def load_basics(self):
np.random.seed(0)
camera, frustum = getcam()
mesh = get_earthmesh(trans=np.array([0,0,5]), rotation = np.array([0,0,0]))
lighting_3channel = LambertianPointLight(
f=mesh.f,
num_verts=len(mesh.v),
light_pos=np.array([-1000,-1000,-1000]),
vc=mesh.vc,
light_color=np.array([1., 1., 1.]))
lighting_1channel = LambertianPointLight(
f=mesh.f,
num_verts=len(mesh.v),
light_pos=np.array([-1000,-1000,-1000]),
vc=mesh.vc.mean(axis=1).reshape((-1,1)),
light_color=np.array([1.]))
bgcolor = np.array([0.,0.,0.])
renderers = [
ColoredRenderer(f=mesh.f, camera=camera, frustum=frustum, bgcolor=bgcolor, num_channels=3),
TexturedRenderer(f=mesh.f, camera=camera, frustum=frustum, texture_image=mesh.texture_image, vt=mesh.vt, ft=mesh.ft, bgcolor=bgcolor),
ColoredRenderer(f=mesh.f, camera=camera, frustum=frustum, bgcolor=bgcolor[0], num_channels=1)]
lightings = {1: lighting_1channel, 3: lighting_3channel}
return mesh, lightings, camera, frustum, renderers
def load_basics(self):
np.random.seed(0)
camera, frustum = getcam()
mesh = get_earthmesh(trans=np.array([0,0,5]), rotation = np.array([0,0,0]))
lighting_3channel = LambertianPointLight(
f=mesh.f,
num_verts=len(mesh.v),
light_pos=np.array([-1000,-1000,-1000]),
vc=mesh.vc,
light_color=np.array([1., 1., 1.]))
lighting_1channel = LambertianPointLight(
f=mesh.f,
num_verts=len(mesh.v),
light_pos=np.array([-1000,-1000,-1000]),
vc=mesh.vc.mean(axis=1).reshape((-1,1)),
light_color=np.array([1.]))
bgcolor = np.array([0.,0.,0.])
renderers = [
ColoredRenderer(f=mesh.f, camera=camera, frustum=frustum, bgcolor=bgcolor, num_channels=3),
TexturedRenderer(f=mesh.f, camera=camera, frustum=frustum, texture_image=mesh.texture_image, vt=mesh.vt, ft=mesh.ft, bgcolor=bgcolor),
ColoredRenderer(f=mesh.f, camera=camera, frustum=frustum, bgcolor=bgcolor[0], num_channels=1)]
lightings = {1: lighting_1channel, 3: lighting_3channel}
return mesh, lightings, camera, frustum, renderers
def test_depth_image(self):
# Create renderer
import chumpy as ch
from renderer import DepthRenderer
rn = DepthRenderer()
# Assign attributes to renderer
from util_tests import get_earthmesh
m = get_earthmesh(trans=ch.array([0,0,0]), rotation=ch.zeros(3))
m.v = m.v * .01
m.v[:,2] += 4
w, h = (320, 240)
from camera import ProjectPoints
rn.camera = ProjectPoints(v=m.v, rt=ch.zeros(3), t=ch.zeros(3), f=ch.array([w,w])/2., c=ch.array([w,h])/2., k=ch.zeros(5))
rn.frustum = {'near': 1., 'far': 10., 'width': w, 'height': h}
rn.set(v=m.v, f=m.f, vc=m.vc*0+1, bgcolor=ch.zeros(3))
# import time
# tm = time.time()
# rn.r
# print 'took %es' % (time.time() - tm)
# print np.min(rn.r.ravel())
# print np.max(rn.r.ravel())
self.assertLess(np.abs(np.min(rn.r.ravel()) - 3.98), 1e-5)
self.assertLess(np.abs(np.min(m.v[:,2]) - np.min(rn.r.ravel())), 1e-5)
self.assertLess(np.abs(rn.r[h/2,w/2] - 3.98), 1e-5)
def test_project_points_without_derivatives(self):
from util_tests import get_earthmesh
cam_params = self.get_cam_params()
mesh = get_earthmesh(trans=np.array([0,0,5]), rotation = np.array([0,0,0]))
pp = ProjectPoints(f=cam_params['f'], rt=cam_params['rt'], t=cam_params['t'], k=cam_params['k'], c=cam_params['c'])
pp.v = mesh.v
np.testing.assert_array_equal(pp.r, pp.r_and_derivatives[0].squeeze())
def get_sphere_mesh():
from util_tests import get_earthmesh
mesh = get_earthmesh(np.zeros(3), np.zeros(3)) # load_mesh(filename)
v, f = mesh.v*64., mesh.f
for i in range(3):
mtx, f = loop_subdivider(v, f)
v = mtx.dot(v.ravel()).reshape((-1,3))
v /= 200.
v[:,2] += 2
return v, f
def get_sphere_mesh():
from util_tests import get_earthmesh
mesh = get_earthmesh(np.zeros(3), np.zeros(3)) # load_mesh(filename)
v, f = mesh.v*64., mesh.f
for i in range(3):
mtx, f = loop_subdivider(v, f)
v = mtx.dot(v.ravel()).reshape((-1,3))
v /= 200.
v[:,2] += 2
return v, f
def test_project_points_without_derivatives(self):
from util_tests import get_earthmesh
cam_params = self.get_cam_params()
mesh = get_earthmesh(trans=np.array([0, 0, 5]),
rotation=np.array([0, 0, 0]))
pp = ProjectPoints(f=cam_params['f'],
rt=cam_params['rt'],
t=cam_params['t'],
k=cam_params['k'],
c=cam_params['c'])
pp.v = mesh.v
np.testing.assert_array_equal(pp.r, pp.r_and_derivatives[0].squeeze())
def test_derivatives2(self):
import chumpy as ch
import numpy as np
from renderer import DepthRenderer
rn = DepthRenderer()
# Assign attributes to renderer
from util_tests import get_earthmesh
m = get_earthmesh(trans=ch.array([0,0,4]), rotation=ch.zeros(3))
w, h = (320, 240)
from camera import ProjectPoints
rn.camera = ProjectPoints(v=m.v, rt=ch.zeros(3), t=ch.zeros(3), f=ch.array([w,w])/2., c=ch.array([w,h])/2., k=ch.zeros(5))
rn.frustum = {'near': 1., 'far': 10., 'width': w, 'height': h}
rn.set(v=m.v, f=m.f, bgcolor=ch.zeros(3))
if visualize:
import matplotlib.pyplot as plt
plt.ion()
plt.figure()
for which in range(3):
r1 = rn.r
adder = np.random.rand(rn.v.r.size).reshape(rn.v.r.shape)*.01
change = rn.v.r * 0 + adder
dr_pred = rn.dr_wrt(rn.v).dot(change.ravel()).reshape(rn.shape)
rn.v = rn.v.r + change
r2 = rn.r
dr_emp = r2 - r1
#print np.mean(np.abs(dr_pred-dr_emp))
self.assertLess(np.mean(np.abs(dr_pred-dr_emp)), .024)
if visualize:
plt.subplot(2,3,which+1)
plt.imshow(dr_pred)
plt.clim(-.01,.01)
plt.title('emp')
plt.subplot(2,3,which+4)
plt.imshow(dr_emp)
plt.clim(-.01,.01)
plt.title('pred')
plt.draw()
plt.show()
def test_vert_normals(self):
from geometry import VertNormals
import numpy as np
mesh = get_earthmesh(np.zeros(3), np.zeros(3))
v, f = mesh.v*127., mesh.f
vn1 = VertNormals(f=f, v=v)
dr_predicted = vn1.dr_wrt(vn1.v).copy()
eps = .00001 * np.random.randn(v.size).reshape(v.shape)
v += eps
vn2 = VertNormals(v=v, f=f)
empirical_diff = (vn2.r - vn1.r).reshape((-1,3))
predicted_diff = dr_predicted.dot(eps.flatten()).reshape((-1,3))
if False:
print np.max(np.abs(empirical_diff-predicted_diff))
print empirical_diff[:6]
print predicted_diff[:6]
self.assertTrue(np.max(np.abs(empirical_diff-predicted_diff)) < 6e-13)
def test_occlusion(self):
if visualize:
import matplotlib.pyplot as plt
plt.ion()
# Create renderer
import chumpy as ch
import numpy as np
from opendr.renderer import TexturedRenderer, ColoredRenderer
#rn = TexturedRenderer()
rn = ColoredRenderer()
# Assign attributes to renderer
from util_tests import get_earthmesh
m = get_earthmesh(trans=ch.array([0,0,4]), rotation=ch.zeros(3))
rn.texture_image = m.texture_image
rn.ft = m.ft
rn.vt = m.vt
m.v[:,2] = np.mean(m.v[:,2])
# red is front and zero
# green is back and 1
t0 = ch.array([1,0,.1])
t1 = ch.array([-1,0,.1])
v0 = ch.array(m.v) + t0
if False:
v1 = ch.array(m.v*.4 + np.array([0,0,3.8])) + t1
else:
v1 = ch.array(m.v) + t1
vc0 = v0*0 + np.array([[.4,0,0]])
vc1 = v1*0 + np.array([[0,.4,0]])
vc = ch.vstack((vc0, vc1))
v = ch.vstack((v0, v1))
f = np.vstack((m.f, m.f+len(v0)))
w, h = (320, 240)
rn.camera = ProjectPoints(v=v, rt=ch.zeros(3), t=ch.zeros(3), f=ch.array([w,w])/2., c=ch.array([w,h])/2., k=ch.zeros(5))
rn.camera.t = ch.array([0,0,-2.5])
rn.frustum = {'near': 1., 'far': 10., 'width': w, 'height': h}
m.vc = v.r*0 + np.array([[1,0,0]])
rn.set(v=v, f=f, vc=vc)
t0[:] = np.array([1.4, 0, .1-.02])
t1[:] = np.array([-0.6, 0, .1+.02])
target = rn.r
if visualize:
plt.figure()
plt.imshow(target)
plt.title('target')
plt.figure()
plt.show()
im_orig = rn.r.copy()
from cvwrap import cv2
tr = t0
eps_emp = .02
eps_pred = .02
#blur = lambda x : cv2.blur(x, ksize=(5,5))
blur = lambda x : x
for tr in [t0, t1]:
if tr is t0:
sum_limits = np.array([2.1e+2, 6.9e+1, 1.6e+2])
else:
sum_limits = [1., 5., 4.]
if visualize:
plt.figure()
for i in range(3):
dr_pred = np.array(rn.dr_wrt(tr[i]).todense()).reshape(rn.shape) * eps_pred
dr_pred = blur(dr_pred)
# central differences
tr[i] = tr[i].r + eps_emp/2.
rn_greater = rn.r.copy()
tr[i] = tr[i].r - eps_emp/1.
rn_lesser = rn.r.copy()
tr[i] = tr[i].r + eps_emp/2.
dr_emp = blur((rn_greater - rn_lesser) * eps_pred / eps_emp)
dr_pred_shown = np.clip(dr_pred, -.5, .5) + .5
dr_emp_shown = np.clip(dr_emp, -.5, .5) + .5
if visualize:
plt.subplot(3,3,i+1)
plt.imshow(dr_pred_shown)
plt.title('pred')
plt.axis('off')
plt.subplot(3,3,3+i+1)
plt.imshow(dr_emp_shown)
plt.title('empirical')
plt.axis('off')
plt.subplot(3,3,6+i+1)
diff = np.abs(dr_emp - dr_pred)
if visualize:
plt.imshow(diff)
diff = diff.ravel()
if visualize:
plt.title('diff (sum: %.2e)' % (np.sum(diff)))
plt.axis('off')
# print 'dr pred sum: %.2e' % (np.sum(np.abs(dr_pred.ravel())),)
# print 'dr emp sum: %.2e' % (np.sum(np.abs(dr_emp.ravel())),)
#import pdb; pdb.set_trace()
self.assertTrue(np.sum(diff) < sum_limits[i])
def test_color_derivatives(self):
mesh, lightings, camera, frustum, renderers = self.load_basics()
for renderer in renderers:
im_shape = renderer.shape
lighting = lightings[renderer.num_channels]
# Get pixels and dI/dC
mesh = get_earthmesh(trans=np.array([0,0,5]), rotation = np.array([math.pi/2.,0,0]))
mesh_verts = Ch(mesh.v)
mesh_colors = Ch(mesh.vc)
camera.set(v=mesh_verts)
# import pdb; pdb.set_trace()
# print '-------------------------------------------'
#lighting.set(vc=mesh_colors, v=mesh_verts)
lighting.vc = mesh_colors[:,:renderer.num_channels]
lighting.v = mesh_verts
renderer.set(v=mesh_verts, vc=lighting)
r = renderer.r
dr = renderer.dr_wrt(mesh_colors).copy()
# Establish a random direction
eps = .4
direction = (np.random.randn(mesh.v.size).reshape(mesh.v.shape)*.1 + np.sin(mesh.v*19)*.1).flatten()
# Find empirical forward derivatives in that direction
mesh_colors = Ch(mesh.vc+direction.reshape(mesh.vc.shape)*eps/2.)
lighting.set(vc=mesh_colors[:,:renderer.num_channels])
renderer.set(vc=lighting)
rfwd = renderer.r
# Find empirical backward derivatives in that direction
mesh_colors = Ch(mesh.vc-direction.reshape(mesh.vc.shape)*eps/2.)
lighting.set(vc=mesh_colors[:,:renderer.num_channels])
renderer.set(vc=lighting)
rbwd = renderer.r
dr_empirical = (np.asarray(rfwd, np.float64) - np.asarray(rbwd, np.float64)).ravel() / eps
dr_predicted = dr.dot(col(direction.flatten())).reshape(dr_empirical.shape)
images = OrderedDict()
images['shifted colors'] = np.asarray(rfwd, np.float64)-.5
images[r'empirical colors $\left(\frac{dI}{dC}\right)$'] = dr_empirical
images[r'predicted colors $\left(\frac{dI}{dC}\right)$'] = dr_predicted
images['difference colors'] = dr_predicted-dr_empirical
nonzero = images['difference colors'][np.nonzero(images['difference colors']!=0)[0]]
if visualize:
matplotlib.rcParams.update({'font.size': 18})
plt.figure(figsize=(6*3, 2*3))
for idx, title in enumerate(images.keys()):
plt.subplot(1,len(images.keys()), idx+1)
im = process(images[title].reshape(im_shape), vmin=-.5, vmax=.5)
plt.title(title)
plt.imshow(im)
plt.show()
print 'color: median nonzero %.2e' % (np.median(np.abs(nonzero)),)
print 'color: mean nonzero %.2e' % (np.mean(np.abs(nonzero)),)
self.assertLess(np.mean(np.abs(nonzero)), 2e-2)
self.assertLess(np.median(np.abs(nonzero)), 4.5e-3)
def test_lightpos_derivatives(self):
mesh, lightings, camera, frustum, renderers = self.load_basics()
for renderer in renderers:
im_shape = renderer.shape
lighting = lightings[renderer.num_channels]
# Render a rotating mesh
mesh = get_earthmesh(trans=np.array([0,0,5]), rotation = np.array([math.pi/2.,0,0]))
mesh_verts = Ch(mesh.v.flatten())
camera.set(v=mesh_verts)
# Get predicted derivatives wrt light pos
light1_pos = Ch(np.array([-1000,-1000,-1000]))
lighting.set(light_pos=light1_pos, v=mesh_verts)
renderer.set(vc=lighting, v=mesh_verts)
dr = renderer.dr_wrt(light1_pos).copy()
# Establish a random direction for the light
direction = (np.random.rand(3)-.5)*1000.
eps = 1.
# Find empirical forward derivatives in that direction
lighting.set(light_pos = light1_pos.r + direction*eps/2.)
renderer.set(vc=lighting)
rfwd = renderer.r
# Find empirical backward derivatives in that direction
lighting.set(light_pos = light1_pos.r - direction*eps/2.)
renderer.set(vc=lighting)
rbwd = renderer.r
# Establish empirical and predicted derivatives
dr_empirical = (np.asarray(rfwd, np.float64) - np.asarray(rbwd, np.float64)).ravel() / eps
dr_predicted = dr.dot(col(direction.flatten())).reshape(dr_empirical.shape)
images = OrderedDict()
images['shifted lightpos'] = np.asarray(rfwd, np.float64)-.5
images[r'empirical lightpos $\left(\frac{dI}{dL_p}\right)$'] = dr_empirical
images[r'predicted lightpos $\left(\frac{dI}{dL_p}\right)$'] = dr_predicted
images['difference lightpos'] = dr_predicted-dr_empirical
nonzero = images['difference lightpos'][np.nonzero(images['difference lightpos']!=0)[0]]
if visualize:
matplotlib.rcParams.update({'font.size': 18})
plt.figure(figsize=(6*3, 2*3))
for idx, title in enumerate(images.keys()):
plt.subplot(1,len(images.keys()), idx+1)
im = process(images[title].reshape(im_shape), vmin=-.5, vmax=.5)
plt.title(title)
plt.imshow(im)
plt.show()
print 'lightpos: median nonzero %.2e' % (np.median(np.abs(nonzero)),)
print 'lightpos: mean nonzero %.2e' % (np.mean(np.abs(nonzero)),)
self.assertLess(np.mean(np.abs(nonzero)), 2.4e-2)
self.assertLess(np.median(np.abs(nonzero)), 1.2e-2)
def test_vert_derivatives(self):
mesh, lightings, camera, frustum, renderers = self.load_basics()
for renderer in renderers:
lighting = lightings[renderer.num_channels]
im_shape = renderer.shape
# Render a rotating mesh
mesh = get_earthmesh(trans=np.array([0,0,5]), rotation = np.array([math.pi/2.,0,0]))
mesh_verts = Ch(mesh.v.flatten())
camera.set(v=mesh_verts)
lighting.set(v=mesh_verts)
renderer.set(camera=camera)
renderer.set(vc=lighting)
# Get pixels and derivatives
r = renderer.r
dr = renderer.dr_wrt(mesh_verts)
# Establish a random direction
direction = (np.random.rand(mesh.v.size).reshape(mesh.v.shape)-.5)*.1 + np.sin(mesh.v*10)*.2
direction *= .5
eps = .2
# Render going forward in that direction
mesh_verts = Ch(mesh.v+direction*eps/2.)
lighting.set(v=mesh_verts)
renderer.set(v=mesh_verts, vc=lighting)
rfwd = renderer.r
# Render going backward in that direction
mesh_verts = Ch(mesh.v-direction*eps/2.)
lighting.set(v=mesh_verts)
renderer.set(v=mesh_verts, vc=lighting)
rbwd = renderer.r
# Establish empirical and predicted derivatives
dr_empirical = (np.asarray(rfwd, np.float64) - np.asarray(rbwd, np.float64)).ravel() / eps
dr_predicted = dr.dot(col(direction.flatten())).reshape(dr_empirical.shape)
images = OrderedDict()
images['shifted verts'] = np.asarray(rfwd, np.float64)-.5
images[r'empirical verts $\left(\frac{dI}{dV}\right)$'] = dr_empirical
images[r'predicted verts $\left(\frac{dI}{dV}\right)$'] = dr_predicted
images['difference verts'] = dr_predicted - dr_empirical
nonzero = images['difference verts'][np.nonzero(images['difference verts']!=0)[0]]
if visualize:
matplotlib.rcParams.update({'font.size': 18})
plt.figure(figsize=(6*3, 2*3))
for idx, title in enumerate(images.keys()):
plt.subplot(1,len(images.keys()), idx+1)
im = process(images[title].reshape(im_shape), vmin=-.5, vmax=.5)
plt.title(title)
plt.imshow(im)
print 'verts: median nonzero %.2e' % (np.median(np.abs(nonzero)),)
print 'verts: mean nonzero %.2e' % (np.mean(np.abs(nonzero)),)
plt.draw()
plt.show()
self.assertLess(np.mean(np.abs(nonzero)), 7e-2)
self.assertLess(np.median(np.abs(nonzero)), 4e-2)
def test_cam_derivatives(self):
mesh, lightings, camera, frustum, renderers = self.load_basics()
camparms = {
'c': {'mednz' : 2.2e-2, 'meannz': 4.2e-2, 'desc': 'center of proj diff', 'eps0': 4., 'eps1': .1},
#'f': {'mednz' : 2.5e-2, 'meannz': 6e-2, 'desc': 'focal diff', 'eps0': 100., 'eps1': .1},
't': {'mednz' : 1.2e-1, 'meannz': 3.0e-1, 'desc': 'trans diff', 'eps0': .25, 'eps1': .1},
'rt': {'mednz' : 8e-2, 'meannz': 1.8e-1, 'desc': 'rot diff', 'eps0': 0.02, 'eps1': .5},
'k': {'mednz' : 7e-2, 'meannz': 5.1e-1, 'desc': 'distortion diff', 'eps0': .5, 'eps1': .05}
}
for renderer in renderers:
im_shape = renderer.shape
lighting = lightings[renderer.num_channels]
# Render a rotating mesh
mesh = get_earthmesh(trans=np.array([0,0,5]), rotation = np.array([math.pi/2.,0,0]))
mesh_verts = Ch(mesh.v.flatten())
camera.v = mesh_verts
lighting.v = mesh_verts
renderer.vc = lighting
renderer.camera = camera
for atrname, info in camparms.items():
# Get pixels and derivatives
r = renderer.r
atr = lambda : getattr(camera, atrname)
satr = lambda x : setattr(camera, atrname, x)
atr_size = atr().size
dr = renderer.dr_wrt(atr())
# Establish a random direction
tmp = np.random.rand(atr().size) - .5
direction = (tmp / np.linalg.norm(tmp))*info['eps0']
#direction = np.sin(np.ones(atr_size))*info['eps0']
#direction = np.zeros(atr_size)
# try:
# direction[4] = 1.
# except: pass
#direction *= info['eps0']
eps = info['eps1']
# Render going forward in that direction
satr(atr().r + direction*eps/2.)
rfwd = renderer.r
# Render going backward in that direction
satr(atr().r - direction*eps/1.)
rbwd = renderer.r
# Put back
satr(atr().r + direction*eps/2.)
# Establish empirical and predicted derivatives
dr_empirical = (np.asarray(rfwd, np.float64) - np.asarray(rbwd, np.float64)).ravel() / eps
dr_predicted = dr.dot(col(direction.flatten())).reshape(dr_empirical.shape)
images = OrderedDict()
images['shifted %s' % (atrname,)] = np.asarray(rfwd, np.float64)-.5
images[r'empirical %s' % (atrname,)] = dr_empirical
images[r'predicted %s' % (atrname,)] = dr_predicted
images[info['desc']] = dr_predicted - dr_empirical
nonzero = images[info['desc']][np.nonzero(images[info['desc']]!=0)[0]]
mederror = np.median(np.abs(nonzero))
meanerror = np.mean(np.abs(nonzero))
if visualize:
matplotlib.rcParams.update({'font.size': 18})
plt.figure(figsize=(6*3, 2*3))
for idx, title in enumerate(images.keys()):
plt.subplot(1,len(images.keys()), idx+1)
im = process(images[title].reshape(im_shape), vmin=-.5, vmax=.5)
plt.title(title)
plt.imshow(im)
print '%s: median nonzero %.2e' % (atrname, mederror,)
print '%s: mean nonzero %.2e' % (atrname, meanerror,)
plt.draw()
plt.show()
self.assertLess(meanerror, info['meannz'])
self.assertLess(mederror, info['mednz'])
def test_distortion(self):
mesh, lightings, camera, frustum, renderers = self.load_basics()
renderer = renderers[1]
lighting = lightings[renderer.num_channels]
lighting.light_pos = -lighting.light_pos * 100.
mesh = get_earthmesh(trans=np.array([0,0,-8]), rotation = np.array([math.pi/2.,0,0]))
mesh_verts = Ch(mesh.v.flatten())
renderer.camera = camera
camera.v = mesh_verts
lighting.v = mesh_verts
renderer.vc = lighting
renderer.camera = camera
camera.rt = np.array([np.pi, 0, 0])
# Get pixels and derivatives
im_original = renderer.r.copy()
#camera.k = np.zeros(5)
#camera.k = np.arange(8,0,-1)*.1
#camera.k = np.array([ 0.00249999, 0.42208098, 0.45360267, 0.06808415, -0.38003062])
camera.k = np.array([ 5., 25., .3, .4, 1000., 5., 0., 0.])
im_distorted = renderer.r
cr = renderer
cmtx = np.array([
[cr.camera.f.r[0], 0, cr.camera.c.r[0]],
[0, cr.camera.f.r[1], cr.camera.c.r[1]],
[0, 0, 1]
])
from cvwrap import cv2
im_undistorted = cv2.undistort(im_distorted, cmtx, cr.camera.k.r)
d1 = (im_original - im_distorted).ravel()
d2 = (im_original - im_undistorted).ravel()
d1 = d1[d1 != 0.]
d2 = d2[d2 != 0.]
self.assertGreater(np.mean(d1**2) / np.mean(d2**2), 44.)
self.assertLess(np.mean(d2**2), 0.0016)
self.assertGreater(np.median(d1**2) / np.median(d2**2), 650)
self.assertLess(np.median(d2**2), 1.9e-5)
if visualize:
import matplotlib.pyplot as plt
plt.ion()
matplotlib.rcParams.update({'font.size': 18})
plt.figure(figsize=(6*3, 2*3))
plt.subplot(1,4,1)
plt.imshow(im_original)
plt.title('original')
plt.subplot(1,4,2)
plt.imshow(im_distorted)
plt.title('distorted')
plt.subplot(1,4,3)
plt.imshow(im_undistorted)
plt.title('undistorted by opencv')
plt.subplot(1,4,4)
plt.imshow(im_undistorted - im_original + .5)
plt.title('diff')
plt.draw()
plt.show()
def test_distortion(self):
mesh, lightings, camera, frustum, renderers = self.load_basics()
renderer = renderers[1]
lighting = lightings[renderer.num_channels]
lighting.light_pos = -lighting.light_pos * 100.
mesh = get_earthmesh(trans=np.array([0, 0, -8]),
rotation=np.array([math.pi / 2., 0, 0]))
mesh_verts = Ch(mesh.v.flatten())
renderer.camera = camera
camera.v = mesh_verts
lighting.v = mesh_verts
renderer.vc = lighting
renderer.camera = camera
camera.rt = np.array([np.pi, 0, 0])
# Get pixels and derivatives
im_original = renderer.r.copy()
#camera.k = np.zeros(5)
#camera.k = np.arange(8,0,-1)*.1
#camera.k = np.array([ 0.00249999, 0.42208098, 0.45360267, 0.06808415, -0.38003062])
camera.k = np.array([5., 25., .3, .4, 1000., 5., 0., 0.])
im_distorted = renderer.r
cr = renderer
cmtx = np.array([[cr.camera.f.r[0], 0, cr.camera.c.r[0]],
[0, cr.camera.f.r[1], cr.camera.c.r[1]], [0, 0, 1]])
from cvwrap import cv2
im_undistorted = cv2.undistort(im_distorted, cmtx, cr.camera.k.r)
d1 = (im_original - im_distorted).ravel()
d2 = (im_original - im_undistorted).ravel()
d1 = d1[d1 != 0.]
d2 = d2[d2 != 0.]
self.assertGreater(np.mean(d1**2) / np.mean(d2**2), 44.)
self.assertLess(np.mean(d2**2), 0.0016)
self.assertGreater(np.median(d1**2) / np.median(d2**2), 650)
self.assertLess(np.median(d2**2), 1.9e-5)
if visualize:
import matplotlib.pyplot as plt
plt.ion()
matplotlib.rcParams.update({'font.size': 18})
plt.figure(figsize=(6 * 3, 2 * 3))
plt.subplot(1, 4, 1)
plt.imshow(im_original)
plt.title('original')
plt.subplot(1, 4, 2)
plt.imshow(im_distorted)
plt.title('distorted')
plt.subplot(1, 4, 3)
plt.imshow(im_undistorted)
plt.title('undistorted by opencv')
plt.subplot(1, 4, 4)
plt.imshow(im_undistorted - im_original + .5)
plt.title('diff')
plt.draw()
plt.show()
def test_cam_derivatives(self):
mesh, lightings, camera, frustum, renderers = self.load_basics()
camparms = {
'c': {
'mednz': 2.2e-2,
'meannz': 4.3e-2,
'desc': 'center of proj diff',
'eps0': 4.,
'eps1': .1
},
#'f': {'mednz' : 2.5e-2, 'meannz': 6e-2, 'desc': 'focal diff', 'eps0': 100., 'eps1': .1},
't': {
'mednz': 1.2e-1,
'meannz': 3.0e-1,
'desc': 'trans diff',
'eps0': .25,
'eps1': .1
},
'rt': {
'mednz': 8e-2,
'meannz': 1.8e-1,
'desc': 'rot diff',
'eps0': 0.02,
'eps1': .5
},
'k': {
'mednz': 7e-2,
'meannz': 5.1e-1,
'desc': 'distortion diff',
'eps0': .5,
'eps1': .05
}
}
for renderer in renderers:
im_shape = renderer.shape
lighting = lightings[renderer.num_channels]
# Render a rotating mesh
mesh = get_earthmesh(trans=np.array([0, 0, 5]),
rotation=np.array([math.pi / 2., 0, 0]))
mesh_verts = Ch(mesh.v.flatten())
camera.v = mesh_verts
lighting.v = mesh_verts
renderer.vc = lighting
renderer.camera = camera
for atrname, info in camparms.items():
# Get pixels and derivatives
r = renderer.r
atr = lambda: getattr(camera, atrname)
satr = lambda x: setattr(camera, atrname, x)
atr_size = atr().size
dr = renderer.dr_wrt(atr())
# Establish a random direction
tmp = np.random.rand(atr().size) - .5
direction = (tmp / np.linalg.norm(tmp)) * info['eps0']
#direction = np.sin(np.ones(atr_size))*info['eps0']
#direction = np.zeros(atr_size)
# try:
# direction[4] = 1.
# except: pass
#direction *= info['eps0']
eps = info['eps1']
# Render going forward in that direction
satr(atr().r + direction * eps / 2.)
rfwd = renderer.r
# Render going backward in that direction
satr(atr().r - direction * eps / 1.)
rbwd = renderer.r
# Put back
satr(atr().r + direction * eps / 2.)
# Establish empirical and predicted derivatives
dr_empirical = (np.asarray(rfwd, np.float64) -
np.asarray(rbwd, np.float64)).ravel() / eps
dr_predicted = dr.dot(col(direction.flatten())).reshape(
dr_empirical.shape)
images = OrderedDict()
images['shifted %s' %
(atrname, )] = np.asarray(rfwd, np.float64) - .5
images[r'empirical %s' % (atrname, )] = dr_empirical
images[r'predicted %s' % (atrname, )] = dr_predicted
images[info['desc']] = dr_predicted - dr_empirical
nonzero = images[info['desc']][np.nonzero(
images[info['desc']] != 0)[0]]
mederror = np.median(np.abs(nonzero))
meanerror = np.mean(np.abs(nonzero))
if visualize:
matplotlib.rcParams.update({'font.size': 18})
plt.figure(figsize=(6 * 3, 2 * 3))
for idx, title in enumerate(images.keys()):
plt.subplot(1, len(images.keys()), idx + 1)
im = process(images[title].reshape(im_shape),
vmin=-.5,
vmax=.5)
plt.title(title)
plt.imshow(im)
print '%s: median nonzero %.2e' % (
atrname,
mederror,
)
print '%s: mean nonzero %.2e' % (
atrname,
meanerror,
)
plt.draw()
plt.show()
self.assertLess(meanerror, info['meannz'])
self.assertLess(mederror, info['mednz'])
def test_occlusion(self):
if visualize:
import matplotlib.pyplot as plt
plt.ion()
# Create renderer
import chumpy as ch
import numpy as np
from opendr.renderer import TexturedRenderer, ColoredRenderer
#rn = TexturedRenderer()
rn = ColoredRenderer()
# Assign attributes to renderer
from util_tests import get_earthmesh
m = get_earthmesh(trans=ch.array([0,0,4]), rotation=ch.zeros(3))
rn.texture_image = m.texture_image
rn.ft = m.ft
rn.vt = m.vt
m.v[:,2] = np.mean(m.v[:,2])
# red is front and zero
# green is back and 1
t0 = ch.array([1,0,.1])
t1 = ch.array([-1,0,.1])
v0 = ch.array(m.v) + t0
if False:
v1 = ch.array(m.v*.4 + np.array([0,0,3.8])) + t1
else:
v1 = ch.array(m.v) + t1
vc0 = v0*0 + np.array([[.4,0,0]])
vc1 = v1*0 + np.array([[0,.4,0]])
vc = ch.vstack((vc0, vc1))
v = ch.vstack((v0, v1))
f = np.vstack((m.f, m.f+len(v0)))
w, h = (320, 240)
rn.camera = ProjectPoints(v=v, rt=ch.zeros(3), t=ch.zeros(3), f=ch.array([w,w])/2., c=ch.array([w,h])/2., k=ch.zeros(5))
rn.camera.t = ch.array([0,0,-2.5])
rn.frustum = {'near': 1., 'far': 10., 'width': w, 'height': h}
m.vc = v.r*0 + np.array([[1,0,0]])
rn.set(v=v, f=f, vc=vc)
t0[:] = np.array([1.4, 0, .1-.02])
t1[:] = np.array([-0.6, 0, .1+.02])
target = rn.r
if visualize:
plt.figure()
plt.imshow(target)
plt.title('target')
plt.figure()
plt.show()
im_orig = rn.r.copy()
from cvwrap import cv2
tr = t0
eps_emp = .02
eps_pred = .02
#blur = lambda x : cv2.blur(x, ksize=(5,5))
blur = lambda x : x
for tr in [t0, t1]:
if tr is t0:
sum_limits = np.array([2.1e+2, 6.9e+1, 1.6e+2])
else:
sum_limits = [1., 5., 4.]
if visualize:
plt.figure()
for i in range(3):
dr_pred = np.array(rn.dr_wrt(tr[i]).toarray()).reshape(rn.shape) * eps_pred
dr_pred = blur(dr_pred)
# central differences
tr[i] = tr[i].r + eps_emp/2.
rn_greater = rn.r.copy()
tr[i] = tr[i].r - eps_emp/1.
rn_lesser = rn.r.copy()
tr[i] = tr[i].r + eps_emp/2.
dr_emp = blur((rn_greater - rn_lesser) * eps_pred / eps_emp)
dr_pred_shown = np.clip(dr_pred, -.5, .5) + .5
dr_emp_shown = np.clip(dr_emp, -.5, .5) + .5
if visualize:
plt.subplot(3,3,i+1)
plt.imshow(dr_pred_shown)
plt.title('pred')
plt.axis('off')
plt.subplot(3,3,3+i+1)
plt.imshow(dr_emp_shown)
plt.title('empirical')
plt.axis('off')
plt.subplot(3,3,6+i+1)
diff = np.abs(dr_emp - dr_pred)
if visualize:
plt.imshow(diff)
diff = diff.ravel()
if visualize:
plt.title('diff (sum: %.2e)' % (np.sum(diff)))
plt.axis('off')
# print 'dr pred sum: %.2e' % (np.sum(np.abs(dr_pred.ravel())),)
# print 'dr emp sum: %.2e' % (np.sum(np.abs(dr_emp.ravel())),)
#import pdb; pdb.set_trace()
self.assertTrue(np.sum(diff) < sum_limits[i])
def test_vert_derivatives(self):
mesh, lightings, camera, frustum, renderers = self.load_basics()
for renderer in renderers:
lighting = lightings[renderer.num_channels]
im_shape = renderer.shape
# Render a rotating mesh
mesh = get_earthmesh(trans=np.array([0, 0, 5]),
rotation=np.array([math.pi / 2., 0, 0]))
mesh_verts = Ch(mesh.v.flatten())
camera.set(v=mesh_verts)
lighting.set(v=mesh_verts)
renderer.set(camera=camera)
renderer.set(vc=lighting)
# Get pixels and derivatives
r = renderer.r
dr = renderer.dr_wrt(mesh_verts)
# Establish a random direction
direction = (np.random.rand(mesh.v.size).reshape(mesh.v.shape) -
.5) * .1 + np.sin(mesh.v * 10) * .2
direction *= .5
eps = .2
# Render going forward in that direction
mesh_verts = Ch(mesh.v + direction * eps / 2.)
lighting.set(v=mesh_verts)
renderer.set(v=mesh_verts, vc=lighting)
rfwd = renderer.r
# Render going backward in that direction
mesh_verts = Ch(mesh.v - direction * eps / 2.)
lighting.set(v=mesh_verts)
renderer.set(v=mesh_verts, vc=lighting)
rbwd = renderer.r
# Establish empirical and predicted derivatives
dr_empirical = (np.asarray(rfwd, np.float64) -
np.asarray(rbwd, np.float64)).ravel() / eps
dr_predicted = dr.dot(col(direction.flatten())).reshape(
dr_empirical.shape)
images = OrderedDict()
images['shifted verts'] = np.asarray(rfwd, np.float64) - .5
images[
r'empirical verts $\left(\frac{dI}{dV}\right)$'] = dr_empirical
images[
r'predicted verts $\left(\frac{dI}{dV}\right)$'] = dr_predicted
images['difference verts'] = dr_predicted - dr_empirical
nonzero = images['difference verts'][np.nonzero(
images['difference verts'] != 0)[0]]
if visualize:
matplotlib.rcParams.update({'font.size': 18})
plt.figure(figsize=(6 * 3, 2 * 3))
for idx, title in enumerate(images.keys()):
plt.subplot(1, len(images.keys()), idx + 1)
im = process(images[title].reshape(im_shape),
vmin=-.5,
vmax=.5)
plt.title(title)
plt.imshow(im)
print 'verts: median nonzero %.2e' % (np.median(
np.abs(nonzero)), )
print 'verts: mean nonzero %.2e' % (np.mean(np.abs(nonzero)), )
plt.draw()
plt.show()
self.assertLess(np.mean(np.abs(nonzero)), 7e-2)
self.assertLess(np.median(np.abs(nonzero)), 4e-2)
def test_color_derivatives(self):
mesh, lightings, camera, frustum, renderers = self.load_basics()
for renderer in renderers:
im_shape = renderer.shape
lighting = lightings[renderer.num_channels]
# Get pixels and dI/dC
mesh = get_earthmesh(trans=np.array([0, 0, 5]),
rotation=np.array([math.pi / 2., 0, 0]))
mesh_verts = Ch(mesh.v)
mesh_colors = Ch(mesh.vc)
camera.set(v=mesh_verts)
# import pdb; pdb.set_trace()
# print '-------------------------------------------'
#lighting.set(vc=mesh_colors, v=mesh_verts)
lighting.vc = mesh_colors[:, :renderer.num_channels]
lighting.v = mesh_verts
renderer.set(v=mesh_verts, vc=lighting)
r = renderer.r
dr = renderer.dr_wrt(mesh_colors).copy()
# Establish a random direction
eps = .4
direction = (
np.random.randn(mesh.v.size).reshape(mesh.v.shape) * .1 +
np.sin(mesh.v * 19) * .1).flatten()
# Find empirical forward derivatives in that direction
mesh_colors = Ch(mesh.vc +
direction.reshape(mesh.vc.shape) * eps / 2.)
lighting.set(vc=mesh_colors[:, :renderer.num_channels])
renderer.set(vc=lighting)
rfwd = renderer.r
# Find empirical backward derivatives in that direction
mesh_colors = Ch(mesh.vc -
direction.reshape(mesh.vc.shape) * eps / 2.)
lighting.set(vc=mesh_colors[:, :renderer.num_channels])
renderer.set(vc=lighting)
rbwd = renderer.r
dr_empirical = (np.asarray(rfwd, np.float64) -
np.asarray(rbwd, np.float64)).ravel() / eps
dr_predicted = dr.dot(col(direction.flatten())).reshape(
dr_empirical.shape)
images = OrderedDict()
images['shifted colors'] = np.asarray(rfwd, np.float64) - .5
images[
r'empirical colors $\left(\frac{dI}{dC}\right)$'] = dr_empirical
images[
r'predicted colors $\left(\frac{dI}{dC}\right)$'] = dr_predicted
images['difference colors'] = dr_predicted - dr_empirical
nonzero = images['difference colors'][np.nonzero(
images['difference colors'] != 0)[0]]
if visualize:
matplotlib.rcParams.update({'font.size': 18})
plt.figure(figsize=(6 * 3, 2 * 3))
for idx, title in enumerate(images.keys()):
plt.subplot(1, len(images.keys()), idx + 1)
im = process(images[title].reshape(im_shape),
vmin=-.5,
vmax=.5)
plt.title(title)
plt.imshow(im)
plt.show()
print 'color: median nonzero %.2e' % (np.median(
np.abs(nonzero)), )
print 'color: mean nonzero %.2e' % (np.mean(np.abs(nonzero)), )
self.assertLess(np.mean(np.abs(nonzero)), 2e-2)
self.assertLess(np.median(np.abs(nonzero)), 4.5e-3)
def test_lightpos_derivatives(self):
mesh, lightings, camera, frustum, renderers = self.load_basics()
for renderer in renderers:
im_shape = renderer.shape
lighting = lightings[renderer.num_channels]
# Render a rotating mesh
mesh = get_earthmesh(trans=np.array([0, 0, 5]),
rotation=np.array([math.pi / 2., 0, 0]))
mesh_verts = Ch(mesh.v.flatten())
camera.set(v=mesh_verts)
# Get predicted derivatives wrt light pos
light1_pos = Ch(np.array([-1000, -1000, -1000]))
lighting.set(light_pos=light1_pos, v=mesh_verts)
renderer.set(vc=lighting, v=mesh_verts)
dr = renderer.dr_wrt(light1_pos).copy()
# Establish a random direction for the light
direction = (np.random.rand(3) - .5) * 1000.
eps = 1.
# Find empirical forward derivatives in that direction
lighting.set(light_pos=light1_pos.r + direction * eps / 2.)
renderer.set(vc=lighting)
rfwd = renderer.r
# Find empirical backward derivatives in that direction
lighting.set(light_pos=light1_pos.r - direction * eps / 2.)
renderer.set(vc=lighting)
rbwd = renderer.r
# Establish empirical and predicted derivatives
dr_empirical = (np.asarray(rfwd, np.float64) -
np.asarray(rbwd, np.float64)).ravel() / eps
dr_predicted = dr.dot(col(direction.flatten())).reshape(
dr_empirical.shape)
images = OrderedDict()
images['shifted lightpos'] = np.asarray(rfwd, np.float64) - .5
images[
r'empirical lightpos $\left(\frac{dI}{dL_p}\right)$'] = dr_empirical
images[
r'predicted lightpos $\left(\frac{dI}{dL_p}\right)$'] = dr_predicted
images['difference lightpos'] = dr_predicted - dr_empirical
nonzero = images['difference lightpos'][np.nonzero(
images['difference lightpos'] != 0)[0]]
if visualize:
matplotlib.rcParams.update({'font.size': 18})
plt.figure(figsize=(6 * 3, 2 * 3))
for idx, title in enumerate(images.keys()):
plt.subplot(1, len(images.keys()), idx + 1)
im = process(images[title].reshape(im_shape),
vmin=-.5,
vmax=.5)
plt.title(title)
plt.imshow(im)
plt.show()
print 'lightpos: median nonzero %.2e' % (np.median(
np.abs(nonzero)), )
print 'lightpos: mean nonzero %.2e' % (np.mean(
np.abs(nonzero)), )
self.assertLess(np.mean(np.abs(nonzero)), 2.4e-2)
self.assertLess(np.median(np.abs(nonzero)), 1.2e-2)
