Generate watercolor-like paintings

Dyamic SMC can be used to generate watercolor-like paintings. One such example is shown below.

# After necessary imports
# Read image and convert to density
image = imread('lady_painting.jpg')
image_dens = 255.0 - image.T
image_dens = np.fliplr(image_dens)

# Define probability distribution
xmin, ymin = (0, 0)
xmax, ymax = image_dens.shape
prob_dist = ProbDist(xmin=xmin, xmax=xmax, ymin=ymin, ymax=ymax, Nx=xmax, Ny=ymax)
prob_dist.set_prob_dist_from_array(image_dens)

# setting up simulation parameters
n_agents = 20
n_time_steps = 10000
paint_rate = sum(sum(image_dens)) / float(n_time_steps * n_agents)

# initializing paint density array
paint_dens = np.zeros(image_dens.shape)

# Define DynamicSMC coverage object
dynamic_smc = DynamicSMC(prob_dist)

# add agents to coverage object
for _ in range(n_agents):
    dynamic_smc.add_agent(Agent(xmin + random.random() * (xmax-xmin),
    				ymin + random.random() * (ymax-ymin)))

# Run the algorithm 
for time_ind in range(n_time_steps):

    dynamic_smc.time_steps(1, 1.0)

    # code here to use current agent locations for updating 
    # paint density array and display

Below is shown an animation of the evolving painting. To see the full code for this example, look here. The original painting used for this experiment is this: lady_painting . More complex and color paintings can be generated using Dynamic SMC as seen in this paper.

Evolving painting	Final painting