The goal of OpenPanorama is to give to the panoramic imaging world a complete and free to use file format based on XML.
   We also believe that it is important to be able to convert other proprietary file formats to OpenPanorama without loss of data or image quality (no change of projection type or recompression of the image).

    This document provides a summary of the different types of panoramas and their functionality , all of which will be supported by the OpenPanorama format.
    In this document we assume you are familiar with panoramic photography. For more information about panoramic imaging, see James Rigg's web site http://www.panoguide.com

    Panoramas are composed of :

I. Projections

    There are four projections in general use today:

    Cylindrical projection provides a 360 degree panoramic view but is limited in the vertical axis - some of the sky or ceiling and of the floor or ground is omitted.
    To display a cylindrical image, it is projected onto a cylinder. The image is then viewed from the center of the cylinder. The source image is mapped  on the cylinder using a gnomonic projection identical to gnomonic projections used to create earth maps.
    This system is used by QuickTime® and other companies to create non 360°x180° panoramas. 

    The source image can be horizontal or vertical and in certain cases be segmented:



    Today, many different solutions exist for storing cylindrical images. OpenPanorama's image description syntax is flexible enough to accomodate all existing image formats in the context of panoramic imaging. See “Image Format part”, below.

Attributes of a cylindrical panorama:
    Because the panorama can be a partial view of the scene (less than 360 degrees), it is important to be able to specify the horizontal field of view (without which the image cannot be displayed correctly).  The vertical  field of view is deduced automatically from the relationship between the height and the width of the image. 

    Points in the panoramic scene can be described in terms of their horizontal and vertical angles within the cylinder. For example, to define 'north' in the image compared to the north of other objects in the panorama.


    Spherical projection involves mapping the image of the environment on a sphere. Unlike cylindrical images, spherical images provide a complete representation of the environment - i.e. including the ground below the camera and the sky. However, it is also possible to use this system for panorama whose vertical field is lower than 180° with specifying a max/min vertical tilt. 

    The source image can be of various types, but the most commonly used is that that requires equirectangular projection.  For a panorama of 360°x180°, the image has a constant aspect ratio of 2:1 (exactly twice as wide as it is high).


    This system is used a lot by people who use PTViewer to display their panoramas. The images used with PTViewer are generally stored in only one image file, as represented in the example above.



Attributes of the spherical panorama:
        Because the panorama can be a partial view of the scene (less than 360 degrees) , it is important to be able to specify the horizontal field of view (without which the image cannot be displayed correctly). The vertical  field of view is deduced automatically from the relationship between the height and the width of the image.

        Points in the panoramic scene can be described in terms of their horizontal and vertical angles within the sphere. For example, to define 'north' in the image compared to the north of other objects in the panorama.


    Very popular since QuickTime® 5, cubic projection consists of projecting the environment on a cube, and positioning the viewpoint in the centre. The visualization software’s projection algorithm corrects the perspectives of the 6 cube faces and gives the impression that the user is in a spherical view.


    The six faces of the cube can be stored in various ways: 

    The OpenPanorama image description syntax is flexible enough to accommodate all existing image formats in the context of panoramic imaging.  See “Image Format part”, below.


Attributes of the cubic panorama:
    Because the panorama can be a partial view of the scene (less than 360 or 180 degrees) , it is important to be able to specify the horizontal and vertical field of view.

    Points in the panoramic scene can be described in terms of their horizontal and vertical angles within the cube. For example, to define 'north' in the image compared to the north of other objects in the panorama.



    Flat projection consists in placing a planar image in the panoramic environment.  (Note that a cubic panorama is made up of 6 flat projections.)
    Flat projection does not provide a useful means of displaying a full panoramic image, but is very useful for adding images to the panoramic scene, or for certain types of hotspots, as we will see later.

    Note that “flat” or “planar” projection is sometimes used to describe how some software displays panoramic images without projecting the image at all – i.e. providing a non-perspectively correct view of a panoramic image.


Attributes of a flat projection:
    We consider that the diagonals of the image are perpendicular to the  line passing by the intersection of the diagonals and the point of visualization. 
    To position the mapped image in our panoramic environment, we need: 

II. Hotspots

    Currently, three major types of hotspots exist: 

   To define hostpots using a mask, a mask image the same size as the panoramic image contains blocks of colour that correspond to each clickable zone (hotspot) within the panoramic scene. Each colour corresponds to a different hotspot. 
    These masks use the same type of projection as the image(s) for which they define hotspots. 
    This approach is very flexible in that a hotspot can be any shape.  However manually defining hotspots in this way is time consuming – ideally software tools automate the process.

Here is an example:




    The following diagrams show a hotspot defined by a 3D rectangle (illustrated by the translucent blue rectangle).
    The size of this hotspot is defined by the angles γ and ε.
    It is positioned in 3d thanks to the two angles α and β.
    It is possible to turn this rectangle around axis OC.  (This angle is not shown on the diagrams in order to keep them simple.  In the following example, this angle is considered to be null.)




Advantages:

Disadvantages:


    A hotspot can be defined on appoint using just two angles (α and β). The sensitivity of the hotspot (i.e.how close the user must click to this point in order to activate the hotspot) is defined by the viewer.


Advantages:

Disadvantages:

III. Sounds

        Sounds are not often used in panoramas or virtual tours.  However, they can make the panorama come to life. 
    Two types of sound can be defined: 
    2D sounds are generally used to give a global environment to the panorama.
    They require only few parameters: 

    Volume is not used here to control the total volume of the sounds in the panoramic scene, but to control the relative volume of different sounds

    Spacial sounds are used even less than 2D sounds in panoramic imaging, but they can make a panorama seem even more ‘alive’ in that they appear to interact with the user.

    The parameters used are exactly the same as those for 2D sound, but with two more angles (α and β) which make it possible to position  the sound in 3D. (see diagram).

IV. Light effects

       Light effects were introduced by ImmerVision.  Like 3D sounds, they make it possible to make panoramas more dynamic, and add a feeling of reality by simulating the effects of real light sources.  The user gets the impression of looking around a non-static environment.

    A light is positioned simply using two angles (α and β),  as shown on the following diagram:



    There are two types of light effects, which can be used simultaneously: 


    Dazzle simulates the effect we perceive have when we look at the sun with the naked eye or with a camera and are momentarily “blinded” by the intensity of the light.


    Lens flare is the aureoles, which appear when a multiple-element optical system has a source of intense light in its field.


    

This effect is obtained with using an image that defines the series of optical artefacts that simulate the refraction of light within a lens.  The image below is an example:

       This means that a user can easily define his or her own lens flare using some very simple parameters.

V. Other useful data

    Various useful information can be added to the panorama such as: 


        In the majority of panoramic file formats there is no means of specifying copyright ownership or licence terms.  However, this information is important, because it makes it possible for the author to assert ownership and the terms for the use and the redistribution of the work

        Description simply gives more information about the panorama.  This can be a description of the scene, conditions of the shoot, etc.... 

    GPS co-ordinates give the geographical position of the panorama as well as the date and time when the photos were taken. 
They can also be used to automatically position a point on a map.

    Although we mean magnetic north, the north direction can be an arbitrary direction other than north.
    By indicating the north direction in each panorama in a virtual tour, it is possible to keep the same direction of view when we pass from one panorama to another (without having to separately specify “leaving” and “entry” points for hotspots).  This can also provide a more realistic virtual tour experience for the end user.  Moreover, it makes it possible to automatically control the direction of a compass whose role would be to show the direction of view on a map (a good example of this can be seen in Poor Man’s VR (PMVR) at http://www.duckware.com/).

        The default point of view is given by the value of the zoom, horizontal and vertical rotations.  It makes it possible to specify the initial view when the panorama is first loaded.  This entry point is not used when a person uses a hotspots in a virtual visit to pass from a panorama to another.


        Movement limits make it possible to limit horizontal and vertical panning within the viewer as well as the zoom.  Limit on zoom is typically used to prevent the end user zooming in beyond the actual resolution of the image (which would only result in a pixelated view).

VI. Scripting language

    Currently, several viewers of panoramic images support basic script languages.
    OpenPanorama will support its own simple script language allowing the control of and interaction with the various objects in the panorama.  However compatibility with complex languages like AppleScript® used by QuickTime® will not be provided because of their complexity.

VII. OpenPanorama image format

    Why create an OpenPanorama image format? 
    There are, at least, two reasons to the use of such an image format: 

    To better understand the goal of this functionality, let us take again the cylindrical example of panoramas used by QuickTime®. The original images are organized as shown below:
    For use in an OpenPanorama panorama, we want the images to be organized in the following form:

    Thanks to the OpenPanorama image XML specification it's simple to give the size of the image we want, as well as the way to position each image of figure 1 inside.  That is done simply with using source and destination rectangles.

    In this way a segmented or “tiled” image can be treated as if it is a single image, without having to modify the original images, or recompress it (Note that the majority of the current systems of compression are lossy, and recompression decreases the quality of the image considerably).

    This principle can also be applied to cubic, and even spherical image mappings.



    Today, the most widely used image formats for panoramas are GIF and JPEG, because they are natively supported by Java. However, these image formats have limited functionality: 

        It should be possible to use these traditional file formats (or others including video formats) to create images with more capabilities.
    For example, it is possible to apply a variable transparency to a JPEG image using a simple gray level mask.


    This is a simple example of what such a system would allow.  Note, it would be possible to optimize the size of the panorama file by compressing the mask more heavily than the panoramic image. (This is because a mask is unlikely to need to be as good quality as the actual panorama.)

    We also envisage creating animations using several images, adding a transparency mask to a video, etc....