I am trying to produce shadows in a captured image in the Pybullet simulator. The shadows show up in the debugGUI mode but are not visible when calling p.getCameraImage
even with shadow=True
.
Something else I noticed is when shadow=True
in p.getCameraImage
there is a large shadow as if there is a tall object blocking the light rather than a small shadow corresponding to the object.
The issue persists whether connected to direct (p.connect(p.DIRECT)
) or to GUI (p.connect(p.GUI)
) as suggested here: https://github.com/bulletphysics/bullet3/issues/2610 and I have attached a minimum code sample below needed to reproduce the issue.
import pybullet as p
import pybullet_data
from PIL import Image
import math
physicsClient = p.connect(p.DIRECT)
p.setAdditionalSearchPath(pybullet_data.getDataPath())
p.configureDebugVisualizer(p.COV_ENABLE_SHADOWS, 1) # Shadows on/off
planeId = p.loadURDF("plane.urdf")
p.loadURDF("sphere_small.urdf",[0,0,0.2])
projectionMatrix = p.computeProjectionMatrixFOV(
fov=90.0,
aspect=16. / 9.,
nearVal=0.1,
farVal=30)
for i in range(360):
viewMatrix = p.computeViewMatrix(
cameraEyePosition=[2*math.cos(math.radians(i)), 2*math.sin(math.radians(i)),2],
cameraTargetPosition=[0.0, 0, 0],
cameraUpVector=[0, 0, 1])
width, height, rgbImg, depthImg, segImg = p.getCameraImage(
width=int(1280),
height=int(720),
viewMatrix=viewMatrix,
projectionMatrix=projectionMatrix,
shadow=True
)
im = Image.fromarray(rgbImg)
im.save('imgs/test/' + str(i) + ".png")