Mapillary enables anyone, anywhere to create their own street-level imagery and use it for updating maps. For viewing, interacting with, and navigating between the imagery we provide the Mapillary viewer, which you can access on the Mapillary platform as well as add to your own integrations.

For a long time it has been possible to look around in panoramas on Mapillary by panning the image. For regular images, we've been limited by the field of view of the image when trying to pan. The only way to look around has been to use the turn navigation arrows. At the same time, we know that lots of people are using multi-camera rigs to capture a surround view of an area.

To make looking around in regular images more natural and the interaction more similar to the one in panoramas, we have developed the combined panning feature with the recent update of MapillaryJS (the library used for building our viewer). With the combined panning, it is possible to look around seamlessly by panning between a combination of multiple regular images.

The combined panning interaction

Besides the combined panning interaction, we have also improved the compass icon to indicate the possible panning field. The possible panning field determines how much you can look around by panning left or right for any given image, any looking direction, and any zoom level.

How it works

The combined panning feature makes the viewing experience more natural by panning in regular images to look around. In many places, it will be possible to pan to the left or right instead of having to use the turn arrows. The experience will be more coherent across image types when interacting with the viewer.

The possible panning field indication will tell you if the image you are looking at is a 360° panorama or not. It will also help you realize if it's possible to pan to the left or right at any given moment. Previously, you could only guess from the image appearance if it was a 360° panorama or a regular image and whether you could pan or not. Now that information is always available.

The compass indicator consists of three elements:

• The circle sector indicating the field of view of the viewport camera, i.e. the angle between the left border and right border of the image viewer. If you zoom in, the field of view becomes more narrow, and if you zoom out, the field of view becomes wider.
• The circular arc indicating the possible panning field. If it's a full circle, you are looking at a panorama. If it is an arc, it indicates how much you can pan in each direction in relation to the direction you are currently looking at.
• The triangle indicating the north direction relative to where you are currently looking.

The three elements of the compass indicator. The circle sector indicating the field of view, the circular arc indicating the possible panning field, and the triangle indicating north.

Four different examples of the compass indicator. From left to right: 1. A panorama as indicated by the full panning field circle. 2. An image with a narrow field of view where it's not possible to pan in any direction. 3. An image where combined panning is available. It's possible to pan in both directions. 4. An image where combined panning is only available to the left.

You can try the combined panning in the viewer below. Pan by dragging the mouse to the left or right. Release the mouse when you want to navigate.

Drag to pan, release to navigate, observe the compass indicator to verify if panning is possible

The combined panning functionality is now also available in the Mapillary web app, where we also have added zoom controls to the viewer for convenience.

Capture tips for enabling combined panning

There are some criteria that need to be met for combined panning to be available for an image on Mapillary.

• The technology we use for 3D reconstruction—Structure from Motion (SfM)—is able to find enough correspondences between the images.
• Images have been captured in multiple directions sufficiently close to each other in the area according to the 3D reconstruction. Only images captured within a radius of 5 meters from the current image in the viewer will be considered as candidates.
• There is sufficient, but not too much, overlap between the images captured in the different directions.

The combined panning feature works best for images with a wide field of view, for example GoPro or Sony action cameras. For the best panning experience, the cameras should preferably be mounted close together and synchronized when capturing, but images from sequences captured at different times are also candidates if they are close enough to each other.

Take a look at the MapillaryJS documentation and examples if you want to integrate MapillaryJS in your own application.

/Oscar, Computer Vision Developer