Share |

Distributed rendering

Complex animations and computer graphics, a field long driven by gamers and movie makers, requires computationally expensive ‘rendering’. This involves generating a single image from a model that contains information about an object – its geometry, viewpoint, lighting, etc.

As well as being used to create computer animated movies and shows, such as Shrek, The Penguins of Madagascar, and Happy Feet, rendering is also useful for areas such as medicine and architecture.

This short video, with Blender Foundation’s beloved open source character Big Buck Bunny acting as host, explains what rendering is and why distributed rendering is so useful.

The rendering of complex computer generated imagery is often carried out through high-performance computing, but it is by its nature a highly parallelizable activity and so can be done through volunteer computing.

Renderfarm.fi is currently the world's most popular volunteer computing-based rendering service. It uses the BURP (Big and Ugly Rendering Project) open source middleware software together with BOINC in order to distribute rendering over the Internet.

Renderfarm.fi is run in Espoo, Finland by the Laurea University of Applied Sciences, a member of the International Desktop Grid Federation. The project has so far gathered over 5,000 registered users who are at the time of writing sharing some 2,500-3,000 CPU cores between each other. Renderfarm.fi was recently awarded the UN backed World Summit Award in the category of E-science and Technology.

Renderfarm.fi is distinguishable from many currently run BOINC projects in two main ways. First of all, Renderfarm.fi provides an openly available interface through which registered users can themselves commit clusters of work (called ‘sessions’) to be rendered by their peers.

Second, unlike BOINC projects that focus on using scavenged computing cycles for scientific computations useful solely to researchers, Renderfarm.fi by its nature can offer its participants regular visual feedback in the form of rendered images and animations.

Renderfarm.fi is inherently open by design, which means that current technical limitations imply that model and texture data will be exposed. This openness can be seen as both a limitation and a feature of any such rendering service.

For example, it is likely to be seen as a limitation for commercial usage such as movies with large budgets, but as an asset for projects that aim for maximum visibility, such as low-budget movies, commercials and product endorsements.

Interestingly, the adoption of Creative Commons licenses, which potentially enable the use of volunteer computing resources for commercial rendering, has not raised alarm among volunteers nor decreased interest towards Renderfarm.fi. This would seem to indicate that as long as volunteers feel empowered, they do not mind participating in commercial works.

Volunteer computing projects often exhibit a top-down model for job submission traditionally witnessed in grid computing. While enabling the donation of computing resources, this does not offer volunteers the possibility to submit jobs themselves.

By adopting the bottom-up model, Renderfarm.fi is making it possible for its users to benefit from the volunteered computing power of their peers. For the writer it appears accessibility is a key tackling point in the process of 'democratizing' distributed computing, that is when making the technology itself available to wider audiences, independent of their technical understanding, economical situation or geographic location.

In the case of Renderfarm.fi, we have solved the issue of accessibility by supporting Blender, the world's most popular open source 3D suite. We have integrated a feature we call the "uploader" into the software (available in trunk since version 2.53). The uploader enables users to send work to the service directly from within Blender and helps them go through the steps necessary for a successful render on a complex distributed system.

- Julius Tuomisto, Laurea University of Applied Sciences

Your rating: None Average: 4 (11 votes)

Comments

Post new comment

By submitting this form, you accept the Mollom privacy policy.