How to make animated clones get mixed in the crowdAugust 16th, 2008 - 5:02 pm ICT by ANI
London, Aug 16 (ANI): Making the virtual crowd look more realistic is a tough job headache for any animator, as audiences easily point out copied characters. Now, a group of researchers has identified tricks to solve this problem.
While studying two types of animated clones like appearance clones that are identical, and motion clones, which appear different but move in the same way, the team led by Rachel McDonnell at Trinity College Dublin, Ireland found that appearance clones were quickly detected.
The scientists found that participants shown a group of 12 static characters, two of which were identical, needed on average just 5.7 seconds to spot the clones.
However, for participants shown a grid of 12 walking mannequin models two of which shared the same walking, it was difficult to spot the clones.
They took more than 60 seconds to identify the clones.
McDonnell suggests that the animators could use just a few walking styles to generate a crowd scene with a low risk that the clones would be detected.
Furthermore, the volunteers were also shown crowds of 20 characters walking out of step, up to 10 of which were motion clones.
Even with 50pct of the crowd sharing the same walking style, they took around 10 seconds to spot an identical pair.
While analysing appearance clones, the researchers found they could almost double the detection time by choosing the relative positions of clones in a crowd.
When diagonally from one another on screen, detection time jumped from around 5 to 10 seconds.
“Appearance clones on the same vertical or horizontal plane were easily detected,” New Scientist quoted McDonnell, as saying.
“That could be important for reconstructing a stadium crowd you”d make sure no clones lie side by side either vertically or horizontally,” she added.
The work was presented at the SIGGRAPH computer graphics conference in Los Angeles this week. (ANI)
Tags: 7 seconds, animator, animators, appearance, audiences, clones, crowd scene, crowds, detection time, horizontal plane, mannequin models, new scientist, relative positions, siggraph computer graphics, stadium crowd, tough job, trinity college dublin, trinity college dublin ireland, virtual crowd, volunteers