Panoramic Street Images on Yandex.Maps

Panoramic images on Yandex.Maps let visitors take a virtual walk around a city’s streets. To show how a city looks in reality, thousands of photographs have to be taken. Naturally, this is no easy task. It requires a detailed shoot plan to be drafted for every locale. For big cities, a basic range of locations to be photographed normally covers the centre, all the main streets, interesting places on the outskirts and popular nature spots or historic sites, such as one-time official residence of the Russian Emperors, the town of Gatchina outside St. Petersburg, for example.
The photos are mostly taken from a GPS-equipped motor vehicle with an unusual contraption on its roof – a special device holding several cameras with resolution of no less than 10 megapixels. The car drives slowly along its designated route, and every 20-50 metres all of its cameras take a picture – which is why you can look in any direction in the panoramic images. Photos are taken every 20-30 metres in the historical centre or on a narrow lane, and less often – every 50 metres – in areas where there are no landmarks of note.

Working on mistakes

In areas that are closed to motor vehicles, cameras are attached to a bicycle or carried by hand. Moscow’s Poklonnaya Gora park was photographed from a bicycle, and the famous Odessa steps were shot by hand with the aid of a tripod.
Yandex also photographs some cities from the sky, to provide birds'-eye view panoramas of historical and cultural sights. In St. Petersburg, for example, Yandex took aerial pictures of hundreds of landmarks and the most beautiful locations around the city.
Aerial photographs are taken using a helicopter or an airship. Just like on a car, they are fitted with a device holding cameras that look in different directions. The photos are taken from 150-200 metres, a height at which all monuments and buildings are visible. When photographing from a helicopter, another special device is also used to suppress the vibrations which would otherwise negatively impact on photo quality.
The Mi-2 is one of the most suitable helicopters for aerial photography:
Not surprisingly, panoramas look best when the streets are free of traffic jams and when the weather is good. For on-ground panoramas, we try to shoot on sunny days, early in the morning when there are few cars on the roads. For aerial photography, soft dispersed light is best; in bright sunlight the shadows appear to be sharper, so pictures from the sky are best taken in the morning or afternoon.
After a photo shoot has been completed, the pictures have to be collated into panoramas. Photos taken from an automobile or bicycle are put together to form 360-degree views, allowing the viewer to look all around while taking a virtual stroll around a city. One circular panorama consists of four photos and looks like this:
Photos taken from the sky produce spherical panoramas. These consist of five photographs and allow the viewer to look at the city from above. The radius of a panoramic view ranges from 2 km to 10 km.
The photos are joined together into panoramas automatically using data on the coordinates and direction of the shot (which is why GPS is required). This information allows connections between pictures to be made. Any pictures with erroneous GPS or direction data are discarded.
Street Panoramas allow users to virtually move around a city. The impression of strolling about is created by the way one panorama replaces another, as if the user was actually moving through the scene. To speed things up, Yandex doesn’t load the whole panorama, but just the fragment visible to the user. In order to do this, panoramas are cut into smaller squares, or tiles, of 256 x 256 megapixels each.
As the initial photographs are of very high quality, the panoramas may be examined in detail. There are several zoom levels, and as you move in on one, the next is already loading. Every level has its own set of tiles.
For ease of navigation, streets are labelled by name, and buildings by number. Because the view coordinates and direction are known, buildings and other objects can be automatically identified on the map and marked on the panoramas. Metro exits, railway stations and airports are also labelled, as are city landmarks on aerial views.