diff --git a/camera_calibration.py b/camera_calibration.py
new file mode 100644
index 0000000..4ac3949
--- /dev/null
+++ b/camera_calibration.py
@@ -0,0 +1,66 @@
+import numpy as np
+import cv2
+
+# termination criteria
+criteria = (cv2.TERM_CRITERIA_EPS + cv2.TERM_CRITERIA_MAX_ITER, 30, 0.001)
+# prepare object points, like (0,0,0), (1,0,0), (2,0,0) ....,(6,5,0)
+objp = np.zeros((6*7, 3), np.float32)
+objp[:, :2] = np.mgrid[0:7, 0:6].T.reshape(-1, 2)
+# Arrays to store object points and image points from all the images.
+objpoints = []  # 3d point in real world space
+imgpoints = []  # 2d points in image plane.
+
+w = 1280
+h = 720
+
+target_size = 720
+
+s_x = (w//2-target_size//2)
+e_x = (w//2+target_size//2)
+s_y = 0
+e_y = target_size
+
+cap = cv2.VideoCapture(0)
+cap.set(cv2.CAP_PROP_FRAME_WIDTH, 1280)
+cap.set(cv2.CAP_PROP_FRAME_HEIGHT, 720)
+
+while cap.isOpened():
+    success, img = cap.read()
+    if not success:
+        print("Ignoring empty camera frame.")
+        # If loading a video, use 'break' instead of 'continue'.
+        continue
+
+    img = img[s_y:e_y, s_x:e_x]
+    img = cv2.resize(img, (1080, 1080))
+    # print(img.shape)
+
+    width = img.shape[0]
+    height = img.shape[1]
+
+    gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
+    # Find the chess board corners
+    ret, corners = cv2.findChessboardCorners(gray, (7, 6), None)
+    # If found, add object points, image points (after refining them)
+    if ret == True:
+        objpoints.append(objp)
+        corners2 = cv2.cornerSubPix(
+            gray, corners, (11, 11), (-1, -1), criteria)
+        imgpoints.append(corners)
+        # Draw and display the corners
+        cv2.drawChessboardCorners(img, (7, 6), corners2, ret)
+    cv2.imshow('img', img)
+
+    if cv2.waitKey(5) & 0xFF == 27:
+        break
+
+cv2.destroyAllWindows()
+
+ret, mtx, dist, rvecs, tvecs = cv2.calibrateCamera(
+    objpoints, imgpoints, gray.shape[::-1], None, None)
+
+print(f"width:{width}")
+print(f"height:{height}")
+print(f"ret:{ret}")
+print(f"mtx:{mtx}")
+print(f"dist:{dist}")
diff --git a/environment.yml b/environment.yml
new file mode 100644
index 0000000..f183d42
--- /dev/null
+++ b/environment.yml
@@ -0,0 +1,42 @@
+name: 360_vdo_reconstruck
+channels:
+  - defaults
+dependencies:
+  - _libgcc_mutex=0.1
+  - _openmp_mutex=4.5
+  - ca-certificates=2022.4.26
+  - certifi=2021.10.8
+  - ld_impl_linux-64=2.35.1
+  - libffi=3.3
+  - libgcc-ng=9.3.0
+  - libgomp=9.3.0
+  - libstdcxx-ng=9.3.0
+  - ncurses=6.3
+  - openssl=1.1.1o
+  - pip=21.2.2
+  - python=3.7.13
+  - readline=8.1.2
+  - setuptools=61.2.0
+  - sqlite=3.38.3
+  - tk=8.6.11
+  - wheel=0.37.1
+  - xz=5.2.5
+  - zlib=1.2.12
+  - pip:
+    - cached-property==1.5.2
+    - chumpy==0.70
+    - cycler==0.11.0
+    - cython==0.29.28
+    - h5py==3.6.0
+    - kiwisolver==1.4.2
+    - matplotlib==2.2.5
+    - numpy==1.21.6
+    - opencv-python==4.5.5.64
+    - opendr==0.77
+    - pyparsing==3.0.8
+    - python-dateutil==2.8.2
+    - pytz==2022.1
+    - scipy==1.7.3
+    - six==1.16.0
+    - tqdm==4.64.0
+    - typing-extensions==4.2.0
diff --git a/hdf5_read.py b/hdf5_read.py
new file mode 100644
index 0000000..17bc1e9
--- /dev/null
+++ b/hdf5_read.py
@@ -0,0 +1,17 @@
+import h5py
+import numpy as np
+filename = "reconstructed_poses_noratap.hdf5"
+
+with h5py.File(filename, "r") as f:
+    # List all groups
+    print("Keys: %s" % f.keys())
+    a_group_key = list(f.keys())[1]
+
+    # Get the data
+    data = list(f[a_group_key])
+    pose_data = data[0].reshape((24, 3))
+
+    for i in range(0, 24):
+        print(f"{i}:{np.degrees(pose_data[i])}")
+
+    print(len(data[0]))
diff --git a/lib/rays.py b/lib/rays.py
index 0e69c85..96582a1 100644
--- a/lib/rays.py
+++ b/lib/rays.py
@@ -27,8 +27,14 @@ def unpose_and_select_rays(rays, Vi, smpl, rn_b, rn_m):
     v_ids = visible_boundary_edge_verts(rn_b, rn_m)
     verts = smpl.r[v_ids]
 
+    print(rn_b)
+    print(rn_m)
+    print(v_ids)
+    print(verts)
+
     n, m = plucker(rays)
-    dist = np.linalg.norm(np.cross(verts.reshape(-1, 1, 3), n, axisa=2, axisb=1) - m, axis=2)
+    dist = np.linalg.norm(np.cross(verts.reshape(-1, 1, 3),
+                          n, axisa=2, axisb=1) - m, axis=2)
 
     ray_matches = np.argmin(dist, axis=0)
     vert_matches = np.argmin(dist, axis=1)
@@ -38,27 +44,35 @@ def unpose_and_select_rays(rays, Vi, smpl, rn_b, rn_m):
     M = Vi[:, :, v_ids]
     T = smpl.v_posevariation[v_ids].r
 
-    tmp0 = M[:, :, ray_matches] * np.hstack((rays[:, 0], np.ones((rays.shape[0], 1)))).T.reshape(1, 4, -1)
-    tmp1 = M[:, :, ray_matches] * np.hstack((rays[:, 1], np.ones((rays.shape[0], 1)))).T.reshape(1, 4, -1)
+    tmp0 = M[:, :, ray_matches] * \
+        np.hstack((rays[:, 0], np.ones((rays.shape[0], 1)))
+                  ).T.reshape(1, 4, -1)
+    tmp1 = M[:, :, ray_matches] * \
+        np.hstack((rays[:, 1], np.ones((rays.shape[0], 1)))
+                  ).T.reshape(1, 4, -1)
 
     rays_u_r[:, 0] = np.sum(tmp0, axis=1).T[:, :3] - T[ray_matches]
     rays_u_r[:, 1] = np.sum(tmp1, axis=1).T[:, :3] - T[ray_matches]
 
     rays_u_v = np.zeros_like(rays[vert_matches])
 
-    tmp0 = M * np.hstack((rays[vert_matches, 0], np.ones((verts.shape[0], 1)))).T.reshape(1, 4, -1)
-    tmp1 = M * np.hstack((rays[vert_matches, 1], np.ones((verts.shape[0], 1)))).T.reshape(1, 4, -1)
+    tmp0 = M * np.hstack((rays[vert_matches, 0],
+                         np.ones((verts.shape[0], 1)))).T.reshape(1, 4, -1)
+    tmp1 = M * np.hstack((rays[vert_matches, 1],
+                         np.ones((verts.shape[0], 1)))).T.reshape(1, 4, -1)
 
     rays_u_v[:, 0] = np.sum(tmp0, axis=1).T[:, :3] - T
     rays_u_v[:, 1] = np.sum(tmp1, axis=1).T[:, :3] - T
 
-    valid_rays = dist[np.vstack((ray_matches, range(dist.shape[1]))).tolist()] < 0.12
-    valid_verts = dist[np.vstack((range(dist.shape[0]), vert_matches)).tolist()] < 0.03
+    valid_rays = dist[np.vstack(
+        (ray_matches, range(dist.shape[1]))).tolist()] < 0.12
+    valid_verts = dist[np.vstack(
+        (range(dist.shape[0]), vert_matches)).tolist()] < 0.03
 
     ray_matches = ray_matches[valid_rays]
 
     return np.concatenate((v_ids[ray_matches], v_ids[valid_verts])), \
-           np.concatenate((rays_u_r[valid_rays], rays_u_v[valid_verts]))
+        np.concatenate((rays_u_r[valid_rays], rays_u_v[valid_verts]))
 
 
 def rays_from_points(points, camera):
@@ -73,9 +87,12 @@ def rays_from_points(points, camera):
 def rays_from_silh(mask, camera):
 
     if cv2.__version__[0] == '2':
-        contours, _ = cv2.findContours(mask, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE)
+        contours, _ = cv2.findContours(
+            mask, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE)
     else:
-        _, contours, _ = cv2.findContours(mask, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE)
+        #_, contours, _ = cv2.findContours(mask, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE)
+        contours, _ = cv2.findContours(
+            mask, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE)
 
     silh = np.zeros_like(mask)
 
@@ -92,7 +109,8 @@ def ray_objective(f, sigma, base_smpl, camera, vis_rn_b, vis_rn_m):
     base_smpl.pose[:] = f.pose
     camera.t[:] = f.trans
 
-    f.v_ids, f.rays_u = unpose_and_select_rays(f.rays, f.Vi, base_smpl, vis_rn_b, vis_rn_m)
+    f.v_ids, f.rays_u = unpose_and_select_rays(
+        f.rays, f.Vi, base_smpl, vis_rn_b, vis_rn_m)
     f.verts = base_smpl.v_shaped_personal[f.v_ids]
     f.dist = distance_function(f.rays_u, f.verts)
 
diff --git a/mask_numpy2pic.py b/mask_numpy2pic.py
new file mode 100644
index 0000000..a967993
--- /dev/null
+++ b/mask_numpy2pic.py
@@ -0,0 +1,30 @@
+import numpy as np
+import cv2
+import h5py
+import argparse
+
+parser = argparse.ArgumentParser()
+parser.add_argument('input', type=str)
+parser.add_argument('output', type=str)
+
+args = parser.parse_args()
+
+input_file = args.input
+output_folder = args.output
+
+
+with h5py.File(input_file, "r") as f:
+    # List all groups
+    print("Keys: %s" % f.keys())
+    a_group_key = list(f.keys())[0]
+
+    # Get the data
+    data = list(f[a_group_key])
+    for frame, mask in enumerate(data):
+        print(frame)
+        print(mask.shape)
+        white_img = np.ones((1080, 1080), dtype='uint8')*255
+        mask = cv2.bitwise_and(white_img, white_img, mask=mask)
+        print(mask.dtype)
+        #mask *= 255
+        cv2.imwrite(f"{output_folder}/{frame}.png", mask)
diff --git a/mask_person_by_bg_subtrack.py b/mask_person_by_bg_subtrack.py
new file mode 100644
index 0000000..5c6c82a
--- /dev/null
+++ b/mask_person_by_bg_subtrack.py
@@ -0,0 +1,72 @@
+import cv2
+import numpy as np
+
+vdo_file = "E:/CMU Med Data Analysis/3d_reconstruck/dataset/input/Noratap/noratap_360.mp4"
+bg_file = "E:/CMU Med Data Analysis/3d_reconstruck/dataset/input/Noratap/bg.jpg"
+
+mask_out_path = "E:/CMU Med Data Analysis/3d_reconstruck/dataset/input/Noratap/mask"
+
+# Set Up Vdo Write
+vdo_name = "E:/CMU Med Data Analysis/3d_reconstruck/dataset/input/Noratap/noratap_360_1080.avi"
+print("VDO Save ", vdo_name)
+vdo_out = cv2.VideoWriter(vdo_name, cv2.VideoWriter_fourcc(
+    *'DIVX'), 25, (1080, 1080))
+
+cap = cv2.VideoCapture(vdo_file)
+img_bg = cv2.imread(bg_file)
+
+w = img_bg.shape[0]
+h = img_bg.shape[1]
+
+target_size = 720
+
+s_x = (w//2-target_size//2) + 350
+e_x = (w//2+target_size//2) + 350
+s_y = 0
+e_y = target_size
+
+img_bg = img_bg[s_y:e_y, s_x:e_x]
+
+frame = 0
+
+if (cap.isOpened() == False):
+    print("Error opening video stream or file")
+
+while(cap.isOpened()):
+    ret, img = cap.read()
+    if ret == True:
+
+        img = img[s_y:e_y, s_x:e_x]
+
+        cv2.imshow('Img', img)
+
+        # bg Subtrack
+        img_mask = cv2.subtract(img_bg, img)
+        img_mask = cv2.cvtColor(img_mask, cv2.COLOR_BGR2GRAY)
+
+       # img_mask = cv2.equalizeHist(img_mask)
+
+        ret_th, img_mask = cv2.threshold(img_mask, 30, 255, cv2.THRESH_BINARY)
+
+        # img_mask = cv2.dilate(img_mask, np.ones((10, 10)))
+
+        cv2.imshow('Mask', img_mask)
+
+        img_mask = cv2.resize(img_mask, (1080, 1080))
+        img = cv2.resize(img, (1080, 1080))
+
+        # Save Img
+        cv2.imwrite(f"{mask_out_path}/{frame}.png", img_mask)
+        vdo_out.write(img)
+
+        frame += 1
+
+        if cv2.waitKey(25) & 0xFF == ord('q'):
+            break
+
+    else:
+        break
+
+vdo_out.release()
+cap.release()
+cv2.destroyAllWindows()
diff --git a/models/bodyparts.py b/models/bodyparts.py
index 65c0b7a..e5b65e4 100644
--- a/models/bodyparts.py
+++ b/models/bodyparts.py
@@ -1,7 +1,7 @@
 #!/usr/bin/env python2
 # -*- coding: utf-8 -*-
 
-import cPickle as pkl
+import pickle as pkl
 import numpy as np
 
 _cache = None
@@ -12,7 +12,9 @@ def get_bodypart_vertex_ids():
 
     if _cache is None:
         with open('assets/bodyparts.pkl', 'rb') as fp:
-            _cache = pkl.load(fp)
+            u = pkl._Unpickler(fp)
+            u.encoding = 'latin1'
+            _cache = u.load()
 
     return _cache
 
diff --git a/models/smpl.py b/models/smpl.py
index 8233217..5c36f86 100644
--- a/models/smpl.py
+++ b/models/smpl.py
@@ -3,7 +3,7 @@
 
 import chumpy as ch
 import numpy as np
-import cPickle as pkl
+import pickle as pkl
 import scipy.sparse as sp
 from chumpy.ch import Ch
 from vendor.smpl.posemapper import posemap, Rodrigues
@@ -73,8 +73,10 @@ def on_changed(self, which):
     def _set_up(self):
         self.v_shaped = self.shapedirs.dot(self.betas) + self.v_template
         self.v_shaped_personal = self.v_shaped + self.v_personal
-        self.J = ch.sum(self.J_regressor.T.reshape(-1, 1, 24) * self.v_shaped.reshape(-1, 3, 1), axis=0).T
-        self.v_posevariation = self.posedirs.dot(posemap(self.bs_type)(self.pose))
+        self.J = ch.sum(self.J_regressor.T.reshape(-1, 1, 24)
+                        * self.v_shaped.reshape(-1, 3, 1), axis=0).T
+        self.v_posevariation = self.posedirs.dot(
+            posemap(self.bs_type)(self.pose))
         self.v_poseshaped = self.v_shaped_personal + self.v_posevariation
 
         self.A, A_global = self._global_rigid_transformation()
@@ -83,24 +85,31 @@ def _set_up(self):
 
         self.V = self.A.dot(self.weights.T)
 
-        rest_shape_h = ch.hstack((self.v_poseshaped, ch.ones((self.v_poseshaped.shape[0], 1))))
-        self.v_posed = ch.sum(self.V.T * rest_shape_h.reshape(-1, 4, 1), axis=1)[:, :3]
+        rest_shape_h = ch.hstack(
+            (self.v_poseshaped, ch.ones((self.v_poseshaped.shape[0], 1))))
+        self.v_posed = ch.sum(
+            self.V.T * rest_shape_h.reshape(-1, 4, 1), axis=1)[:, :3]
         self.v = self.v_posed + self.trans
 
     def _global_rigid_transformation(self):
         results = {}
         pose = self.pose.reshape((-1, 3))
-        parent = {i: self.kintree_table[0, i] for i in range(1, self.kintree_table.shape[1])}
+        parent = {i: self.kintree_table[0, i]
+                  for i in range(1, self.kintree_table.shape[1])}
 
-        with_zeros = lambda x: ch.vstack((x, ch.array([[0.0, 0.0, 0.0, 1.0]])))
-        pack = lambda x: ch.hstack([ch.zeros((4, 3)), x.reshape((4, 1))])
+        def with_zeros(x): return ch.vstack(
+            (x, ch.array([[0.0, 0.0, 0.0, 1.0]])))
 
-        results[0] = with_zeros(ch.hstack((Rodrigues(pose[0, :]), self.J[0, :].reshape((3, 1)))))
+        def pack(x): return ch.hstack([ch.zeros((4, 3)), x.reshape((4, 1))])
+
+        results[0] = with_zeros(
+            ch.hstack((Rodrigues(pose[0, :]), self.J[0, :].reshape((3, 1)))))
 
         for i in range(1, self.kintree_table.shape[1]):
             results[i] = results[parent[i]].dot(with_zeros(ch.hstack((
                 Rodrigues(pose[i, :]),      # rotation around bone endpoint
-                (self.J[i, :] - self.J[parent[i], :]).reshape((3, 1))     # bone
+                (self.J[i, :] - self.J[parent[i], :]
+                 ).reshape((3, 1))     # bone
             ))))
 
         results = [results[i] for i in sorted(results.keys())]
@@ -162,7 +171,8 @@ def joints_coco(smpl):
 
 
 def model_params_in_camera_coords(trans, pose, J0, camera_t, camera_rt):
-    root = Rodrigues(np.matmul(Rodrigues(camera_rt).r, Rodrigues(pose[:3]).r)).r.reshape(-1)
+    root = Rodrigues(np.matmul(Rodrigues(camera_rt).r,
+                     Rodrigues(pose[:3]).r)).r.reshape(-1)
     pose[:3] = root
 
     trans = (Rodrigues(camera_rt).dot(J0 + trans) - J0 + camera_t).r
@@ -171,7 +181,8 @@ def model_params_in_camera_coords(trans, pose, J0, camera_t, camera_rt):
 
 
 if __name__ == '__main__':
-    smpl = Smpl(model='../vendor/smpl/models/basicModel_f_lbs_10_207_0_v1.0.0.pkl')
+    smpl = Smpl(
+        model='../vendor/smpl/models/basicModel_f_lbs_10_207_0_v1.0.0.pkl')
     smpl.pose[:] = np.random.randn(72) * .2
     smpl.pose[0] = np.pi
     # smpl.v_personal[:] = np.random.randn(*smpl.shape) / 500.
@@ -204,6 +215,6 @@ def model_params_in_camera_coords(trans, pose, J0, camera_t, camera_rt):
     import cv2
 
     cv2.imshow('render_SMPL', rn.r)
-    print ('..Print any key while on the display window')
+    print('..Print any key while on the display window')
     cv2.waitKey(0)
     cv2.destroyAllWindows()
diff --git a/opendr b/opendr
new file mode 160000
index 0000000..bc16a6a
--- /dev/null
+++ b/opendr
@@ -0,0 +1 @@
+Subproject commit bc16a6a51771d6e062d088ba5cede66649b7c7ec
diff --git a/plk_read.py b/plk_read.py
new file mode 100644
index 0000000..45b95ed
--- /dev/null
+++ b/plk_read.py
@@ -0,0 +1,9 @@
+import pickle as pkl
+
+plk_file = 'camera.pkl'
+
+with open(plk_file, 'rb') as fp:
+    u = pkl._Unpickler(fp)
+    u.encoding = 'latin1'
+    data = u.load()
+    print(data)
diff --git a/prepare_data/create_camera.py b/prepare_data/create_camera.py
index 0505ded..7c4cd04 100644
--- a/prepare_data/create_camera.py
+++ b/prepare_data/create_camera.py
@@ -3,7 +3,7 @@
 
 import numpy as np
 import argparse
-import cPickle as pkl
+import pickle as pkl
 
 """
 
@@ -18,9 +18,12 @@
 parser.add_argument('out', type=str, help="Output file (.pkl)")
 parser.add_argument('width', type=int, help="Frame width in px")
 parser.add_argument('height', type=int, help="Frame height in px")
-parser.add_argument('-f', type=float, nargs='*', help="Focal length in px (2,)")
-parser.add_argument('-c', type=float, nargs='*', help="Principal point in px (2,)")
-parser.add_argument('-k', type=float, nargs='*', help="Distortion coefficients (5,)")
+parser.add_argument('-f', type=float, nargs='*',
+                    help="Focal length in px (2,)")
+parser.add_argument('-c', type=float, nargs='*',
+                    help="Principal point in px (2,)")
+parser.add_argument('-k', type=float, nargs='*',
+                    help="Distortion coefficients (5,)")
 
 args = parser.parse_args()
 
diff --git a/requirements.txt b/requirements.txt
index 8df10d5..c94a25a 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -1,7 +1,18 @@
-opendr>=0.76
-numpy>=1.11.0
-scipy>=0.19.0
-matplotlib==2.*
-h5py==2.7.*
-tqdm==4.*
-chumpy>=0.67
\ No newline at end of file
+cached-property==1.5.2
+certifi==2021.10.8
+chumpy==0.70
+cycler==0.11.0
+Cython==0.29.28
+h5py==3.6.0
+kiwisolver==1.4.2
+matplotlib==2.2.5
+numpy==1.21.6
+opencv-python==4.5.5.64
+opendr==0.77
+pyparsing==3.0.8
+python-dateutil==2.8.2
+pytz==2022.1
+scipy==1.7.3
+six==1.16.0
+tqdm==4.64.0
+typing_extensions==4.2.0
diff --git a/resample_hdf5.py b/resample_hdf5.py
new file mode 100644
index 0000000..48259fc
--- /dev/null
+++ b/resample_hdf5.py
@@ -0,0 +1,22 @@
+import h5py
+filename = "masks"
+
+with h5py.File(f"{filename}.hdf5", "r") as f:
+    # List all groups
+    print("Keys: %s" % f.keys())
+    a_group_key = list(f.keys())[0]
+
+    # Get the data
+    data = list(f[a_group_key])
+
+    # Resample 25 fps to 10 fps
+    # 5 = (5/25)*25
+    resample_result = []
+    for i in range(0, len(data), 5):
+        resample_result.append(data[i])
+
+    print(f"Resample Size:{len(resample_result)}")
+
+    # Write data to HDF5
+    with h5py.File(f"{filename}_resample.hdf5", "w") as data_file:
+        data_file.create_dataset(a_group_key, data=resample_result)
diff --git a/run_step1.sh b/run_step1.sh
index c1009f8..aa0af9d 100755
--- a/run_step1.sh
+++ b/run_step1.sh
@@ -9,7 +9,7 @@ SUBJ="$1"
 OUT="$2"
 
 if [[ $SUBJ = *"female"* ]]; then
-  MODEL='--model vendor/smpl/models/basicModel_f_lbs_10_207_0_v1.0.0.pkl'
+  MODEL='--model vendor/smpl/models/basicModel_m_lbs_10_207_0_v1.0.0.pkl'
 fi
 
 python step1_pose.py $SUBJ/keypoints.hdf5 $SUBJ/masks.hdf5 $SUBJ/camera.pkl $OUT/reconstructed_poses.hdf5 $MODEL ${@:3}
\ No newline at end of file
diff --git a/run_step2.sh b/run_step2.sh
index 9c77cef..a574d9f 100755
--- a/run_step2.sh
+++ b/run_step2.sh
@@ -9,7 +9,7 @@ SUBJ="$1"
 OUT="$2"
 
 if [[ $SUBJ = *"female"* ]]; then
-  MODEL='--model vendor/smpl/models/basicModel_f_lbs_10_207_0_v1.0.0.pkl'
+  MODEL='--model vendor/smpl/models/basicModel_m_lbs_10_207_0_v1.0.0.pkl'
 fi
 
-python step2_consensus.py $SUBJ/reconstructed_poses.hdf5 $SUBJ/masks.hdf5 $SUBJ/camera.pkl $OUT/consensus.pkl --obj_out $OUT/consensus.obj $MODEL ${@:3}
+python step2_consensus.py $SUBJ/reconstructed_poses.hdf5 $SUBJ/masks.hdf5 $SUBJ/camera.pkl $OUT/consensus.pkl --obj_out $OUT/consensus.obj $MODEL ${@:3} 
diff --git a/run_step3.sh b/run_step3.sh
index dcb8d1e..befabc5 100755
--- a/run_step3.sh
+++ b/run_step3.sh
@@ -9,7 +9,7 @@ SUBJ="$1"
 OUT="$2"
 
 if [[ $SUBJ = *"female"* ]]; then
-  MODEL='--model vendor/smpl/models/basicModel_f_lbs_10_207_0_v1.0.0.pkl'
+  MODEL='--model vendor/smpl/models/basicModel_m_lbs_10_207_0_v1.0.0.pkl'
 fi
 
-python step3_texture.py $SUBJ/consensus.pkl  $SUBJ/camera.pkl $SUBJ/$(basename $SUBJ).mp4 $SUBJ/reconstructed_poses.hdf5 $SUBJ/masks.hdf5 $OUT/tex-$(basename $SUBJ).jpg $MODEL ${@:3}
+python step3_texture.py $SUBJ/consensus.pkl  $SUBJ/camera.pkl $SUBJ/noratap_360_1080.avi $SUBJ/reconstructed_poses.hdf5 $SUBJ/masks.hdf5 $OUT/tex-$(basename $SUBJ).jpg $MODEL ${@:3}
diff --git a/step1_pose.py b/step1_pose.py
index c264c1d..c6b93b2 100644
--- a/step1_pose.py
+++ b/step1_pose.py
@@ -1,31 +1,29 @@
 #!/usr/bin/env python2
 # -*- coding: utf-8 -*-
 
-import cv2
-import h5py
-import argparse
-import numpy as np
-import chumpy as ch
-import cPickle as pkl
-
-from opendr.camera import ProjectPoints
-from opendr.lighting import LambertianPointLight
-from opendr.renderer import ColoredRenderer
-from opendr.filters import gaussian_pyramid
-
-from util import im
-from util.logger import log
-from lib.frame import FrameData
-from models.smpl import Smpl, copy_smpl, joints_coco
-from models.bodyparts import faces_no_hands
-
-from vendor.smplify.sphere_collisions import SphereCollisions
 from vendor.smplify.robustifiers import GMOf
+from vendor.smplify.sphere_collisions import SphereCollisions
+from models.bodyparts import faces_no_hands
+from models.smpl import Smpl, copy_smpl, joints_coco
+from lib.frame import FrameData
+from util.logger import log
+from util import im
+from opendr.filters import gaussian_pyramid
+from opendr.renderer import ColoredRenderer
+from opendr.lighting import LambertianPointLight
+from opendr.camera import ProjectPoints
+import pickle as pkl
+import chumpy as ch
+import numpy as np
+import argparse
+import h5py
+import cv2
 
 
 def get_cb(viz_rn, f):
     if viz_rn is not None:
-        viz_rn.set(v=f.smpl, background_image=np.dstack((f.mask, f.mask, f.mask)))
+        viz_rn.set(v=f.smpl, background_image=np.dstack(
+            (f.mask, f.mask, f.mask)))
         viz_rn.vc.set(v=f.smpl)
 
         def cb(_):
@@ -44,7 +42,8 @@ def cb(_):
 
 
 def collision_obj(smpl, regs):
-    sp = SphereCollisions(pose=smpl.pose, betas=smpl.betas, model=smpl, regs=regs)
+    sp = SphereCollisions(
+        pose=smpl.pose, betas=smpl.betas, model=smpl, regs=regs)
     sp.no_hands = True
 
     return sp
@@ -63,7 +62,7 @@ def init(frames, body_height, b2m, viz_rn):
 
     E_height = None
     if body_height is not None:
-         E_height = height_predictor(b2m, betas) - body_height * 1000.
+        E_height = height_predictor(b2m, betas) - body_height * 1000.
 
     # first get a rough pose for all frames individually
     for i, f in enumerate(frames):
@@ -112,7 +111,8 @@ def init(frames, body_height, b2m, viz_rn):
 
         for i, f in enumerate(frames):
             if np.sum(f.keypoints[[0, 2, 5, 8, 11], 2]) > 3.:
-                x0.extend([f.smpl.pose[range(21) + range(27, 30) + range(36, 60)], f.smpl.trans])
+                x0.extend(
+                    [f.smpl.pose[list(range(21, 0)) + list(range(27, 30)) + list(range(36, 60))], f.smpl.trans])
                 E['pose_{}'.format(i)] = f.pose_obj
                 E['prior_{}'.format(i)] = f.pose_prior_obj * w_prior
 
@@ -163,9 +163,11 @@ def reinit_frame(frame, null_pose, nohands, viz_rn):
         x0 = [frame.smpl.trans]
 
         if nohands:
-            x0.append(frame.smpl.pose[range(21) + range(27, 30) + range(36, 60)])
+            x0.append(frame.smpl.pose[list(range(21)) +
+                      list(range(27, 30)) + list(range(36, 60))])
         else:
-            x0.append(frame.smpl.pose[range(21) + range(27, 30) + range(36, 72)])
+            x0.append(frame.smpl.pose[list(range(21)) +
+                      list(range(27, 30)) + list(range(36, 72))])
 
         ch.minimize(
             E,
@@ -190,7 +192,8 @@ def fit_pose(frame, last_smpl, frustum, nohands, viz_rn):
     dist_i = cv2.distanceTransform(np.uint8(frame.mask * 255), dst_type, 5) - 1
     dist_i[dist_i < 0] = 0
     dist_i[dist_i > 50] = 50
-    dist_o = cv2.distanceTransform(255 - np.uint8(frame.mask * 255), dst_type, 5)
+    dist_o = cv2.distanceTransform(
+        255 - np.uint8(frame.mask * 255), dst_type, 5)
     dist_o[dist_o > 50] = 50
 
     rn_m = ColoredRenderer(camera=frame.camera, v=frame.smpl, f=faces, vc=np.ones_like(frame.smpl), frustum=frustum,
@@ -208,9 +211,11 @@ def fit_pose(frame, last_smpl, frustum, nohands, viz_rn):
         E['last_trans'] = GMOf(frame.smpl.trans - last_smpl.trans, 0.05) * 50.
 
     if nohands:
-        x0 = [frame.smpl.pose[range(21) + range(27, 30) + range(36, 60)], frame.smpl.trans]
+        x0 = [frame.smpl.pose[list(range(
+            21)) + list(range(27, 30)) + list(range(36, 60))], frame.smpl.trans]
     else:
-        x0 = [frame.smpl.pose[range(21) + range(27, 30) + range(36, 72)], frame.smpl.trans]
+        x0 = [frame.smpl.pose[list(range(
+            21)) + list(range(27, 30)) + list(range(36, 72))], frame.smpl.trans]
 
     ch.minimize(
         E,
@@ -228,19 +233,27 @@ def main(keypoint_file, masks_file, camera_file, out, model_file, prior_file, re
 
     # load data
     with open(model_file, 'rb') as fp:
-        model_data = pkl.load(fp)
+        u = pkl._Unpickler(fp)
+        u.encoding = 'latin1'
+        model_data = u.load()
 
     with open(camera_file, 'rb') as fp:
-        camera_data = pkl.load(fp)
+        u = pkl._Unpickler(fp)
+        u.encoding = 'latin1'
+        camera_data = u.load()
 
     with open(prior_file, 'rb') as fp:
-        prior_data = pkl.load(fp)
+        u = pkl._Unpickler(fp)
+        u.encoding = 'latin1'
+        prior_data = u.load()
 
     if 'basicModel_f' in model_file:
-        regs = np.load('vendor/smplify/models/regressors_locked_normalized_female.npz')
+        regs = np.load(
+            'vendor/smplify/models/regressors_locked_normalized_female.npz')
         b2m = np.load('assets/b2m_f.npy')
     else:
-        regs = np.load('vendor/smplify/models/regressors_locked_normalized_male.npz')
+        regs = np.load(
+            'vendor/smplify/models/regressors_locked_normalized_male.npz')
         b2m = np.load('assets/b2m_m.npy')
 
     keypoints = h5py.File(keypoint_file, 'r')['keypoints']
@@ -259,10 +272,12 @@ def main(keypoint_file, masks_file, camera_file, out, model_file, prior_file, re
                'width': int(camera_data['width'] * resize), 'height': int(camera_data['height'] * resize)}
 
     if display:
-        debug_cam = ProjectPoints(v=base_smpl, t=camera.t, rt=camera.rt, c=camera.c, f=camera.f, k=camera.k)
+        debug_cam = ProjectPoints(
+            v=base_smpl, t=camera.t, rt=camera.rt, c=camera.c, f=camera.f, k=camera.k)
         debug_light = LambertianPointLight(f=base_smpl.f, v=base_smpl, num_verts=len(base_smpl), light_pos=np.zeros(3),
                                            vc=np.ones(3), light_color=np.ones(3))
-        debug_rn = ColoredRenderer(camera=debug_cam, v=base_smpl, f=base_smpl.f, vc=debug_light, frustum=frustum)
+        debug_rn = ColoredRenderer(
+            camera=debug_cam, v=base_smpl, f=base_smpl.f, vc=debug_light, frustum=frustum)
     else:
         debug_rn = None
 
@@ -271,17 +286,21 @@ def create_frame(i, smpl, copy=True):
         f = FrameData()
 
         f.smpl = copy_smpl(smpl, model_data) if copy else smpl
-        f.camera = ProjectPoints(v=f.smpl, t=camera.t, rt=camera.rt, c=camera.c, f=camera.f, k=camera.k)
+        f.camera = ProjectPoints(
+            v=f.smpl, t=camera.t, rt=camera.rt, c=camera.c, f=camera.f, k=camera.k)
 
-        f.keypoints = np.array(keypoints[i]).reshape(-1, 3) * np.array([resize, resize, 1])
+        f.keypoints = np.array(
+            keypoints[i]).reshape(-1, 3) * np.array([resize, resize, 1])
         f.J = joints_coco(f.smpl)
-        f.J_proj = ProjectPoints(v=f.J, t=camera.t, rt=camera.rt, c=camera.c, f=camera.f, k=camera.k)
+        f.J_proj = ProjectPoints(
+            v=f.J, t=camera.t, rt=camera.rt, c=camera.c, f=camera.f, k=camera.k)
         f.mask = cv2.resize(np.array(masks[i], dtype=np.float32), (0, 0),
                             fx=resize, fy=resize, interpolation=cv2.INTER_NEAREST)
 
         f.collision_obj = collision_obj(f.smpl, regs)
         f.pose_prior_obj = pose_prior_obj(f.smpl, prior_data)
-        f.pose_obj = (f.J_proj - f.keypoints[:, :2]) * f.keypoints[:, 2].reshape(-1, 1)
+        f.pose_obj = (
+            f.J_proj - f.keypoints[:, :2]) * f.keypoints[:, 2].reshape(-1, 1)
 
         return f
 
@@ -291,7 +310,8 @@ def create_frame(i, smpl, copy=True):
     log.info('Initial fit')
 
     num_init = 5
-    indices_init = np.ceil(np.arange(num_init) * num_frames * 1. / num_init).astype(np.int)
+    indices_init = np.ceil(np.arange(num_init) *
+                           num_frames * 1. / num_init).astype(np.int32)
 
     init_frames = [base_frame]
     for i in indices_init[1:]:
@@ -302,11 +322,14 @@ def create_frame(i, smpl, copy=True):
     # get pose frame by frame
     with h5py.File(out, 'w') as fp:
         last_smpl = None
-        poses_dset = fp.create_dataset("pose", (num_frames, 72), 'f', chunks=True, compression="lzf")
-        trans_dset = fp.create_dataset("trans", (num_frames, 3), 'f', chunks=True, compression="lzf")
-        betas_dset = fp.create_dataset("betas", (10,), 'f', chunks=True, compression="lzf")
-
-        for i in xrange(num_frames):
+        poses_dset = fp.create_dataset(
+            "pose", (num_frames, 72), 'f', chunks=True, compression="lzf")
+        trans_dset = fp.create_dataset(
+            "trans", (num_frames, 3), 'f', chunks=True, compression="lzf")
+        betas_dset = fp.create_dataset(
+            "betas", (10,), 'f', chunks=True, compression="lzf")
+
+        for i in range(num_frames-1):
             if i == 0:
                 current_frame = base_frame
             else:
diff --git a/step2_consensus.py b/step2_consensus.py
index c13ef12..06f2117 100644
--- a/step2_consensus.py
+++ b/step2_consensus.py
@@ -5,7 +5,7 @@
 import argparse
 import numpy as np
 import chumpy as ch
-import cPickle as pkl
+import pickle as pkl
 
 from opendr.camera import ProjectPoints
 from opendr.renderer import BoundaryRenderer, ColoredRenderer
@@ -43,7 +43,8 @@ def fit_consensus(frames, base_smpl, camera, frustum, model_data, nohands, icp_c
     else:
         faces = base_smpl.f
 
-    vis_rn_b = BoundaryRenderer(camera=camera, frustum=frustum, f=faces, num_channels=1)
+    vis_rn_b = BoundaryRenderer(
+        camera=camera, frustum=frustum, f=faces, num_channels=1)
     vis_rn_m = ColoredRenderer(camera=camera, frustum=frustum, f=faces, vc=np.zeros_like(base_smpl), bgcolor=1,
                                num_channels=1)
 
@@ -55,8 +56,10 @@ def fit_consensus(frames, base_smpl, camera, frustum, model_data, nohands, icp_c
     g_symmetry = regularize_symmetry()
 
     for step, (w_laplace, w_model, w_symmetry, sigma) in enumerate(zip(
-            np.linspace(6.5, 4.0, icp_count) if naked else np.linspace(4.0, 2.0, icp_count),
-            np.linspace(0.9, 0.6, icp_count) if naked else np.linspace(0.6, 0.3, icp_count),
+            np.linspace(6.5, 4.0, icp_count) if naked else np.linspace(
+                4.0, 2.0, icp_count),
+            np.linspace(0.9, 0.6, icp_count) if naked else np.linspace(
+                0.6, 0.3, icp_count),
             np.linspace(3.6, 1.8, icp_count),
             np.linspace(0.06, 0.003, icp_count),
     )):
@@ -78,7 +81,11 @@ def fit_consensus(frames, base_smpl, camera, frustum, model_data, nohands, icp_c
 
         log.info('## Matching rays with contours')
         for current, f in enumerate(tqdm(frames)):
-            E['silh_{}'.format(current)] = ray_objective(f, sigma, base_smpl, camera, vis_rn_b, vis_rn_m)
+            try:
+                E['silh_{}'.format(current)] = ray_objective(
+                    f, sigma, base_smpl, camera, vis_rn_b, vis_rn_m)
+            except:
+                continue
 
         log.info('## Run optimization')
         ch.minimize(
@@ -86,7 +93,8 @@ def fit_consensus(frames, base_smpl, camera, frustum, model_data, nohands, icp_c
             [base_smpl.v_personal, model_template.betas],
             method='dogleg',
             options={'maxiter': 15, 'e_3': 0.001},
-            callback=get_cb(frames[0], base_smpl, camera, frustum) if display else None
+            callback=get_cb(frames[0], base_smpl, camera,
+                            frustum) if display else None
         )
 
 
@@ -95,10 +103,14 @@ def main(pose_file, masks_file, camera_file, out, obj_out, num, icp_count, model
 
     # load data
     with open(model_file, 'rb') as fp:
-        model_data = pkl.load(fp)
+        u = pkl._Unpickler(fp)
+        u.encoding = 'latin1'
+        model_data = u.load()
 
     with open(camera_file, 'rb') as fp:
-        camera_data = pkl.load(fp)
+        u = pkl._Unpickler(fp)
+        u.encoding = 'latin1'
+        camera_data = u.load()
 
     pose_data = h5py.File(pose_file, 'r')
     poses = pose_data['pose'][first_frame:last_frame]
@@ -106,7 +118,8 @@ def main(pose_file, masks_file, camera_file, out, obj_out, num, icp_count, model
     masks = h5py.File(masks_file, 'r')['masks'][first_frame:last_frame]
     num_frames = masks.shape[0]
 
-    indices_consensus = np.ceil(np.arange(num) * num_frames * 1. / num).astype(np.int)
+    indices_consensus = np.ceil(
+        np.arange(num) * num_frames * 1. / num).astype(np.int64)
 
     # init
     base_smpl = Smpl(model_data)
@@ -116,7 +129,8 @@ def main(pose_file, masks_file, camera_file, out, obj_out, num, icp_count, model
                            f=camera_data['camera_f'], k=camera_data['camera_k'], v=base_smpl)
     camera_t = camera_data['camera_t']
     camera_rt = camera_data['camera_rt']
-    frustum = {'near': 0.1, 'far': 1000., 'width': int(camera_data['width']), 'height': int(camera_data['height'])}
+    frustum = {'near': 0.1, 'far': 1000., 'width': int(
+        camera_data['width']), 'height': int(camera_data['height'])}
     frames = []
 
     for i in indices_consensus:
@@ -126,11 +140,13 @@ def main(pose_file, masks_file, camera_file, out, obj_out, num, icp_count, model
         pose_i = np.array(poses[i], dtype=np.float32)
         trans_i = np.array(trans[i], dtype=np.float32)
 
-        frames.append(setup_frame_rays(base_smpl, camera, camera_t, camera_rt, pose_i, trans_i, mask))
+        frames.append(setup_frame_rays(base_smpl, camera,
+                      camera_t, camera_rt, pose_i, trans_i, mask))
 
     log.info('Set up complete.')
     log.info('Begin consensus fit...')
-    fit_consensus(frames, base_smpl, camera, frustum, model_data, nohands, icp_count, naked, display)
+    fit_consensus(frames, base_smpl, camera, frustum,
+                  model_data, nohands, icp_count, naked, display)
 
     with open(out, 'wb') as fp:
         pkl.dump({
diff --git a/step3_texture.py b/step3_texture.py
index 4042d79..2a53e41 100644
--- a/step3_texture.py
+++ b/step3_texture.py
@@ -5,7 +5,7 @@
 import h5py
 import argparse
 import numpy as np
-import cPickle as pkl
+import pickle as pkl
 
 from opendr.renderer import ColoredRenderer
 from opendr.camera import ProjectPoints
@@ -21,20 +21,27 @@ def main(consensus_file, camera_file, video_file, pose_file, masks_file, out, mo
          first_frame, last_frame, display):
     # load data
     with open(model_file, 'rb') as fp:
-        model_data = pkl.load(fp)
+        u = pkl._Unpickler(fp)
+        u.encoding = 'latin1'
+        model_data = u.load()
 
     with open(camera_file, 'rb') as fp:
-        camera_data = pkl.load(fp)
+        u = pkl._Unpickler(fp)
+        u.encoding = 'latin1'
+        camera_data = u.load()
 
     with open(consensus_file, 'rb') as fp:
-        consensus_data = pkl.load(fp)
+        u = pkl._Unpickler(fp)
+        u.encoding = 'latin1'
+        consensus_data = u.load()
 
     pose_data = h5py.File(pose_file, 'r')
     poses = pose_data['pose'][first_frame:last_frame]
     trans = pose_data['trans'][first_frame:last_frame]
     masks = h5py.File(masks_file, 'r')['masks'][first_frame:last_frame]
     num_frames = masks.shape[0]
-    indices_texture = np.ceil(np.arange(num) * num_frames * 1. / num).astype(np.int)
+    indices_texture = np.ceil(
+        np.arange(num) * num_frames * 1. / num).astype(np.int64)
 
     vt = np.load('assets/basicModel_vt.npy')
     ft = np.load('assets/basicModel_ft.npy')
@@ -49,8 +56,10 @@ def main(consensus_file, camera_file, video_file, pose_file, masks_file, out, mo
     iso_vis = IsoColoredRenderer(vt, ft, base_smpl.f, resolution)
     camera = ProjectPoints(t=camera_data['camera_t'], rt=camera_data['camera_rt'], c=camera_data['camera_c'],
                            f=camera_data['camera_f'], k=camera_data['camera_k'], v=base_smpl)
-    frustum = {'near': 0.1, 'far': 1000., 'width': int(camera_data['width']), 'height': int(camera_data['height'])}
-    rn_vis = ColoredRenderer(f=base_smpl.f, frustum=frustum, camera=camera, num_channels=1)
+    frustum = {'near': 0.1, 'far': 1000., 'width': int(
+        camera_data['width']), 'height': int(camera_data['height'])}
+    rn_vis = ColoredRenderer(
+        f=base_smpl.f, frustum=frustum, camera=camera, num_channels=1)
 
     cap = cv2.VideoCapture(video_file)
     for _ in range(first_frame):
@@ -84,18 +93,25 @@ def main(consensus_file, camera_file, video_file, pose_file, masks_file, out, mo
 
             proj = camera.r
             in_viewport = np.logical_and(
-                np.logical_and(np.round(camera.r[:, 0]) >= 0, np.round(camera.r[:, 0]) < frustum['width']),
-                np.logical_and(np.round(camera.r[:, 1]) >= 0, np.round(camera.r[:, 1]) < frustum['height']),
+                np.logical_and(np.round(camera.r[:, 0]) >= 0, np.round(
+                    camera.r[:, 0]) < frustum['width']),
+                np.logical_and(np.round(camera.r[:, 1]) >= 0, np.round(
+                    camera.r[:, 1]) < frustum['height']),
             )
             in_mask = np.zeros(camera.shape[0], dtype=np.bool)
-            idx = np.round(proj[in_viewport][:, [1, 0]].T).astype(np.int).tolist()
+            idx = np.round(proj[in_viewport][:, [1, 0]].T).astype(
+                np.int64).tolist()
             in_mask[in_viewport] = mask[idx]
 
             faces_in_mask = np.where(np.min(in_mask[base_smpl.f], axis=1))[0]
             visible_faces = np.intersect1d(faces_in_mask, visibility[visible])
 
             # get the current unwrap
-            part_tex = iso.render(frame / 255., camera, visible_faces)
+            try:
+                part_tex = iso.render(frame / 255., camera, visible_faces)
+            except:
+                i += 1
+                continue
 
             # angle under which the texels have been seen
             points = np.hstack((proj, np.ones((proj.shape[0], 1))))
@@ -114,9 +130,12 @@ def main(consensus_file, camera_file, video_file, pose_file, masks_file, out, mo
             # update best seen texels
             where = np.argmax(np.atleast_3d(iso_normals) - normal_agg, axis=2)
 
-            idx = np.dstack((static_indices[0], static_indices[1], where))[part_mask == 1]
-            tex_agg[list(idx[:, 0]), list(idx[:, 1]), list(idx[:, 2])] = part_tex[part_mask == 1]
-            normal_agg[list(idx[:, 0]), list(idx[:, 1]), list(idx[:, 2])] = iso_normals[part_mask == 1]
+            idx = np.dstack((static_indices[0], static_indices[1], where))[
+                part_mask == 1]
+            tex_agg[list(idx[:, 0]), list(idx[:, 1]), list(
+                idx[:, 2])] = part_tex[part_mask == 1]
+            normal_agg[list(idx[:, 0]), list(idx[:, 1]), list(
+                idx[:, 2])] = iso_normals[part_mask == 1]
 
             if display:
                 im.show(part_tex, id='part_tex', waittime=1)
@@ -140,7 +159,8 @@ def main(consensus_file, camera_file, video_file, pose_file, masks_file, out, mo
     kernel = np.ones((kernel_size, kernel_size), np.uint8)
     inpaint_area = cv2.dilate(tex_mask, kernel) - mask_final
 
-    tex_final = cv2.inpaint(np.uint8(tex_median * 255), np.uint8(inpaint_area * 255), 3, cv2.INPAINT_TELEA)
+    tex_final = cv2.inpaint(np.uint8(tex_median * 255),
+                            np.uint8(inpaint_area * 255), 3, cv2.INPAINT_TELEA)
 
     cv2.imwrite(out, tex_final)
     log.info('Done.')