how does google produce streetscene?
March 21, 2009 11:36 AM Subscribe
As a photographer with particular interest in 'stitcher' type panoramic images I was interested in the hardware and software that Google have used for their street scene application. Does anyone have any info? What type of camera and lens combo and specialist software, even if it was written just for Google.
I once saw a presentation by one of the guys behind Street View. There's a lot of processing that goes into making the perspective look right. A naive implementation, i.e. pasting together vertical strips from the centers of lots of overlapping photos, tends to make near things look "scrunched" (like trees on the sidewalk), and far things too wide (alleyways, stuff in the distance).
I can't find the presentation online, but I bet it's available somewhere
posted by qxntpqbbbqxl at 12:19 PM on March 21, 2009
I can't find the presentation online, but I bet it's available somewhere
posted by qxntpqbbbqxl at 12:19 PM on March 21, 2009
I have a friend who is responsible for coordinating the image processing done for Microsoft virtual earth and can attest, it does require a huge amount of computation to get things stitched up nicely.
posted by Good Brain at 5:16 PM on March 21, 2009
posted by Good Brain at 5:16 PM on March 21, 2009
This thread is closed to new comments.
posted by rr at 11:45 AM on March 21, 2009