New datasets are always welcome for Science On a Sphere®! Here are a few simple guidelines to assist in the creation of content. The SOS team has significant experience with technical and aesthetic issues when creating content for spherical display systems, and is happy to provide feedback and suggestions during production to ensure quality content.
The Content Creation Guidelines are intended for anyone who is creating new datasets for Science On a Sphere®, whether or not you have access to a Sphere.
This documentation covers the basics of how to create a new dataset for Science On a Sphere®. More details about dataset creation can be found in the Datasets Manual, and information on creating presentation playlists from datasets is available in the Presentation Manual.
The map projection used is the Equatorial Cylindrical Equidistant projection (wikipedia). This is a simple latitude/longitude grid with the lines evenly spaced where the image is twice as wide as it is tall. To be consistent with the SOS Data Catalog, it is recommended that the 0° line go through the middle of the map, with the +/-180° lines at the edges. In order for the data to wrap properly around the sphere, it is imperative that you follow the specifications for the data format closely. Images in the wrong format will project on the sphere, but they will not correctly represent the size of the continents.
Textures can be displayed on the sphere and rotate around the poles of the sphere, as in planetary rotation. A good example of a texture is Mars. In the Mars folder you will find just one image, named for its resolution, which is projected on the sphere. Often, the textures are available in several different resolutions. As the resolution increases, so does the loading time on the sphere. Textures can be rotated in any way using the functions either in the Control menu of the SOS Stream GUI or in the remote, and by default rotate around the poles of the sphere. Textures will rotate indefinitely until pause is pressed or another dataset is loaded.
SOS will accept most common formats (JPEG, PNG, GIF, TIFF, etc.), but JPEG and PNG are preferred for images and image sequences. In addition to pointing to files that are local on your computer, you can also use URLs, such as “http://example.com/image.jpg”.
There are several different ways that time series work. In its raw form a time series is an image sequence. SOS will display in sequential order all the images in a single directory. The images can also be converted to an MPEG4. This is the preferred format for showing time series because the system can play the MPEG4 files at a higher resolution and faster frame rate than the image sequences and they take up less disk space. There is no limit to the length of a time series.
Time series by default don’t rotate, but can also be set to rotate while they are animating through a feature in the playlist and a button on the remote. Transitions, special effects, and other computer graphics techniques can be added to a sequence through the use of off the shelf software like Final Cut Pro. Time series will loop indefinitely until pause is pressed or another dataset is loaded.
Render the video with the MPEG4 video codec at a minimum of 25 mbps. Just because a file has a .mp4 extension does not mean it will play perfectly on SOS. Be sure to check the codec that was used to render the file. The H.264 codec should not be used because it can cause errors in the SOS software.
The software also has the ability to display KML (Keyhole Markup Language) files and load images directly from an Open GeoSpatial Consortium (OGC) Web Mapping Service (WMS) on Science On a Sphere. The Datasets Manual contains more details about these formats.
The recommended resolutions are 2048x1024 for time series and 4096x2048 for textures, though other 2x1 resolutions, such as 3000x1500 will work. Higher resolutions are possible for the animations, but not all SOS systems in the SOS Users Collaborative Network are able to animate at 30 frames per second for higher resolutions. Make sure to render the videos with square (1:1) pixels for proper playback. If possible, the highest resolution of the animation that is generated should also be provided, in addition to the 2048x1024 file, so that it is available for new and upgraded sites that have the ability to play them. If an image sequence is provided, it will be rendered into a MPEG4 file that is 2048x1024 unless there is a specific reason to keep the animation as an image sequence. For single images, resolutions above 4096x2048 are possible, though load time increases with resolution size.
As of release 5.0, the SOS software is capable of working with higher resolution 4K projectors. If you wish to create higher resolution images and movies to take advantage of the higher resolution quality of 4K projection, the same 2:1 ratio rules apply. Movies with 4096x2048 resolutions will work on such systems, although decreased FPS rates may be necessary and the load time will be longer. Single images of 8192x4096 will display well on 4K projectors and images as large as 16384x8192 should still work, although resolutions above 8192x4096 will result in only slightly increased quality. We recommend that you test any high resolution images and movies you create on a full 4K projector system to ensure they will display as you expect.
The audio player for SOS is fairly versatile and most common formats will work with SOS, such as mp3, mpeg4, wav, ogg, and aif. The audio file can either be a standalone file or embedded in the same MPEG4 file as the video.
The maps created for SOS should be centered on the Prime Meridian, so that 0°N,0°E is the center of the image, as seen below left. The map on the next page is from the dataset in the extras category called SOS Coordinate System that is useful when learning how datasets are loaded on the sphere. The center of the map loads between projectors three and four with the edges of the map lining up between projectors one and two, as seen in the diagram to the right of the map.
It is important that the data fill the entire image space. If there are borders or extra space around the edges of the image then a seam will appear on the sphere as a vertical white line with spots on the poles. It is also important that the data match on either side of the edges of the image, otherwise you’ll end up with a discontinuity running vertically along the visualization.
In addition to making sure that the data fills the entire frame, from 180° West to 180° East, also make sure that the data fills the entire frame from 90° South to 90° North. If there is missing data at the poles, fill in the area with a solid color or a basic land/ocean background to ensure that the dataset wraps properly around the sphere without stretching vertically.
When working with a spherical surface, warping is always something to consider. The least amount of warping occurs near the equator, while the most warping occurs at the poles. Because of this, it is recommended that any text and labels are placed near the equator. Supplementary text, labels and images that are displayed as PIPs don’t warp if their position is set with the pipcoords attribute in the playlist. Datasets can be tested for warping issues using CC Sphere in Adobe After Effects or 3D Sphere in Photoshop. There are some plugins for After Effects such as Cycore Effect’s Sphere Utilities that can also help with spherical warping. In addition, these programs can be used to check for seams.
A color scale can dramatically change the emphasis and message of a dataset. Because of this, the Science On a Sphere Users Collaborative Network has had many discussions on the color scales that are used for SOS datasets. The goal is to create datasets with well-chosen color scales that are meaningful, intuitive, and scientifically accurate. Several conclusions are the result of these discussions:
It can be confusing to users when the same color scheme and their associated color bars are used for two completely unrelated datasets
The same color should not be used to represent more than one thing. i.e. if ice is shaded white, then white should not also be used for areas of missing data
Using rainbow-colored legends and color schemes is often confusing to the audience and hard to parse. Instead, consider using shades of green to represent phytoplankton, and blue and red gradations to represent temperature anomalies. i.e. use “meaningful” colors
Avoid using full sphere backgrounds that are completely or pre-dominantly white. The seams between projectors become more apparent when using solid white/bright backgrounds. f using bright background colors, consider adding some noise/texture to them. Same goes for PIPs, especially those displayed at the seams between projectors
ColorBrewer is a good resource for helping you choose a color scheme for your visualization.
You can animate a time series at any rate, but 30 frames per second is the recommended speed. We try to create our time series so that they look smooth and animate well at 30 fps. The frame rate is sometimes limited based on the time resolution of the data and the type of data. It is important to keep this in mind when creating a time series so that you make enough images to ensure that the dataset plays for a reasonable length. If you only make 30 images, then it will only take one second to loop through the dataset at 30 frames per second. The optimal playback speed is chosen based on the number of frames and the degree of change between each frame in the sequence. To get smooth time series, the changes between each frame should be small and the playback speed high. If a time series is coarse, then it might animate better at a slower frame rate such as 10–15 fps.
Single images are typically named for their resolution, such as “4096.jpg”. Image sequences are kept in folders that are named for their resolution, and the images themselves should be named to sort in ascending order from earliest to latest. This can either be done with a time stamp in the file name, or a frame number in the file name with a sufficient number of leading zeros in ensure proper sorting, shown in the example below. Videos should be named based on content and resolution, such as hurricanes_2048.mp4. By including the resolution in the file names, SOS users are able to easily determine what is available and appropriate for their system. The SOS software does not handle spaces and special characters in file names well. Do not use spaces and special characters in file names!
Labels and colorbars are important for providing context. Labels and color bars can be in the frames or projected on top of them externally. It is recommended that you do not add the labels and color bars directly to the frames that you create. By keeping them as external images, you have much more flexibility with their size and position within the playlist.sos file. If you do choose to put your labels and colorbars directly on the images that you create (we call this burning them in), make sure that you make them big enough so that they are legible on the sphere and that they are far enough from the poles that they don’t get too warped.
One of the nice things about keeping the labels and color bars external is that they don’t move as you rotate a dataset. They stay in the same position relative to the projectors. Labels and color bars that are part of the frame rotate with the frame, which can cause viewing trouble for the audience as you move the sphere about.
SOS version 5.3 and after: Use the SOS Visual Playlist Editor to set the label position to a non default value, and use the Label Editor to fully customize the format (color, font, etc) of label text.
SOS version 5.2 and before: Within the playlist.sos file you can set the position using the labelposition attribute, which is set by the x and y position as a pair of coordinates (x,y). Both x and y can vary from -1 to 1. The default position is (-0.3, -0.5). The label color can be changed with the labelColor attribute which can be set to R, G, B, Alpha, (or the symbolic names: white, black, red, green, blue, …). The default color for the labels is white. While you can change the position and color, you cannot change the size or font.
For every dataset that needs a timestamp, a simple text file called labels.txt should be generated that contains one line for each frame in the animation. A labels.txt file cannot be used with a single image. If you have labels for a times series that contains 2000 frames, then you need a text file that has 2000 lines. In the playlist.sos file, if label = default is included, then the image file names appear as the labels on the sphere. The labels file should be stored in the dataset folder and is typically named labels.txt. The labels usually contain the date and maybe a title.
If you don’t want to include the title in your labels file, then you can make an image of title that you can include as a Picture in a Picture. This allows you to choose the font and color scheme of your choosing when you make the title image. This is also a nice option because then you don’t need to insert the title into every line of your labels file.
Alternatively, you can directly create a title—or any text in general—on SOS by using a Text PIP. In addition to using the labels.txt file, there are other ways of labeling the content that is on the sphere. Colorbars and legends can be added using the Picture in a Picture feature. As with the timestamps, do not “burn in” colorbars and legends. Make sure to test the size and fonts of colorbars and legends on the sphere to ensure legibility. Most colorbars and legends are generally too small when originally displayed on SOS and have to be increased in size. Also, make sure to test the size to ensure that the colorbars aren’t so large that they hinder the audience from seeing the underlying dataset.
The purpose of adding labels to the sphere is to aid the visitor in understanding the dataset. Consider using pictographs for scales because they have been found to be intuitive and beneficial for visitor understanding. Also, using country or city labels can help visitors orient themselves in a global context. Another suggestion is to use vertical temperature colorbars because that is how most visitors are accustomed to reading thermometers. Size, orientation and placement of colorbars and legends are important for improving the visitor’s understanding of the dataset.
There is a lot of flexibility with the color bars. They are inserted into the playlist.sos file as a pip. Using this function, you can not only set the position, size and transparency, but also when the color bar appears, how long it stays visible, and how quickly it fades in and out. The color bars can be any common image format such as GIF, JPEG, PNG, TIF, etc. Color bars are generally named color_bar in order to keep all of the various images in the file separate.
Captions can be displayed on SOS from standard SupRip Subtitle (SRT) files that are read in to SOS directly. An SRT file is an industry standard file that contains a sequential set of subtitles with their start and end timecodes. Captions provide text for hard of hearing (the left image below) or non-English speaking audiences (the right image below) and can also be used to show movies in AutoRun with the sound muted. Text formatting may also be fully customized in the SOS Visual Playlist Editor. NOAA datasets will be getting closed captions added to them for your use, but will not be shown by default. The captions may then be turned on in the Remote App or displayed all the time by setting the visibility in your presentations.
Every dataset that is added by the user to the SOS Data Catalog should be in its own individual folder in the directory /shared/sos/media/site-custom. Do not put your site-created content into the folders created by NOAA.
Complementary or similar datasets can be grouped together in a folder that contains individual folders for each dataset. There are many content elements that can be used to create a dataset. For the NOAA-provided datasets you will find the following elements in each folder (not all of these elements are available for each dataset):
An equatorial cylindrical equidistant JPEG or PNG file named for resolution
A folder with an equatorial cylindrical equidistant image sequence named for the resolution of the images
An equatorial cylindrical equidistant video (.mp4) of the data
Text file labeled labels.txt
Text file labeled playlist.sos
Text file in the SubRip Subtilte (.srt) file format
Color bars and other supporting images
Media folder with thumbnails, videos, and supporting documents
In the Aerosol datasets example (a NOAA provided dataset), there are three datasets that are related and all from the same source. To keep them together, a folder was created in the atmosphere category called “aerosol”. In the aerosol folder, each dataset was given its own folder: sulfate, blackcarbon, and blackcarbon_and_sulfate. Notice that there are no spaces in the names! In each of the individual dataset folders there is an image sequence named for the resolution of the frames, an MPEG4 video, a media folder with thumbnails, a colorbar, labels, and a playlist.sos.
A uniform naming convention has been used among the folders in the NOAA provided directories. Images that are projected onto the sphere are named for their resolution, movies that are projected onto the sphere are named for their dataset name and resolution, all labels are named labels.txt, etc. his has been done to make it easy for the user to know what is available in each folder. This naming convention doesn’t need to be used for site-custom datasets.
The only two elements that are required to have a dataset are the playlist.sos file and something to be displayed on the sphere, either a single image, an image sequence, or a MP4 video file. All of the other elements are optional.
Each NOAA provided dataset has a “media” folder that contains a rendered global view of the dataset as well as two thumbnails. These thumbnails are used in the SOS Stream GUI as well as the iPad. The media folder is not required for site-custom datasets, but if you don’t provide a thumbnail, then a question mark icon will show up in the software. In order for custom thumbnails to show up in the software for your datasets, create a folder in your dataset folder called media. The media folder should contain two images in the .jpg format with the following names and resolutions:
Because Science On a Sphere® uses common image and video formats, you can use many tools to create and edit datasets.
Some of the common tools used are Photoshop, FinalCut Pro, ImageMagick, GIMP, etc. You can use whatever you have available and are comfortable using. A program like FinalCut Pro can be used to add transitions, special effects and other computer graphic techniques. At a higher level, tools like IDL, AWIPS, McIDAS, and other image analysis applications are typically used to create imagery from scientific datasets. As an example, we have used AWIPS (Advanced Weather Information Processing System) to create images from numerical forecast models. A graphics designer can use a 3D modeling application, such as 3D Studio, to create advanced visualizations for SOS.
If you have access to a Science On a Sphere®, you can also use the Visual Playlist Editor to create new datasets.
There are many settings that can be included in the playlist.sos file. For a dataset added by a site into the site-custom folder, the playlist.sos file should contain the following elements: name, data, creator, subcategory, keywords, and description. Only name and data are required, but the other settings provide helpful information and allow you to more easily find your dataset.
There is no need to add a category because all datasets in the site-custom folder are automatically tagged for the site-custom category. In order for a new dataset to show up in the site-custom library on the local SOS computer, make sure to update the library by clicking LibraryUpdate Library… in the SOS Stream GUI. After you have updated the library, you have to click the Update Now button in the Settings tab of the iPad in order for it to show up on the iPad. It will not appear on the iPad if it is not stored in the site-custom folder on the computer. In the Data Catalog tab of the iPad you will be able to find the new dataset in the subcategory that you assigned under the site-custom main category. You can create as many different subcategories as you like within site-custom.
We have viewed many datasets and are happy to review your site-custom datasets and provide feedback as desired. Particularly for datasets that will be submitted for consideration for the SOS Catalog, we suggest providing a draft version to the SOS team for feedback.
If you decide to submit a dataset for consideration to be part of the SOS catalog, we will need all the content pieces and the playlist.sos file, along with a written description of the dataset, a list of notable features, and the credits for the dataset. All of this documentation is used to create an entry in the SOS Dataset Catalog. The written description should be a simple overview of the dataset that highlights the source of the data, whether it is modeled or measured, what it shows, and why it’s important. It should be a non-technical description that is easily understood. The “Notable Features” is a bulleted summary of the highlights from the description that presenters can use when showing the dataset to viewers. The credits are listed after the description and should include:
Dataset Visualization Developer
These can be the same for each listing, or all different and can include links to the original sources. For examples, visit the SOS Dataset Catalog.