File types for archiving and dissemination
Archival formats
As with other data types, the most stable formats for preserving 3D date are openly documented, text-based file formats which allow access to data independent of specific software (see Planning for the Creation of Digital Data). For the majority of 3D models the OBJ and PLY formats possess the ability to preserve the geometry and visual surface properties of a 3D object, but they are not suitable for complex scenes with light sources, animation, or complex interactivity. For more complex 3D datasets and visualisations the COLLADA and X3D formats are recommended. Generally not suitable for long-term storage are software specific or binary formats, such as 3DS, MAX, SKP, or BLEND.
If an export function is available, it is advisable to convert data to an archival format in the application in which it was originally created. Where this isn’t an option – e.g. the desired format is not supported by the original software – then the file may have to be converted into an intermediary format which can then be converted to the target format using additional software. For example, the conversion of a 3D object created in Blender[1] to a 3D-PDF for dissemination could be accomplished via an intermediate export to OBJ from Blender and then a successive import and export to U3D in MeshLab.[2]
The Conversion Software Registry[3] enables the discovery of conversion programs for various formats, including intermediary formats. For the conversion of large, complex 3D content there are also specialised commercial programs available such as Polytrans and NuGraph from Okino.[4]
In certain circumstances where data migration proves difficult it is also advised that the original formats be retained and that source files (textures, visualisations, etc.) or problematic elements of a model be captured separately and stored in a suitable archival format.
It is also advised that additional images and/or video files are captured. Such files allow a convenient preview or overview of the model and may preserve elements of ‘look and feel’ that may be difficult to retain in the available preservation formats.
Dissemination
In addition to archival formats, 3D models can be easily disseminated as PDF files. The PDF format supports models in the U3D and PRC formats and can additionally integrate text, images, and links alongside the 3D data. 3D PDF files can be viewed in the standard freely-downloadable Adobe Reader (although currently unavailable for Linux systems) and users can apply measurement and annotation tools in order to measure distances, radii and angles. While 3D PDF files provide a convenient, self-contained format for data dissemination, file sizes may be large and the models often require a large amount of RAM, which may not be available on every system. As a result, the 3D PDF format is better used to disseminate lower resolution or down sampled versions of 3D models with full resolution data being made available in other formats.
Other solutions also exist if direct dissemination of 3D content on the web is required. The first is a commercial system, Sketchfab[5], which supports the upload, publication and visualisation of 3D models on the web. Sketchfab supports inclusion of 3D content in standard web pages and on social media; it already has a very large community of users and is becoming the de-facto standard for the publication of 3D content on the web.
A second option is the academic open-source platform, 3DHOP (3D Heritage On-line Presenter)[6]. 3DHOP is much more flexible than Sketchfab, supporting different presentation layouts and interaction modes, and allows expert users to make changes and configurations. It adopts an efficient multi-resolution format (Nexus) that allows the online publication of full-resolution models via 3D sampling technologies.
A third option is another open-source project, Aton[7]. Aton is based on the same open-source library as SketchFab, focusing on scene-graph concepts and extending capabilities to multi-resolution and LOD, in order to present and visualise complex 3D datasets such as large terrains. The front-end provides support for mobile browsers and multi-touch devices, and offers several options for camera manipulation, spherical panorama support, rich annotation and immersive VR.
[2] http://meshlab.sourceforge.net
[3] http://isda.ncsa.uiuc.edu/NARA/csrAbout.html
[7] http://osiris.itabc.cnr.it/scenebaker/index.php/projects/aton/
