def materialList(): shape_name = "ThreeBox" num_rows = 1 num_cols = len(colorMapFiles()) w = 24 h = w * num_rows / num_cols L = normalizeVector(np.array([-0.4, 0.5, 0.6])) fig, axes = plt.subplots(figsize=(w, h)) font_size = 15 fig.subplots_adjust(left=0.02, right=0.98, top=0.96, bottom=0.04, hspace=0.1, wspace=0.05) fig.suptitle("", fontsize=font_size) plot_grid = SubplotGrid(num_rows, num_cols) Ng_data = shapeFile(shape_name) Ng_data = loadNormal(Ng_data) Ng_32F, A_8U = Ng_data for colormap_file in colorMapFilesSortedReflectanceError(): M_32F = loadColorMap(colormap_file) C0_32F = ColorMapShader(M_32F).diffuseShading(L, Ng_32F) plot_grid.showImage(setAlpha(C0_32F, to32F(A_8U)), "") file_path = shapeResultFile("ShapeEstimation", "MaterialList") fig.savefig(file_path, transparent=True)
def shapeList(): num_rows = 1 num_cols = len(shapeNames()) w = 20 h = w * num_rows / num_cols cmap_id = 10 colormap_file = colorMapFile(cmap_id) M_32F = loadColorMap(colormap_file) L = normalizeVector(np.array([-0.4, 0.5, 0.6])) fig, axes = plt.subplots(figsize=(w, h)) font_size = 15 fig.subplots_adjust(left=0.02, right=0.98, top=0.96, bottom=0.04, hspace=0.15, wspace=0.1) fig.suptitle("", fontsize=font_size) plot_grid = SubplotGrid(num_rows, num_cols) for shape_name in shapeNames(): Ng_data = shapeFile(shape_name) Ng_data = loadNormal(Ng_data) Ng_32F, A_8U = Ng_data C0_32F = ColorMapShader(M_32F).diffuseShading(L, Ng_32F) plot_grid.showImage(setAlpha(C0_32F, to32F(A_8U)), "") file_path = shapeResultFile("ShapeEstimation", "ShapeList") fig.savefig(file_path, transparent=True)
def _runLayer(self, layer_file): C0_8U = loadRGBA(layer_file) if C0_8U is None: return A_8U = alpha(C0_8U) if A_8U is None: return C0_32F = to32F(rgb(C0_8U)) L = normalizeVector(np.array([-0.2, 0.3, 0.7])) sfs_method = Wu08SFS(L, C0_32F, A_8U) sfs_method.run() N_32F = sfs_method.normal() fig, axes = plt.subplots(figsize=(11, 5)) font_size = 15 fig.subplots_adjust(left=0.05, right=0.95, top=0.9, hspace=0.12, wspace=0.05) fig.suptitle(self.name(), fontsize=font_size) num_rows = 1 num_cols = 2 plot_grid = SubplotGrid(num_rows, num_cols) plot_grid.showImage(C0_8U, r"Shading: $C$") plot_grid.showImage(normalToColor(N_32F, A_8U), r"Estimated Normal: $N$") showMaximize()
def _runImp(self): normal_data = loadNormal(self._data_file) if normal_data is None: return N0_32F, A_8U = normal_data # N0_32F = cv2.resize(N0_32F, (64, 64)) # A_8U = cv2.resize(A_8U, N0_32F.shape[:2]) A_32F = to32F(A_8U) L = normalizeVector(np.array([-0.2, 0.3, 0.7])) # C0_32F = ToonShader().diffuseShading(L, N0_32F) C0_32F = LambertShader().diffuseShading(L, N0_32F) sfs_method = Wu08SFS(L, C0_32F, A_8U) sfs_method.run() N_32F = sfs_method.normal() saveNormal(self.resultFile(self._data_file_name, result_name="Wu08"), N_32F, A_8U) C_error = sfs_method.shadingError() I_32F = sfs_method.brightness() I_32F = gray2rgb(I_32F) C_32F = sfs_method.shading() N0_32F = trim(N0_32F, A_8U) C0_32F = trim(C0_32F, A_8U) C_32F = trim(C_32F, A_8U) N_32F = trim(N_32F, A_8U) C_error = trim(C_error, A_8U) I_32F = trim(I_32F, A_8U) A_32F = trim(A_32F, A_8U) A_8U = trim(A_8U, A_8U) h, w = N_32F.shape[:2] N_error = angleErros(N_32F.reshape(-1, 3), N0_32F.reshape(-1, 3)).reshape(h, w) N_error[A_8U < np.max(A_8U)] = 0.0 fig, axes = plt.subplots(figsize=(11, 5)) font_size = 15 fig.subplots_adjust(left=0.05, right=0.95, top=0.9, hspace=0.12, wspace=0.05) fig.suptitle(self.name(), fontsize=font_size) num_rows = 2 num_cols = 3 plot_grid = SubplotGrid(num_rows, num_cols) plot_grid.showImage(normalToColor(N0_32F, A_8U), r"Ground Truth Normal: $N_g$") plot_grid.showImage(normalToColor(N_32F, A_8U), r"Estimated Normal: $N$") plot_grid.showColorMap(N_error, r"Angle Error: $N_g, N$", v_min=0, v_max=30.0) plot_grid.showImage(setAlpha(C0_32F, A_32F), r"Shading: $C$") plot_grid.showImage(setAlpha(C_32F, A_32F), r"Estimated Shading: $C$") plot_grid.showColorMap(C_error, r"Shading Error: $C_g, C$", v_min=0, v_max=0.1) showMaximize()
def _compute(self): image = self._image img_32F = to32F(self._image) self._A_32F = alpha(img_32F) sigma_xy = 100.0 xy = positionFeatures(img_32F) / sigma_xy Lab = LabFeatures(img_32F) sigma_L = 1.0 Lab[:, 0] /= sigma_L foreground = foreGroundFeatures(img_32F) Labxy = np.concatenate((Lab, xy), axis=1) Labxy_samples = shuffle(Labxy[foreground, :], random_state=0)[:1000] kmeans = sklearn.cluster.KMeans(n_clusters=self._num_colors, random_state=0).fit(Labxy_samples) self._centers = kmeans.cluster_centers_ self._centers[:, 0] *= sigma_L labels = kmeans.predict(Labxy) self._labels = labels.reshape(image.shape[:2])
def _runImp(self): image = self._scene.image() I_32F = to32F(rgb2gray(rgb(image))) N_32F, D_32F = estimateNormal(I_32F) self._scene.setNormal(N_32F) self._scene.setDepth(D_32F) self._scene.setDisplayMode(Scene.DisplayNormal)
def _strokeEdited(self, stroke_sets): image = self._scene.image() C_32F = to32F(rgb(image)) for stroke_set in stroke_sets.strokeSets(): for stroke in stroke_set.strokes(): if stroke.empty(): continue mask = np.zeros(image.shape[:2], dtype=np.uint8) points = stroke.points() points = np.int32(points) brush_size = int(stroke.brushSize()) cv2.polylines(mask, [points], 0, 255, brush_size) Cs = C_32F[mask > 0, :] Is = np.arange(len(Cs), dtype=np.float32) Is = np.array(Is) Cs = np.array(Cs) M = ColorMapEstimation(Cs, Is, num_samples=1000) M_img = M.mapImage(image_size=(256, 32)) def plotFunc(): plt.imshow(M_img) self._matplot_view.drawPlots(plotFunc) saveImage(self._currentColorMapFile(), M_img) self._newColorMapID()
def __init__(self, I, radius=5, epsilon=0.4): radius = int(radius) I_32F = to32F(I) if _isGray(I): self._guided_filter = GuidedFilterGray(I_32F, radius, epsilon) else: self._guided_filter = GuidedFilterColor(I_32F, radius, epsilon)
def baseDetailSeparationMedian(I_32F, ksize=5): ksize = 2 * (ksize / 2) + 1 B = to32F(cv2.medianBlur(to8U(I_32F), ksize)) D = I_32F - B return B, D
def _runLayer(self, layer_file): C0_8U = loadRGBA(layer_file) if C0_8U is None: return A_8U = alpha(C0_8U) # if A_8U is None: # return C0_32F = to32F(rgb(C0_8U)) I_32F = luminance(C0_32F) Lab_32F = rgb2Lab(C0_32F) th_specular = 0.2 th_contour = 0.02 th_material = 0.1 E_32F = DoG(I_32F, sigma=2.0) contour = th_contour * np.min(E_32F) - E_32F contour *= 1.0 / np.max(contour) contour = np.clip(contour, 0.0, 1.0) specular = E_32F - th_specular * np.max(E_32F) specular *= 1.0 / np.max(specular) specular = np.clip(specular, 0.0, 1.0) material = rgb(C0_8U) # edge_mask = np.zeros(I_32F.shape, dtype=np.uint8) # edge_mask[contour > 0.0] = 1.0 # material = cv2.inpaint(material, edge_mask, 3, cv2.INPAINT_TELEA) for i in xrange(1): material = cv2.medianBlur(material, ksize=7) # material = th_material * np.max(np.abs(E_32F)) - np.abs(E_32F) # material *= 1.0 / np.max(material) # material = np.clip(material, 0.0, 1.0) # material[material > 0.0] = 1.0 E_32F[E_32F < 0.0] = 0.0 fig, axes = plt.subplots(figsize=(11, 5)) font_size = 15 fig.subplots_adjust(left=0.05, right=0.95, top=0.9, hspace=0.12, wspace=0.05) fig.suptitle(self.name(), fontsize=font_size) num_rows = 1 num_cols = 4 plot_grid = SubplotGrid(num_rows, num_cols) plot_grid.showImage(C0_8U, r'$C$') #plot_grid.showImage(setAlpha(C0_32F, material), r'$Material$') plot_grid.showImage(setAlpha(material, A_8U), r'$Material$') plot_grid.showImage(setAlpha(C0_32F, contour), r'$Contour$') plot_grid.showImage(setAlpha(C0_32F, specular), r'$Specular$') showMaximize()
def slicSegmentation(image, num_segments=600, sigma=5): C_8U = rgb(image) C_32F = to32F(C_8U) segments = slic(C_32F, n_segments=num_segments, sigma=sigma) fig = plt.figure("Superpixels -- %d segments" % (num_segments)) ax = fig.add_subplot(1, 1, 1) ax.imshow(mark_boundaries(C_32F, segments)) plt.axis("off") plt.show()
def filter(self, p): p_32F = to32F(p) if _isGray(p_32F): return self._filterGray(p_32F) cs = p.shape[2] q = np.array(p_32F) for ci in range(cs): q[:, :, ci] = self._filterGray(p_32F[:, :, ci]) return q
def HSVFeatures(image, w_H=1.0, w_S=1.0, w_V=1.0): img_32F = to32F(image) rgb_32F = img_32F[:, :, :3] hsv_32F = rgb2hsv(rgb_32F) h, w, cs = image.shape features = hsv_32F.reshape(h * w, -1) features[:, 0] *= w_H features[:, 1] *= w_S features[:, 2] *= w_V return features
def LabFeatures(image, w_L=1.0, w_a=1.0, w_b=1.0): img_32F = to32F(image) rgb_32F = img_32F[:, :, :3] Lab_32F = rgb2Lab(rgb_32F) h, w, cs = image.shape features = Lab_32F.reshape(h * w, -1) features[:, 0] *= w_L features[:, 1] *= w_a features[:, 2] *= w_b return features
def colorToNormal(C_8U, fill_background=False): rgb_8U = rgb(C_8U) A_8U = alpha(C_8U) C_32F = to32F(rgb_8U) N_32F = 2.0 * C_32F - 1.0 if fill_background: N_32F[A_8U < 10, :] = np.array([0.0, 0.0, 0.0]) N_32F_normalized = normalizeImage(N_32F) return N_32F_normalized
def __init__(self, I, radius=5, epsilon=0.4, scale=4): I_32F = to32F(I) self._I = I_32F h, w = I.shape[:2] I_sub = _downSample(I_32F, scale) self._I_sub = I_sub radius = int(radius / scale) if _isGray(I): self._guided_filter = GuidedFilterGray(I_sub, radius, epsilon) else: self._guided_filter = GuidedFilterColor(I_sub, radius, epsilon)
def filter(self, p): p_32F = to32F(p) shape_original = p.shape[:2] p_sub = _downSample(p_32F, shape=self._I_sub.shape[:2]) if _isGray(p_sub): return self._filterGray(p_sub, shape_original) cs = p.shape[2] q = np.array(p_32F) for ci in range(cs): q[:, :, ci] = self._filterGray(p_sub[:, :, ci], shape_original) return q
def _runImp(self): normal_data = loadNormal(self._data_file) if normal_data is None: return N0_32F, A_8U = normal_data A_32F = to32F(A_8U) L = normalizeVector(np.array([-0.2, 0.3, 0.7])) # C0_32F = ToonShader().diffuseShading(L, N0_32F) C0_32F = LambertShader().diffuseShading(L, N0_32F) self._C0_32F = C0_32F self._loadImage()
def _runLayer(self, layer_file): if layer_file is None: return C0_8U = loadRGBA(layer_file) if C0_8U is None: C0_8U = loadRGB(layer_file) if C0_8U is None: return h, w = C0_8U.shape[:2] w_low = 1024 h_low = w_low * h / w # C0_8U = cv2.resize(C0_8U, (w_low, h_low)) A_8U = alpha(C0_8U) self._A_8U = A_8U C0_32F = to32F(rgb(C0_8U)) if A_8U is not None: C0_32F[A_8U < 0.9 * np.max(A_8U), :] = np.array([0, 0, 0]) self._C0_32F = C0_32F self._loadImage() self._computeBaseDetalSeparation() self._computeInitialDetailNormal() self.computeDetailNormal() self._computeLumoNormal() self.computeInitialNormal() # plt.savefig(self.characterResultFile("BumpNormal.png")) if self._N_b_smooth is not None: self.cleanCharacterResultDir() saveNormal(self.characterResultFile("N_b.png"), self._N_b, A_8U) saveNormal(self.characterResultFile("N_b_smooth.png"), self._N_b_smooth, A_8U) saveNormal(self.characterResultFile("N_d.png"), self._N_d, A_8U) saveNormal(self.characterResultFile("N_d_smooth.png"), self._N_d_smooth, A_8U) saveNormal(self.characterResultFile("N_lumo.png"), self._N_lumo, A_8U) saveNormal(self.characterResultFile("N0_b.png"), self._N0_b_32F, A_8U) saveNormal(self.characterResultFile("N0_d.png"), self._N0_d_32F, A_8U)
def _runImp(self): normal_data = loadNormal(self._data_file) if normal_data is None: return N0_32F, A_8U = normal_data A_32F = to32F(A_8U) L = normalizeVector(np.array([-0.2, 0.3, 0.7])) C0_32F = LambertShader().diffuseShading(L, N0_32F) self._normal_constraint.clear() if os.path.exists(self.constraintFile()): self._normal_constraint.load(self.constraintFile()) self._tool.setImage(setAlpha(C0_32F, A_32F))
def _runLayer(self, layer_file): C0_8U = loadRGBA(layer_file) if C0_8U is None: return A_8U = alpha(C0_8U) if A_8U is None: return C0_32F = to32F(rgb(C0_8U)) I_32F = luminance(C0_32F) N0_32F, A_8U = loadNormal(self.characterResultFile("N0_d.png", data_name="BaseDetailSepration")) Nd_32F, A_8U = loadNormal(self.characterResultFile("N_d_smooth.png", data_name="BaseDetailSepration")) Nb_32F, A_8U = loadNormal(self.characterResultFile("N_b_smooth.png", data_name="BaseDetailSepration")) W_32F = np.array(Nb_32F[:, :, 2]) W_32F = W_32F W_32F[W_32F < 0.95] = 0.0 L = lightEstimation(I_32F, N0_32F, A_8U) # L = lightEstimationByVoting(I_32F, N0_32F, A_8U) L_txt = 0.01 * np.int32(100 * L) L_img = lightSphere(L) fig, axes = plt.subplots(figsize=(11, 5)) font_size = 15 fig.subplots_adjust(left=0.05, right=0.95, top=0.9, hspace=0.12, wspace=0.05) fig.suptitle(self.name(), fontsize=font_size) num_rows = 1 num_cols = 4 plot_grid = SubplotGrid(num_rows, num_cols) plot_grid.showImage(C0_8U, r'$C$') plot_grid.showImage(normalToColor(N0_32F, A_8U), r'$N$') plot_grid.showImage(setAlpha(C0_32F, W_32F), r'$Nd_z$') plot_grid.showImage(L_img, r'$L: [%s, %s, %s]$' %(L_txt[0], L_txt[1], L_txt[2])) showMaximize()
def _runImp(self): normal_data = loadNormal(self._data_file) if normal_data is None: return N_32F, A_8U = normal_data N_32F = trim(N_32F, A_8U) A_8U = trim(A_8U, A_8U) A_32F = to32F(A_8U) L = normalizeVector(np.array([-0.2, 0.3, 0.7])) I_half = half_lambert.diffuse(N_32F, L) I_half = setAlpha(gray2rgb(I_half), A_32F) I_lambert = lambert.diffuse(N_32F, L) I_lambert = setAlpha(gray2rgb(I_lambert), A_32F) fig, axes = plt.subplots(figsize=(11, 5)) font_size = 15 fig.subplots_adjust(left=0.05, right=0.95, top=0.9, hspace=0.05, wspace=0.05) fig.suptitle("Depth From Normal", fontsize=font_size) plt.subplot(1, 4, 1) plt.title(r'Normal: $N$') plt.imshow(normalToColor(N_32F, A_8U)) plt.axis('off') plt.subplot(1, 4, 2) plt.title(r'Half Lambert: $I_h$') plt.imshow(I_half) plt.axis('off') plt.subplot(1, 4, 3) plt.title(r'Lambert: $I_l$') plt.imshow(I_lambert) plt.axis('off') showMaximize()
def _compute(self): C0_8U = self._image C0_32F = to32F(rgb(C0_8U)) Lab_32F = rgb2Lab(C0_32F) I_32F = luminance(C0_32F) h, w = I_32F.shape[:2] edge_mask = np.zeros((h, w), dtype=np.uint8) # E_32F = DoG(Lab_32F, sigma=7.0) # E_32F[E_32F > 0.0] = 0.0 # E_32F = - E_32F # E_32F /= np.max(E_32F) # th_contour = 0.05 # for ci in xrange(3): # edge_area = E_32F[:, :, ci] > th_contour # edge_mask[edge_area] = 255 E_32F = DoG(I_32F, sigma=3.0) h, w = E_32F.shape[:2] E_norm = -np.array(E_32F) E_norm[E_norm < 0.0] = 0.0 th_contour = 0.1 edge_area = E_norm > th_contour * np.max(E_norm) edge_mask[edge_area] = 255 self._edge_mask = edge_mask labels = np.array(edge_mask) mask = np.ones((h + 2, w + 2), dtype=np.uint8) mask[1:-1, 1:-1] = self._edge_mask for label in xrange(1, 3): regionSeeds = np.where(self._edge_mask == 0) if len(regionSeeds[0]) == 0: break p = (regionSeeds[1][0], regionSeeds[0][0]) cv2.floodFill(labels, mask, p, label) self._edge_mask[labels == label] = label self._labels = labels
def _interpolateNormalImage(self, N0_32F, W_32F, A_8U): constraints = [] constraints.append(image_constraints.laplacianConstraints(w_c=0.1)) constraints.append(image_constraints.normalConstraints(W_32F, N0_32F, w_c=3.0)) L = normalizeVector(np.array([-0.2, 0.3, 0.7])) I_32F = luminance(to32F(rgb(self._image))) I_min, I_max = np.min(I_32F), np.max(I_32F) I_32F = (I_32F - I_min) / (I_max - I_min) # constraints.append(image_constraints.brightnessConstraints(L, I_32F, w_c=0.5)) constraints.append(image_constraints.silhouetteConstraints(A_8U, w_c=0.8)) solver_iter = image_solver.solveIterator(constraints, [postNormalize(th=0.0)]) N_32F = np.array(N0_32F) N_32F = image_solver.solveMG(N_32F, solver_iter, iterations=10) N_32F = image_constraints.NxyToNz(N_32F) return N_32F
def stylizedShadingFigure(): target_shapes = ["Man", "Ogre", "Grog", "Vase"] target_shapes = [shapeFile(shape_name) for shape_name in target_shapes] target_colormaps = [1, 5, 10, 3] target_colormaps = [colorMapFile(cmap_id) for cmap_id in target_colormaps] fig, axes = plt.subplots(figsize=(12, 4)) font_size = 15 fig.subplots_adjust(left=0.02, right=0.98, top=0.98, bottom=0.02, hspace=0.1, wspace=0.1) fig.suptitle("", fontsize=font_size) num_rows = 1 num_cols = 4 plot_grid = SubplotGrid(num_rows, num_cols) Ls = [] Ls.append(normalizeVector(np.array([-0.5, 0.3, 0.7]))) Ls.append(normalizeVector(np.array([0.2, -0.35, 0.4]))) Ls.append(normalizeVector(np.array([-0.2, 0.6, 0.3]))) Ls.append(normalizeVector(np.array([-0.2, 0.6, 0.3]))) for shape_file, colormap_file, L in zip(target_shapes, target_colormaps, Ls): N_32F, A_8U = loadNormal(shape_file) M_32F = loadColorMap(colormap_file) C_32F = ColorMapShader(M_32F).diffuseShading(L, N_32F) # C_32F = trim(C_32F, A_8U) # A_8U = trim(A_8U, A_8U) C_32F = setAlpha(C_32F, to32F(A_8U)) # h, w = C_32F.shape[:2] # # h_t = 512 # w_t = w * h_t / h # C_32F = cv2.resize(C_32F, (w_t, h_t)) plot_grid.showImage(C_32F, "", alpha_clip=True) file_path = os.path.join(shapeResultsDir(), "StylizedShading.png") fig.savefig(file_path, transparent=True)
def _runImp(self): normal_data = loadNormal(self._data_file) if normal_data is None: return N0_32F, A_8U = normal_data A_32F = to32F(A_8U) L = normalizeVector(np.array([-0.2, 0.4, 0.7])) C0_32F = LambertShader().diffuseShading(L, N0_32F) I_32F = luminance(C0_32F) br_field = BrightnessField(I_32F, sigma=5.0) I_smooth_32F = br_field.smoothBrightness() dI = br_field.brightnessDifference() gx, gy = br_field.gradients() N_32F = br_field.field() fig, axes = plt.subplots(figsize=(11, 5)) font_size = 15 fig.subplots_adjust(left=0.05, right=0.95, top=0.9, hspace=0.12, wspace=0.05) fig.suptitle(self.name(), fontsize=font_size) num_rows = 2 num_cols = 4 plot_grid = SubplotGrid(num_rows, num_cols) plot_grid.showImage(I_32F, r'$I$') plot_grid.showColorMap(dI, r'$dI$') plot_grid.showColorMap(gx, r'$gx$') plot_grid.showColorMap(gy, r'$gy$') plot_grid.showImage(normalToColor(N_32F, A_8U), r'$N$') plot_grid.showImage(N_32F[:, :, 2], r'$N_z$') showMaximize()
def _runImp(self): normal_data = loadNormal(self._data_file) if normal_data is None: return N0_32F, A_8U = normal_data A_32F = to32F(A_8U) L = normalizeVector(np.array([-0.2, 0.3, 0.7])) #C0_32F = ToonShader().diffuseShading(L, N0_32F) C0_32F = LambertShader().diffuseShading(L, N0_32F) I0_32F = luminance(C0_32F) I0_low_32F = cv2.resize(I0_32F, (256, 256)) A_low_8U = cv2.resize(A_8U, I0_low_32F.shape) D_32F = depthFromGradient(I0_low_32F, A_low_8U) D_32F = cv2.resize(D_32F, I0_32F.shape) N_32F = depthToNormal(D_32F) self._view.setRGBAD(setAlpha(C0_32F, A_32F), D_32F)
def _runImp(self): normal_data = loadNormal(self._data_file) if normal_data is None: return N0_32F, A_8U = normal_data #N0_32F = cv2.resize(N0_32F, (64, 64)) #A_8U = cv2.resize(A_8U, N0_32F.shape[:2]) A_32F = to32F(A_8U) L = normalizeVector(np.array([-0.2, 0.3, 0.7])) C0_32F = ToonShader().diffuseShading(L, N0_32F) # C0_32F = LambertShader().diffuseShading(L, N0_32F) sfs_method = Wu08SFS(L, C0_32F, A_8U) sfs_method.run() N_32F = sfs_method.normal() saveNormal(self.resultFile(self._data_file_name, result_name="Wu08"), N_32F, A_8U)
def _runCharacterImp(self): print self.fullLayerFile() C0_8U = loadRGBA(self.fullLayerFile()) if C0_8U is None: return C0_32F = to32F(C0_8U) print C0_32F h, w = C0_32F.shape[:2] w_low = 512 h_low = w_low * h / w C0_32F = cv2.resize(C0_32F, (w_low, h_low)) self._normal_constraint.clear() if os.path.exists(self.characterConstraintFile()): self._normal_constraint.load(self.characterConstraintFile()) self._tool.setImage(C0_32F)
def slicSampling(image, num_segments=4000, sigma=5): h, w = image.shape[:2] C_32F = to32F(rgb(image)) segments = slic(C_32F, n_segments=num_segments, sigma=sigma) print segments.shape print np.max(segments) num_centers = np.max(segments) + 1 hist_centers = np.zeros((num_centers), dtype=np.float32) centers = np.zeros((num_centers, 3), dtype=np.float32) hist_centers[segments[:, :]] += 1.0 centers[segments[:, :], :] += C_32F[:, :, :3] hist_positive = hist_centers > 0.0 print np.count_nonzero(hist_positive) for ci in xrange(3): centers[hist_positive, ci] /= hist_centers[hist_positive] map_rgb = to8U(centers[segments[:, :], :].reshape(h, w, -1)) # map_rgb = to8U(mark_boundaries(C_32F, segments)) map_image = setAlpha(map_rgb, alpha(image)) return map_image
def relightingFigure(shape_name="Vase", cmap_id=3): num_methods = 3 num_lights = 2 num_rows = num_lights + 1 num_cols = num_methods + 2 w = 15 h = w * num_rows / num_cols fig, axes = plt.subplots(figsize=(w, h)) font_size = 15 fig.subplots_adjust(left=0.02, right=0.98, top=0.96, bottom=0.04, hspace=0.15, wspace=0.1) fig.suptitle("", fontsize=font_size) plot_grid = SubplotGrid(num_rows, num_cols) Lg = normalizeVector(np.array([-0.2, 0.3, 0.5])) Lg_img = lightSphere(Lg) L1 = normalizeVector(np.array([0.0, 0.7, 0.6])) L2 = normalizeVector(np.array([0.3, 0.5, 0.6])) # Ls = [normalizeVector(Lg * (1.0 - t) + t * L1) for t in np.linspace(0.0, 1.0, num_lights) ] Ls = [L1, L2] Ng_data = shapeFile(shape_name) Ng_data = loadNormal(Ng_data) Ng_32F, A_8U = Ng_data N0_file = shapeResultFile(result_name="InitialNormal", data_name=shape_name) N0_data = loadNormal(N0_file) N0_32F, A_8U = N0_data A_8U = cv2.bilateralFilter(A_8U, 0, 5, 2) colormap_file = colorMapFile(cmap_id) M_32F = loadColorMap(colormap_file) C0_32F = ColorMapShader(M_32F).diffuseShading(Lg, Ng_32F) toon_sfs = ToonSFS(Lg, C0_32F, A_8U) toon_sfs.setInitialNormal(N0_32F) toon_sfs.setNumIterations(iterations=50) toon_sfs.setWeights(w_lap=0.2) toon_sfs.run() N_toon = toon_sfs.normal() C_toon = toon_sfs.shading() C_lumo, N_lumo = lumoSFS(C0_32F, Lg, N0_32F, A_8U) C_wu, N_wu = wuSFS(C0_32F, Lg, N0_32F, A_8U) M_lumo = estimatedReflectance(C0_32F, Lg, N_lumo, A_8U) M_wu = estimatedReflectance(C0_32F, Lg, N_wu, A_8U) plot_grid.showImage(Lg_img, "Light direction") plot_grid.showImage(setAlpha(C0_32F, to32F(A_8U)), "Ground-truth") title = "" plot_grid.showImage(setAlpha(C_lumo, to32F(A_8U)), "Lumo") #plot_grid.showColorMap(C_error_lumo, title, v_min=0, v_max=0.1, with_colorbar=True) plot_grid.showImage(setAlpha(C_wu, to32F(A_8U)), "Lambert assumption") #plot_grid.showColorMap(C_error_wu, title, v_min=0, v_max=0.1, with_colorbar=True) plot_grid.showImage(setAlpha(C_toon, to32F(A_8U)), "Our result") #plot_grid.showColorMap(C_error_toon, title, v_min=0, v_max=0.1, with_colorbar=True) for L in Ls: C1 = ColorMapShader(M_32F).diffuseShading(L, Ng_32F) C1_lumo = M_lumo.shading(LdotN(L, N_lumo).flatten()).reshape(C0_32F.shape) C1_wu = M_wu.shading(LdotN(L, N_wu).flatten()).reshape(C0_32F.shape) C1_toon = toon_sfs.relighting(L) plot_grid.showImage(lightSphere(L), "") plot_grid.showImage(setAlpha(C1, to32F(A_8U)), "") title = "" plot_grid.showImage(setAlpha(C1_lumo, to32F(A_8U)), "") #plot_grid.showColorMap(C_error_lumo, title, v_min=0, v_max=0.1, with_colorbar=True) plot_grid.showImage(setAlpha(C1_wu, to32F(A_8U)), "") #plot_grid.showColorMap(C_error_wu, title, v_min=0, v_max=0.1, with_colorbar=True) plot_grid.showImage(setAlpha(C1_toon, to32F(A_8U)), "") # showMaximize() file_path = shapeResultFile("Relighting", "RelightingComparison", file_ext=".png") fig.savefig(file_path, transparent=True)
def relightingVideo(shape_name="Ogre", cmap_id=3): num_methods = 3 num_rows = 1 num_cols = num_methods + 2 num_lights = 120 w = 10 h = 5 fig, axes = plt.subplots(figsize=(w, h)) font_size = 15 fig.subplots_adjust(left=0.02, right=0.98, top=0.96, bottom=0.04, hspace=0.15, wspace=0.1) fig.suptitle("Shading Analysis", fontsize=font_size) plot_grid = SubplotGrid(num_rows, num_cols) Lg = normalizeVector(np.array([-0.2, 0.3, 0.5])) Lg_img = lightSphere(Lg) L1 = normalizeVector(np.array([0.5, 0.5, 0.6])) Ls = [ normalizeVector(Lg * (1.0 - t) + t * L1) for t in np.linspace(0.0, 1.0, num_lights) ] # Ls = [normalizeVector(Lg + 1.0 * np.cos(t) * np.array([1, 0, 0]) + 1.0 * np.sin(t) * np.array([0, 1, 0])) for t in np.linspace(0.0, 1.0, num_lights) ] Ng_data = shapeFile(shape_name) Ng_data = loadNormal(Ng_data) Ng_32F, A_8U = Ng_data N0_file = shapeResultFile(result_name="InitialNormal", data_name=shape_name) N0_data = loadNormal(N0_file) N0_32F, A_8U = N0_data A_8U = cv2.bilateralFilter(A_8U, 0, 5, 2) colormap_file = colorMapFile(cmap_id) M_32F = loadColorMap(colormap_file) C0_32F = ColorMapShader(M_32F).diffuseShading(Lg, Ng_32F) toon_sfs = ToonSFS(Lg, C0_32F, A_8U) toon_sfs.setInitialNormal(N0_32F) toon_sfs.setNumIterations(iterations=100) toon_sfs.setWeights(w_lap=0.2) toon_sfs.run() N_toon = toon_sfs.normal() C_toon = toon_sfs.shading() C_lumo, N_lumo = lumoSFS(C0_32F, Lg, N0_32F, A_8U) C_wu, N_wu = wuSFS(C0_32F, Lg, N0_32F, A_8U) M_lumo = estimatedReflectance(C0_32F, Lg, N_lumo, A_8U) M_wu = estimatedReflectance(C0_32F, Lg, N_wu, A_8U) plot_grid.showImage(Lg_img, "Light direction") plot_grid.showImage(setAlpha(C0_32F, to32F(A_8U)), "Ground-truth") title = "" plot_grid.showImage(setAlpha(C_lumo, to32F(A_8U)), "Lumo") #plot_grid.showColorMap(C_error_lumo, title, v_min=0, v_max=0.1, with_colorbar=True) plot_grid.showImage(setAlpha(C_wu, to32F(A_8U)), "Lambert assumption") #plot_grid.showColorMap(C_error_wu, title, v_min=0, v_max=0.1, with_colorbar=True) plot_grid.showImage(setAlpha(C_toon, to32F(A_8U)), "Our result") #plot_grid.showColorMap(C_error_toon, title, v_min=0, v_max=0.1, with_colorbar=True) images = [] for i in xrange(48): images.append(figure2numpy(fig)) for li, L in enumerate(Ls): print li fig.clear() fig.suptitle("Relighting", fontsize=font_size) plot_grid.setPlot(1, 1) C1 = ColorMapShader(M_32F).diffuseShading(L, Ng_32F) C1_lumo = M_lumo.shading(LdotN(L, N_lumo).flatten()).reshape(C0_32F.shape) C1_wu = M_wu.shading(LdotN(L, N_wu).flatten()).reshape(C0_32F.shape) C1_toon = toon_sfs.relighting(L) plot_grid.showImage(lightSphere(L), "Light direction") plot_grid.showImage(setAlpha(C1, to32F(A_8U)), "Ground-truth") title = "" plot_grid.showImage(setAlpha(C1_lumo, to32F(A_8U)), "Lumo") #plot_grid.showColorMap(C_error_lumo, title, v_min=0, v_max=0.1, with_colorbar=True) plot_grid.showImage(setAlpha(C1_wu, to32F(A_8U)), "Lambert assumption") #plot_grid.showColorMap(C_error_wu, title, v_min=0, v_max=0.1, with_colorbar=True) plot_grid.showImage(setAlpha(C1_toon, to32F(A_8U)), "Our result") images.append(figure2numpy(fig)) file_path = shapeResultFile("Relighting", "Relighting_%s_%s" % (shape_name, cmap_id), file_ext=".wmv") saveVideo(file_path, images)
def rgbFeatures(image): img_32F = to32F(image) features = img_32F[:, :, :3].reshape(-1, 3) return features
def _computeSegmentaiton(self, C0_8U): C_32F = to32F(rgb(C0_8U)) label0 = np.zeros(C_32F.shape[:3], dtype=np.uint8) colors = [] for stroke_set in self._stroke_sets.strokeSets(): print stroke_set.color() color = np.array(stroke_set.color())[:3] colors.append(color) color = np.int32(255 * color) print type(color[0]) for stroke in stroke_set.strokes(): if stroke.empty(): continue points = stroke.points() points = np.int32(points) brush_size = int(stroke.brushSize()) print color cv2.polylines(label0, [points], 0, (color[0], color[1], color[2]), brush_size) colors = np.array(colors) # h, w = label0.shape[:2] # # w_low = 512 # h_low = w_low * h / w # # gauide_filter = GuidedFilter(cv2.resize(C_32F, (w_low, h_low)), radius=11, epsilon=0.05) # # label0 = cv2.resize(label0, (w_low, h_low)) # h, w = label0.shape[:2] # label = np.array(label0) # # dc = np.zeros((len(colors), h * w), dtype=np.float32) # # for i in xrange(5): # label = gauide_filter.filter(label) # label[label0 > 0] = label0[label0 > 0] # # label_flat = label.reshape(-1, 3) # # for ci, color in enumerate(colors): # dc[ci, :] = normVectors(label_flat - color) # # centers = np.argmin(dc, axis=0) # # label_flat = colors[centers] # label = label_flat.reshape(h, w, 3) # kmeans = KMeans(C0_8U, num_colors=20) # centerImage = kmeans.centerImage() # self._view.render(centerImage) # # histImage = kmeans.histImage() # # plt.imshow(histImage) # plt.show() edgeSeg = EdgeBasedSegmentaiton(C0_8U) labels = edgeSeg.labels() plt.imshow(labels) plt.show()
def overviewFigure(): cmap_id = 10 colormap_file = colorMapFile(cmap_id) num_rows = 1 num_cols = 5 w = 10 h = w * num_rows / num_cols fig, axes = plt.subplots(figsize=(w, h)) font_size = 15 fig.subplots_adjust(left=0.02, right=0.98, top=0.96, bottom=0.02, hspace=0.05, wspace=0.05) fig.suptitle("", fontsize=font_size) plot_grid = SubplotGrid(num_rows, num_cols) L = normalizeVector(np.array([-0.4, 0.6, 0.6])) L_img = lightSphere(L) shape_name = "ThreeBox" Ng_data = shapeFile(shape_name) Ng_data = loadNormal(Ng_data) Ng_32F, A_8U = Ng_data N0_file = shapeResultFile(result_name="InitialNormal", data_name=shape_name) N0_data = loadNormal(N0_file) N0_32F, A_8U = N0_data M_32F = loadColorMap(colormap_file) Cg_32F = ColorMapShader(M_32F).diffuseShading(L, Ng_32F) borders=[0.6, 0.8, 0.92] colors = [np.array([0.2, 0.2, 0.4]), np.array([0.3, 0.3, 0.6]), np.array([0.4, 0.4, 0.8]), np.array([0.5, 0.5, 1.0])] #Cg_32F = ToonShader(borders, colors).diffuseShading(L, Ng_32F) #Cg_32F = cv2.GaussianBlur(Cg_32F, (0,0), 2.0) sfs_method = ToonSFS(L, Cg_32F, A_8U) sfs_method.setInitialNormal(N0_32F) sfs_method.setNumIterations(iterations=40) sfs_method.setWeights(w_lap=10.0) sfs_method.run() N_32F = sfs_method.normal() I_32F = np.float32(np.clip(LdotN(L, N_32F), 0.0, 1.0)) I0_32F = np.float32(np.clip(LdotN(L, N0_32F), 0.0, 1.0)) C_32F = sfs_method.shading() C0_32F = sfs_method.initialShading() M_32F = sfs_method.colorMap().mapImage() L1 = normalizeVector(np.array([0.0, 0.6, 0.6])) L1_img = lightSphere(L1) C1_32F = sfs_method.relighting(L1) L2 = normalizeVector(np.array([0.5, 0.8, 0.6])) L2_img = lightSphere(L2) C2_32F = sfs_method.relighting(L2) N_sil = silhouetteNormal(A_8U, sigma=7.0) N_sil[:, :, 2] = N_sil[:, :, 2] ** 10.0 N_sil = normalizeImage(N_sil) A_sil = 1.0 - N_sil[:, :, 2] A_sil = to8U(A_sil) N_xy = N_sil[:, :, 0] ** 2 + N_sil[:, :, 1] ** 2 A_sil[N_xy < 0.1] = 0 title = "" plot_grid.showImage(setAlpha(Cg_32F, to32F(A_8U)), title) plot_grid.showImage(normalToColor(N0_32F, A_8U), title) plot_grid.showImage(setAlpha(C0_32F, to32F(A_8U)), title) plot_grid.showImage(normalToColor(N_32F, A_8U), title) plot_grid.showImage(setAlpha(C_32F, to32F(A_8U)), title) # plot_grid.showImage(normalToColor(Ng_32F, A_8U), title) #showMaximize() file_path = shapeResultFile("Overview", "Overview") fig.savefig(file_path, transparent=True) file_path = shapeResultFile("Overview", "Cg") saveRGBA(file_path, setAlpha(Cg_32F, to32F(A_8U))) file_path = shapeResultFile("Overview", "L") saveRGB(file_path, gray2rgb(to8U(L_img))) file_path = shapeResultFile("Overview", "L1") saveRGB(file_path, gray2rgb(to8U(L1_img))) file_path = shapeResultFile("Overview", "L2") saveRGB(file_path, gray2rgb(to8U(L2_img))) file_path = shapeResultFile("Overview", "N0") saveNormal(file_path, N0_32F, A_8U) file_path = shapeResultFile("Overview", "N_sil") saveNormal(file_path, N_sil, A_sil) file_path = shapeResultFile("Overview", "N") saveNormal(file_path, N_32F, A_8U) file_path = shapeResultFile("Overview", "C0") saveRGBA(file_path, setAlpha(C0_32F, to32F(A_8U))) file_path = shapeResultFile("Overview", "C") saveRGBA(file_path, setAlpha(C_32F, to32F(A_8U))) file_path = shapeResultFile("Overview", "C1") saveRGBA(file_path, setAlpha(C1_32F, to32F(A_8U))) file_path = shapeResultFile("Overview", "C2") saveRGBA(file_path, setAlpha(C2_32F, to32F(A_8U))) file_path = shapeResultFile("Overview", "I") saveRGBA(file_path, setAlpha(gray2rgb(I_32F), to32F(A_8U))) file_path = shapeResultFile("Overview", "I0") saveRGBA(file_path, setAlpha(gray2rgb(I0_32F), to32F(A_8U))) file_path = shapeResultFile("Overview", "M") saveRGB(file_path, M_32F)
def _runImp(self): normal_data = loadNormal(self._data_file) if normal_data is None: return N0_32F, A_8U = normal_data A_32F = to32F(A_8U) N0_32F = trim(N0_32F, A_8U) A_32F = trim(A_32F, A_8U) A_8U = trim(A_8U, A_8U) L = normalizeVector(np.array([0.5, -0.2, 0.7])) C0_32F = ToonShader().diffuseShading(L, N0_32F) I0_32F = luminance(C0_32F) I_min, I_max = np.min(I0_32F), np.max(I0_32F) I_scale = (I0_32F - I_min) / (I_max - I_min) I_L = cv2.Laplacian(cv2.GaussianBlur(I_scale, (0, 0), 31.0), cv2.CV_32F, ksize=1) I_L_avg = np.average(np.abs(I_L)) Ix = cv2.Sobel(I0_32F, cv2.CV_64F, 1, 0, ksize=1) Ix = cv2.GaussianBlur(Ix, (0, 0), 3.0) Ixx = cv2.Sobel(Ix, cv2.CV_64F, 1, 0, ksize=1) Ixx = cv2.GaussianBlur(Ixx, (0, 0), 5.0) Iy = -cv2.Sobel(I0_32F, cv2.CV_64F, 0, 1, ksize=1) Iy = cv2.GaussianBlur(Iy, (0, 0), 3.0) Iyy = -cv2.Sobel(Iy, cv2.CV_64F, 0, 1, ksize=1) Iyy = cv2.GaussianBlur(Iyy, (0, 0), 5.0) fig, axes = plt.subplots(figsize=(11, 5)) font_size = 15 fig.subplots_adjust(left=0.05, right=0.95, top=0.9, hspace=0.12, wspace=0.05) fig.suptitle(self.name(), fontsize=font_size) num_rows = 2 num_cols = 5 plot_grid = SubplotGrid(num_rows, num_cols) Nx = cv2.Sobel(N0_32F[:, :, 0], cv2.CV_64F, 1, 0, ksize=1) Nx = cv2.GaussianBlur(Nx, (0, 0), 3.0) Nxx = cv2.Sobel(Nx, cv2.CV_64F, 1, 0, ksize=1) Nxx = cv2.GaussianBlur(Nxx, (0, 0), 5.0) Ny = -cv2.Sobel(N0_32F[:, :, 1], cv2.CV_64F, 0, 1, ksize=1) Ny = cv2.GaussianBlur(Ny, (0, 0), 3.0) Nyy = -cv2.Sobel(Ny, cv2.CV_64F, 0, 1, ksize=1) Nyy = cv2.GaussianBlur(Nyy, (0, 0), 5.0) Nz_L = cv2.Laplacian(cv2.GaussianBlur(N0_32F[:, :, 2], (0, 0), 5.0), cv2.CV_32F, ksize=5) Nz_L_avg = np.average(np.abs(Nz_L)) Nz_L *= 1.0 / Nz_L_avg I_L *= 1.0 / I_L_avg print I_L_avg, Nz_L_avg Nz_L = np.clip(Nz_L, -5.0, 5.0) I_L = np.clip(I_L, -5.0, 5.0) plot_grid.showColorMap(N0_32F[:, :, 0], r'$N_{x}$', v_min=-0.01, v_max=0.01) plot_grid.showColorMap(N0_32F[:, :, 1], r'$N_{y}$', v_min=-0.01, v_max=0.01) plot_grid.showColorMap(Nx, r'$N_{xx}$', v_min=-0.01, v_max=0.01) plot_grid.showColorMap(Ny, r'$N_{yy}$', v_min=-0.01, v_max=0.01) plot_grid.showColorMap(Nz_L, r'$Nz_L$') #plot_grid.showColorMap(Nx + Ny, r'$N_{xx} + N_{yy}$', v_min=-0.01, v_max=0.01) # Ixx[Ixx>0] = 1.0 # Ixx[Ixx<0] = -1.0 # Iyy[Iyy>0] = 1.0 # Iyy[Iyy<0] = -1.0 plot_grid.showColorMap(-Ix, r'$I_{x}$', v_min=-0.001, v_max=0.001) plot_grid.showColorMap(-Iy, r'$I_{y}$', v_min=-0.001, v_max=0.001) plot_grid.showColorMap(-Ixx, r'$I_{xx}$', v_min=-0.001, v_max=0.001) plot_grid.showColorMap(-Iyy, r'$I_{yy}$', v_min=-0.001, v_max=0.001) plot_grid.showColorMap(I_L, r'$I_L$') #plot_grid.showColorMap(-Ixx - Iyy, r'$I_{xx} + I_{yy}$', v_min=-0.01, v_max=0.01) #plot_grid.showColorMap(Iy, r'$I_{y}$') showMaximize()
def _runLayer(self, layer_file): C0_8U = loadRGBA(layer_file) if C0_8U is None: return A_8U = alpha(C0_8U) if A_8U is None: return A_32F = to32F(A_8U) C0_32F = to32F(rgb(C0_8U)) I0_32F = luminance(C0_32F) initial_normals = ["N_lumo.png", "N0_d.png"] layer_name = os.path.splitext(os.path.basename(layer_file))[0] for initial_normal in initial_normals: N0_32F, AN_8U = loadNormal( self.characterResultFile(initial_normal, data_name="BaseDetailSepration")) N_32F, L, C_32F, M = self._runSFS(C0_32F, A_8U, N0_32F, AN_8U) L_img = lightSphere(L) M_img = M.mapImage() fig, axes = plt.subplots(figsize=(11, 5)) font_size = 15 fig.subplots_adjust(left=0.02, right=0.98, top=0.9, hspace=0.12, wspace=0.02) fig.suptitle(self.name(), fontsize=font_size) num_rows = 1 num_cols = 4 plot_grid = SubplotGrid(num_rows, num_cols) plot_grid.showImage(normalToColor(N0_32F, A_8U), r'Initial Normal: $N_0$') plot_grid.showImage(normalToColor(N_32F, A_8U), r'Estimated Normal: $N$') plot_grid.showImage(C0_8U, r'Shading: $C_0$') plot_grid.showImage(setAlpha(C_32F, A_32F), r'Recovered Shading: $C$') out_file_path = self.characterResultFile("ToonSFS" + initial_normal, layer_name=layer_name) plt.savefig(out_file_path) N_trim = trim(N_32F, A_8U) N0_trim = trim(N0_32F, A_8U) C0_trim = trim(C0_32F, A_8U) A_trim = trim(A_8U, A_8U) out_file_path = self.characterResultFile(initial_normal, layer_name=layer_name) saveNormal(out_file_path, N_trim, A_trim) images = self._relightingImages(N_trim, A_trim, M) initial_normal_name = os.path.splitext(initial_normal)[0] video_name = "Relighting" + initial_normal_name + ".wmv" out_file_path = self.characterResultFile(video_name, layer_name=layer_name) saveVideo(out_file_path, images) images = self._relightingOffsetImages(L, C0_trim, N0_trim, A_trim, M) video_name = "RelightingOffset" + initial_normal_name + ".wmv" out_file_path = self.characterResultFile(video_name, layer_name=layer_name) saveVideo(out_file_path, images)