Problem with silhouette formation using Kinect

My aim is to form silhouettes of users detected by Kinect, using particles. Initially I tried to map the joint coordinates given by Kinect to a model and tried to map particles on this model. But this resulted in a bad interaction because the side interaction is not smooth enough in Kinect. Now the only way to achieve this is by using the depth image details from kinect. I used the depth values from Kinect, converted them to camera space coordinates thus getting the point cloud. This point cloud has to be mapped to particles. This is possible in two ways. One way is by using Particle system on each point I get from point clouds by Kinect. When I try to map each point to a particle system, there are a lot of points and this results in a huge lag when the game runs. The other way is to get points from point cloud and map those vertices to particles using shaders and compute shaders. Here the issue is I am not able to control the size of particles using shaders. Is there any way to control particle size in shaders so that I can use this method which also reduces the computation effort of CPU. I would also be happy to know if there is any other way to form silhouette out of particles.