Masters Project - Online collaborative 3d art

  • Research

I worked on this project with Kurt Luther under the guidance of Amy Bruckman at Georgia Tech.  It was based on Kurt's previous work looking at collaboration in amateur Flash animations at Newgrounds and several other websites.  Kurt found that collaborative Flash animations tended to be divided up so each person created a segment of some length to fit into a story or theme and then one person—usually the person that came up with the idea—would compile the segments and post it.  Most of these group projects are never completed, many of them abandoned when the only step that remains is compiling the segments because the "director," or the person compiling them, has lost interest in the project and no one else has access to them or is willing to step on the director's toes.

For this project, we were interested in looking at how this plays out in amateur 3d artwork.  On the one hand, 3d is more difficult in general, requiring many different types of specialties; on the other, it lends itself to division of labor into modelers, texturers, animators, etc. so each person can be involved throughout the animation and everyone has access to the entire thing.  Futhermore, if the project falls through, models, textures, and scenes can still be reused, so people might be less wary of joining these projects and wasting their time if nothing came of them.

My expectations did not exactly hold true: projects were abandoned just as often as the flash projects, if not more so, often at very advanced stages.  The division of labor, which might have saved some of them, turned out to be a burden, as it could be difficult to find people with the right set of skills.  Models were sometimes reused, but more often they were not, especially if participants were turned off of collaboration entirely by the difficulty of their experiences.  However, some sorts of projects were more successful than others, and some types of organization seemed to lend themselves to success more than others; I will break that down further below.

Method

To gather data, I read public discussions on various art message boards and conducted phone and email interviews of people who had participated in 3d art collaborations online.  Interviews were transcribed and coded using a bottom-up method.  Interview participants were recruited from discussions on the message boards and from surveys about online art collaboration.  Most of the participants also participated in several other CG communities in addition to the ones I looked at.

Summary of results

There were several different “genres” of projects that people participated in, the most common that we found in sites and interviews being animations, virtual environments (including games but also non-game 3d spaces that someone could navigate through), still images, and 3d models. There was some overlap, but mostly participants were involved in projects in just one of those areas.  These projects varied greatly in length.  Simpler projects like stills and models might be finished in a few weeks, but some of the animations and environments had been in progress for several years.  Like the Flash animations, many of these projects, especially the longer-running ones, were never completed.

To give you an idea of some of the projects I looked at, a video of one of the virtual environments is embedded below.  It's a model of Minas Tirith from Lord of the Rings that viewers can navigate much they way they might in a first person shooter game.  Individual participants chose buildings to model and they were inserted into the full city model.  The project reached a really amazing stage of completion before it fell apart.  This is one of the projects that I was only able to analyze by forum thread.

Different types of projects have different types of social organization. Animations in particular are often highly organized by role, with some people dedicated to modeling, others animating the models, others doing lighting and texturing, etc., much the same way that a professional animations studio might organize itself.  The techniques involved in animation are often very specialized and complex, so it can be difficult to master more than one, and a collaborator may be put into one role and have very little to do with other parts of the creation process.  The organization of these projects may have been influenced by the heavy involvement of professionals and students who are used to this sort of division of labor, relative to other types of project.

Other projects had a more ad-hoc organization, in particular the 3d environments, where collaboration members may each take a building or a prop, build it on their own with feedback from other members on how to improve it and make it fit with the rest, and send it off to someone—or several people—who collects all the parts together into a single scene.  And though the project may never really be finished, as some of them have no definite end-point in mind when the project begins and members may keep making parts until everyone loses interest, the parts of the project can stand on their own so that there is a product that can be shown off in addition to the other benefits of collaboration. Members post their own finished pieces in their personal art galleries, partly in an attempt to drum up interest in the project but also to showcase their own work.

Projects could be paid or unpaid; in paid projects, collaborators might be promised a certainamount of money at the beginning of the project, paid throughout, or offered money should thecompleted project become lucrative in some way; other projects have no discussion of monetary incentives at all. Paid and unpaid collaborations seem to be distributed fairly evenly throughout the different genres.  Participants in unpaid projects emphasized benefits like learning from others and meeting new people, but these were downplayed in favor of working with artists of similar skill levels and producing consistent, high-quality results.  In many cases, the quality of the results mattered less in the unpaid projects, where participants were involved more for the experience, variety, and pushing themselves outside their comfort zones.

But although participants in unpaid collaborations emphasized benefits of the process rather than the end result, many were still wary about joining collaborations that might never be completed.  Almost all of the participants who had worked on multiple collaborations talked about wanting to work with people they knew they could count on to see the project through.  Like the Flash projects, these had fairly high attrition rates as people lost interest as the project dragged on.  One of the participants, a leader of a 3-year project, said the biggest challenge was maintaining his own and others' enthusiasm throughout.

Attrition can kill a project in a couple of different ways.  Several participants told me that when others are posting frequent updates is easy to be enthusiastic, but when one member starts slowing down then everyone starts losing interest and slowing down and the project grinds to a halt and dies.  The project may also die if the member is irreplaceable.  This is especially true of projects divided by role into modelers, animators, etc., where it may be difficult to find another person that knows how to animate well enough to continue.  However, even ad-hoc projects will usually fall apart if the "director," the one responsible for the original vision, organization, and creative drive, drops out.  Regardless of the division of lavor, small projects like still images and single models are most likely to be completed not only because they are less difficult but because they are completed quickly enough that group members are unlikely to lose interest.  Animation projects, the longest running projects and most subject to member attrition over time, almost always tend to be divided by role and thus suffer a doubly by being more likely to lose members overall and more likely to lose members who cannot be replaced.

I had theorized that attrition might be less of a problem in the 3d projects than in the Flash animation, which often fell apart when the project leader disappeared and no one else had access to all the files.  However, the group leader is also an important creative voice, organizer, and motivator for other members.  So even if everyone has all the files in a project that is divided by roles, losing the leader is still a major problem.  I had also underestimated the difficulty of finding replacements.  In the flash animations, anyone can do what anyone else is doing, and often the segments aren't dependent on each other and can be left out; however, participants in 3d animation stressed the difficulty of getting people involved at all, let alone someone with just the right set of knowledge.

3d collaborations also face technical problems.  The most common methods for group members to communicate were forum threads and email.  With the exception of the participants who created 3d models for sale and preferred to keep development secret, participants preferred the forum thread as a way to publicly showcase their work and get others interested in joining the project.  But asynchronous communication can be slow, especially if everyone is in different time zones, and several participants, especially those who also create 3d art professionally, were frustrated with how long it took to get anything done.  Others didn't mind because they got to coordinate with people all over the world.  Instant messaging was also common, but other synchronous communication methods like Skype were rare because of the difficulty and cost in getting them set up and the latency involved over slow networks.  Forum discussions typically were single threads within a more general collaboration forum and could grow to an unwieldy several hundred pages long.  A couple projects had their own dedicated forums, but those are harder to for the casual observer to stumble upon and require technical skill to set up and someone to donate potentially expensive web hosting to the project.

Email was typically used for communication that did not need to be public and to share files, since email attachment size is usually less limiting than forum attachment size.  Some participants had shared hosting and one even used a versioning system, but most did not have access to those and wished for some shared central file repository that could keep track of versions, upload date, creator, etc.

The research discussed here was conducted in 2007 and 2008.  I would actually be very interested in seeing a project like this with something like Google Wave if it were open to everyone and didn't require registration to view.  I think the long conversations that take place in the threads would be much less unwieldy there, where new "wavelets" can be spawned on subtopics like some particular model, and new members and casual observers can see the discussion replayed as it happened.  Being both synchronous and asynchronous and as public or private as you want it, it satisfies a number of different needs that the participants had of different communication media.  To the basic facilities that Wave provides, I'd like to add a widget that can compile all public images and files uploaded to the wave and linked wavelets into a tagged or categorized list to make it easier for people to find the file or reference image they need and show progress and rendered images to inspire members new and old.