wanted to take it to the next level.

Virtual Earth Streetlevel.

That link will take you to a Channel 9 video of Streetlevel that shows off what we did (and you’ll meet the team behind it). They took a van with 10 cameras to Seattle and San Francisco (more cities coming soon) and built quite a remarkable AJAX app. Lots of fun. Drive your own car down the streets of San Francisco. Oh, and you can search for the closest Starbucks, too! It’s like playing a video game. Driving through tunnels is a lot of fun!
How many images? Seattle alone is 10 million images. What a database!

You can check it out for yourself at http://preview.local.live.com/. More info is on the MSN Search blog.

What do you think?

135 thoughts on “Check this out: Virtual Earth Streetlevel

  1. Channel9 should be shut down. It’s a waste of Microsoft shareholder resources and is extremely high-risk from a PR standpoint.

    I visit this blog a lot more often than channel9, I find Scoble’s blog more valuable, and I think channel9 can’t be trusted because it is closer to Microsoft as a purely propaganda arm than the author of this blog.

  2. Channel9 should be shut down. It’s a waste of Microsoft shareholder resources and is extremely high-risk from a PR standpoint.

    I visit this blog a lot more often than channel9, I find Scoble’s blog more valuable, and I think channel9 can’t be trusted because it is closer to Microsoft as a purely propaganda arm than the author of this blog.

  3. It would be interesting to see a transition effect between each image to simulate movement. Maybe it should zoom in to the current image before displaying the next image. Of course, that assumes the images will load much faster than they are now.

  4. It would be interesting to see a transition effect between each image to simulate movement. Maybe it should zoom in to the current image before displaying the next image. Of course, that assumes the images will load much faster than they are now.

  5. Goebbels, why don’t you just swallow your pride and watch the video? i’m sure you will feel a lot better afterwards.

    This is really interesting stuff. Although it’s very rough at the moment, i’ll be very interested to see how this turns out in a year or so’s time. Thing’s like this take a long time to get right, (i can only imagine the amount of work involved) so i don’t think any reasonable person would expect it to be everything they want just yet. It’s definitely got huge potential for the future.

  6. Goebbels, why don’t you just swallow your pride and watch the video? i’m sure you will feel a lot better afterwards.

    This is really interesting stuff. Although it’s very rough at the moment, i’ll be very interested to see how this turns out in a year or so’s time. Thing’s like this take a long time to get right, (i can only imagine the amount of work involved) so i don’t think any reasonable person would expect it to be everything they want just yet. It’s definitely got huge potential for the future.

  7. About the questions re “priorities” for the ongoing accretion and growth of the image database…

    –images around “public service” locations such as police stations, schools (at all levels), libraries, hospitals, cemetaries?, etc (these are all relatively stable, frequently visited sites, often only once). I guess that “emergencies” aren’t the best time to go check a sites location online, though!

    –heavily-visited tourist sites (as already known, by local/state tourism boards). No need for rural state/national parks, perhaps, since they have unambiguous routing.

    –known “everybody-gets-lost” locations, as evaluated by?? automobile-clubs? local police? common wisdom? state DOTs? Perhaps the images should be sure to show the areas good signage?

    –As previously requested, there could also be high-quality panos and 360 virtual panos at these locations, grabbed by the same vehicle which is capturing the more linear images. The UI would indicate where 360′s are available.

  8. About the questions re “priorities” for the ongoing accretion and growth of the image database…

    –images around “public service” locations such as police stations, schools (at all levels), libraries, hospitals, cemetaries?, etc (these are all relatively stable, frequently visited sites, often only once). I guess that “emergencies” aren’t the best time to go check a sites location online, though!

    –heavily-visited tourist sites (as already known, by local/state tourism boards). No need for rural state/national parks, perhaps, since they have unambiguous routing.

    –known “everybody-gets-lost” locations, as evaluated by?? automobile-clubs? local police? common wisdom? state DOTs? Perhaps the images should be sure to show the areas good signage?

    –As previously requested, there could also be high-quality panos and 360 virtual panos at these locations, grabbed by the same vehicle which is capturing the more linear images. The UI would indicate where 360′s are available.

  9. Well, I’ll probably be more impressed when I get past the lame redirect to a 404 error page because my Firefox javascript blocker hasn’t yet been told to allow the site. Because of the redirect, I don’t get the opportunity to allow scripting on the page you linked to with a click on the options button. I’ll have to go in and do it manually.

    It redirects to here: http://preview.local.live.com/Help/NoJavascript.html

    Note the url: Help/NoJavascript

    So it’s intended to help the visitor who has javascript turned off, no? NO! It’s a standard 404 not found page. An excerpt:
    The page cannot be found
    The page you are looking for might have been removed, had its name changed, or is temporarily unavailable.

    etc.etc. Definitely the wrong error message for the situation.

    I got around the site’s lame error-handling and went for a test drive. Kinda cool. Pretty sluggish, but to be expected across the internet.

    -sam

  10. Well, I’ll probably be more impressed when I get past the lame redirect to a 404 error page because my Firefox javascript blocker hasn’t yet been told to allow the site. Because of the redirect, I don’t get the opportunity to allow scripting on the page you linked to with a click on the options button. I’ll have to go in and do it manually.

    It redirects to here: http://preview.local.live.com/Help/NoJavascript.html

    Note the url: Help/NoJavascript

    So it’s intended to help the visitor who has javascript turned off, no? NO! It’s a standard 404 not found page. An excerpt:
    The page cannot be found
    The page you are looking for might have been removed, had its name changed, or is temporarily unavailable.

    etc.etc. Definitely the wrong error message for the situation.

    I got around the site’s lame error-handling and went for a test drive. Kinda cool. Pretty sluggish, but to be expected across the internet.

    -sam

  11. Kudos to Microsoft, this is very cool, the big G must be but-we-do-the-cool-sh*t ticked off. ;-) A disappointment: searching “moma” in San Fran gives no results.

  12. Kudos to Microsoft, this is very cool, the big G must be but-we-do-the-cool-sh*t ticked off. ;-) A disappointment: searching “moma” in San Fran gives no results.

  13. Brian, thanks, your more helpful than that shat video. Amazing how a couple of sentences work.

    What I simply don’t understand is why they start with a DIFFERENT, CRAPPIER UI in the first place if the intention is to return to a more functional one.

  14. Brian, thanks, your more helpful than that shat video. Amazing how a couple of sentences work.

    What I simply don’t understand is why they start with a DIFFERENT, CRAPPIER UI in the first place if the intention is to return to a more functional one.

  15. Goebbels- In the C9 video, they said that when this gets integrated into the actual Windows Live Local, it will have the same UI/capabilities (pushpins, etc…).

    The directional buttons aren’t meant to be navigational- they are just there to show you that you can use your keyboard to move the car. If you click on them, all they do is open up the help bar.

    As far as the cars, I actually think the “sports car” looks pretty good, and it does the best by far at approximating the 3D, 3-pane view.

  16. Goebbels- In the C9 video, they said that when this gets integrated into the actual Windows Live Local, it will have the same UI/capabilities (pushpins, etc…).

    The directional buttons aren’t meant to be navigational- they are just there to show you that you can use your keyboard to move the car. If you click on them, all they do is open up the help bar.

    As far as the cars, I actually think the “sports car” looks pretty good, and it does the best by far at approximating the 3D, 3-pane view.

  17. Another thing I don’t get is why the UI is so different from Virtual Earth (presumably they were done by the same team or at least if just because it was local photos, they split into separate teams, they should have checked out each others work):

    why are the directional buttons up top instead of by the side as an overlay? (They should be with the lower pane anyway: you essentially don’t navigate the top.)

    why is there no zoom slider (it wasn’t working so well for me because it’s so much slower; why? Presumably the lower pane is the same, but the speed difference is enormous.)

    Additionally:
    I don’t see the reason for allowing the car to be moved. Rotation, yes, that’s necessary, but if you panned and moved the scene under it than it’s less confusing than moving the car a little, panning the scene, moving the car, which is what I notice myself and others doing a bit.

    Emphatically:
    Have I said how pathetic the XBox controller and choice of cars is? Get rid of it, simplify it — Cheezy is not cool, cheezy is not cool. Repeat. And if the choice of cars is any factor in the speed issues, that “feature” is insane.

  18. Another thing I don’t get is why the UI is so different from Virtual Earth (presumably they were done by the same team or at least if just because it was local photos, they split into separate teams, they should have checked out each others work):

    why are the directional buttons up top instead of by the side as an overlay? (They should be with the lower pane anyway: you essentially don’t navigate the top.)

    why is there no zoom slider (it wasn’t working so well for me because it’s so much slower; why? Presumably the lower pane is the same, but the speed difference is enormous.)

    Additionally:
    I don’t see the reason for allowing the car to be moved. Rotation, yes, that’s necessary, but if you panned and moved the scene under it than it’s less confusing than moving the car a little, panning the scene, moving the car, which is what I notice myself and others doing a bit.

    Emphatically:
    Have I said how pathetic the XBox controller and choice of cars is? Get rid of it, simplify it — Cheezy is not cool, cheezy is not cool. Repeat. And if the choice of cars is any factor in the speed issues, that “feature” is insane.

  19. Brian, that’s why figuring out your purpose first usually makes sense? If you want landmarks then: sure, swing around the Embarcadero, duck through the Mission (staying on the Street), hit the Castro, but why waste two months going up and down streets in the Sunset that look the same and offer no landmarks? (But why not just have VR panoramics of the sites than, actually accomplish good stitching.)

    If you want it for directions, just shoot intersections. (This “purpose” I think would be utterly absurd in SF and, say, a city like Boston — the image is not going to prepare you for what you face anymore than an aerial. An image doesn’t capture traffic flow, pedestrian traffic and behavior, etc..)

    If it’s for commercial use, it’ll fail too many businesses too many times. (I’d be happy to hear how Scoble treats this, but I’m not watching a WMV; maybe he’ll start posting on YouTube.)

  20. Brian, that’s why figuring out your purpose first usually makes sense? If you want landmarks then: sure, swing around the Embarcadero, duck through the Mission (staying on the Street), hit the Castro, but why waste two months going up and down streets in the Sunset that look the same and offer no landmarks? (But why not just have VR panoramics of the sites than, actually accomplish good stitching.)

    If you want it for directions, just shoot intersections. (This “purpose” I think would be utterly absurd in SF and, say, a city like Boston — the image is not going to prepare you for what you face anymore than an aerial. An image doesn’t capture traffic flow, pedestrian traffic and behavior, etc..)

    If it’s for commercial use, it’ll fail too many businesses too many times. (I’d be happy to hear how Scoble treats this, but I’m not watching a WMV; maybe he’ll start posting on YouTube.)

  21. Out of date doesn’t seem to be that big of a deal if you can at least be around a year or so, the buildings will usually look the same. I just wish we’d have most of the data before releasing a tech preview. The UI is a little hokey, I’m sure it will be cleaner by release. Very cool overall though. Will have to play with it more.

    Would be great to have some international sites also, but some countries might think the vans are working for the CIA. :)

  22. Out of date doesn’t seem to be that big of a deal if you can at least be around a year or so, the buildings will usually look the same. I just wish we’d have most of the data before releasing a tech preview. The UI is a little hokey, I’m sure it will be cleaner by release. Very cool overall though. Will have to play with it more.

    Would be great to have some international sites also, but some countries might think the vans are working for the CIA. :)

  23. met, Goebbels, et. al.: It raises the question- what exactly is this useful for? I could see myself using it for two main purposes:

    1. Getting really good directions, where I can actually see the turns and lane changes clearly.

    2. Getting a feel for landmarks and the lay of tourist spots in cities where I will vacation.

    For both, getting up to the minute data is not essential. Having data within a year or two should be fine for #1- you should get all but the most recent construction changes. For #2- sure, the little shops and restaurants will change, but the things a tourist wants to see are stable. The Space Needle, the Statue of Liberty, a museum or stadium…those aren’t going to change much over a two or three year period.

  24. met, Goebbels, et. al.: It raises the question- what exactly is this useful for? I could see myself using it for two main purposes:

    1. Getting really good directions, where I can actually see the turns and lane changes clearly.

    2. Getting a feel for landmarks and the lay of tourist spots in cities where I will vacation.

    For both, getting up to the minute data is not essential. Having data within a year or two should be fine for #1- you should get all but the most recent construction changes. For #2- sure, the little shops and restaurants will change, but the things a tourist wants to see are stable. The Space Needle, the Statue of Liberty, a museum or stadium…those aren’t going to change much over a two or three year period.

  25. Goebbels: yep, I took this on in the video too. Yes, they will be out of date. This imagery was taken in late 2005, by the way. So it’s already about three months out of date.

  26. Goebbels: yep, I took this on in the video too. Yes, they will be out of date. This imagery was taken in late 2005, by the way. So it’s already about three months out of date.

  27. met, I wouldn’t even try to address the quirks or fundamental usefulness (a big question mark) of this yet. (Hell, even the sat imagery which is “complete” has quirks and flaws.)

    But this? In a city like SF where hundreds of store fronts might actually change per month, they would always be behind (even if they were flawless) considering it took a month to cover a small portion of the city. If the purpose was to have an image well mapped to address and other “meta” data, it wouldn’t be flaws but rather … LAGS would appear instantly and always!

  28. met, I wouldn’t even try to address the quirks or fundamental usefulness (a big question mark) of this yet. (Hell, even the sat imagery which is “complete” has quirks and flaws.)

    But this? In a city like SF where hundreds of store fronts might actually change per month, they would always be behind (even if they were flawless) considering it took a month to cover a small portion of the city. If the purpose was to have an image well mapped to address and other “meta” data, it wouldn’t be flaws but rather … LAGS would appear instantly and always!

  29. why should you allow users to report mistakes?
    check 1st ave and blachard st intersection (in SF) – the pictures are goofed up.

  30. why should you allow users to report mistakes?
    check 1st ave and blachard st intersection (in SF) – the pictures are goofed up.

  31. I have always wondered why A9 had prolems accurately mapping street addresses with their photos.
    Live’s preview is almost flawless in this aspect. The match between the lower screen and the top is good in most places.

    Suggestions for improvement:

    If you would restrict the car to not move (too much) away from the roads, it would be nicer :) I forsee the difficulties of that, but if your Navteq mapdata is perfect, you wouldn’t have to worry too much.

    Allow people to start maching addresses ?

    Allow people to report mistakes easily, by pointing out on the screen, etc. :)

  32. I have always wondered why A9 had prolems accurately mapping street addresses with their photos.
    Live’s preview is almost flawless in this aspect. The match between the lower screen and the top is good in most places.

    Suggestions for improvement:

    If you would restrict the car to not move (too much) away from the roads, it would be nicer :) I forsee the difficulties of that, but if your Navteq mapdata is perfect, you wouldn’t have to worry too much.

    Allow people to start maching addresses ?

    Allow people to report mistakes easily, by pointing out on the screen, etc. :)

  33. Very cool. Last week in Rochester, I used the “birds eye” views for the first time and was FLOORED by how useful they’d become since initial beta (the ability to pivot on location to see different angles was particularly useful, if buggy).

    This stuff is still all about 2 years away from being ready for the mainstream, but the strides being made in 1) capturing useful imagery and 2) presenting it in a way that is increasingly useful to users; is amazing.

    See you on Wed/Thurs at NewComm Robert :)

  34. Very cool. Last week in Rochester, I used the “birds eye” views for the first time and was FLOORED by how useful they’d become since initial beta (the ability to pivot on location to see different angles was particularly useful, if buggy).

    This stuff is still all about 2 years away from being ready for the mainstream, but the strides being made in 1) capturing useful imagery and 2) presenting it in a way that is increasingly useful to users; is amazing.

    See you on Wed/Thurs at NewComm Robert :)

  35. From a technical point of view, it’s very cool. From a user perspective, however, my reaction is closer to “meh”. As one of those people in “flyover country” (Colorado Springs, specifically) I chalk this up as another toy for the major cities that will in all likelihood never be expanded to cover where I live. My additional minor quibble is the appearance of the “walk” mode — the left and right image panels are quite distracting since they don’t tile with the look ahead. I’m not sure why anybody would use the car interfaces: I don’t want or need additional junk between me and the information I’m looking for…which in this case, would be the ground-level imagery.

  36. Since the map isn’t loading, I’ll digress on Jeff’s point. Yes, there are not enough very well trained GIS specialists, and yes, lots of government and other attempts to collect spatial data have been made over decades in disparate forms with different and developing tools. But now shape files are largely XML (or moving there), and ESRI and the entire industry are moving towards geodatabases in which different coordinate systems and other data differences can be resolved.

    Yes, the GISes that have developed over the last couple of decades may be proprietary or canned or whatever, but the issues you raise aren’t very relevant going forward if the right tools are used by skilled GIS specialists.

  37. From a technical point of view, it’s very cool. From a user perspective, however, my reaction is closer to “meh”. As one of those people in “flyover country” (Colorado Springs, specifically) I chalk this up as another toy for the major cities that will in all likelihood never be expanded to cover where I live. My additional minor quibble is the appearance of the “walk” mode — the left and right image panels are quite distracting since they don’t tile with the look ahead. I’m not sure why anybody would use the car interfaces: I don’t want or need additional junk between me and the information I’m looking for…which in this case, would be the ground-level imagery.

  38. Since the map isn’t loading, I’ll digress on Jeff’s point. Yes, there are not enough very well trained GIS specialists, and yes, lots of government and other attempts to collect spatial data have been made over decades in disparate forms with different and developing tools. But now shape files are largely XML (or moving there), and ESRI and the entire industry are moving towards geodatabases in which different coordinate systems and other data differences can be resolved.

    Yes, the GISes that have developed over the last couple of decades may be proprietary or canned or whatever, but the issues you raise aren’t very relevant going forward if the right tools are used by skilled GIS specialists.

Comments are closed.