tog'21 baseline

README.md

@@ -13,18 +13,94 @@ Or ref to https://pytorch.org/get-started/locally/ for install guide
 
 * tensorboard
 
+* (Optional) dash
+
+```
+$ conda install dash pandas
+```
+
 ## 2. Install Pip packages:
 * pyGlm
 * tensorboardX
 * (optional) opencv-python
+* (optional) thop
+* (optional) ConcurrentLogHandler
 
 # Useful commands
 ## 1. Video generate:
 ```
-$ ffmpeg -y -r 30 -i view_%04d.png -c:v libx264 -vf fps=30 -pix_fmt yuv420p ../train.mp4
+$ ffmpeg -y -r 50 -i %04d.png -c:v libx264 -vframes 600 ../classroom_hmd_mono_hint.mp4
 ```
 
 ## 2. Convert onnx to tensorRT
 ```
 $ trtexec --onnx=net@256x256x2.onnx --fp16 --saveEngine=net@256x256x2.trt --workspace=4096
 ```
+
+# Install FFMpeg with Extra Codecs:
+
+```
+sudo apt-get update -qq && sudo apt-get -y install \
+  autoconf \
+  automake \
+  build-essential \
+  cmake \
+  git-core \
+  libass-dev \
+  libfreetype6-dev \
+  libgnutls28-dev \
+  libsdl2-dev \
+  libtool \
+  libva-dev \
+  libvdpau-dev \
+  libvorbis-dev \
+  libxcb1-dev \
+  libxcb-shm0-dev \
+  libxcb-xfixes0-dev \
+  meson \
+  ninja-build \
+  pkg-config \
+  texinfo \
+  wget \
+  yasm \
+  zlib1g-dev \
+  libunistring-dev \
+  libvpx-dev \
+  libfdk-aac-dev \
+  libmp3lame-dev \
+  libopus-dev \
+  nasm \
+  libx264-dev \
+  libx265-dev \
+  libnuma-dev
+
+mkdir -p ~/ffmpeg_sources ~/bin
+
+cd ~/ffmpeg_sources && \
+wget -O ffmpeg-snapshot.tar.bz2 https://ffmpeg.org/releases/ffmpeg-snapshot.tar.bz2 && \
+tar xjvf ffmpeg-snapshot.tar.bz2
+
+cd ffmpeg && \
+PATH="$HOME/bin:$PATH" PKG_CONFIG_PATH="$HOME/ffmpeg_build/lib/pkgconfig" ./configure \
+  --prefix="$HOME/ffmpeg_build" \
+  --pkg-config-flags="--static" \
+  --extra-cflags="-I$HOME/ffmpeg_build/include" \
+  --extra-ldflags="-L$HOME/ffmpeg_build/lib" \
+  --extra-libs="-lpthread -lm" \
+  --bindir="$HOME/bin" \
+  --enable-gpl \
+  --enable-gnutls \
+  --enable-libass \
+  --enable-libfdk-aac \
+  --enable-libfreetype \
+  --enable-libmp3lame \
+  --enable-libopus \
+  --enable-libvorbis \
+  --enable-libvpx \
+  --enable-libx264 \
+  --enable-libx265 \
+  --enable-nonfree && \
+PATH="$HOME/bin:$PATH" make && \
+make install && \
+hash -r
+```
\ No newline at end of file

assets/dash_test.css

+.slider {
+    margin-top: 30px
+}
\ No newline at end of file

batch_collect_video.sh

+#/usr/bin/bash
+curdir=$(pwd)
+datadir="$curdir/data/__new/classroom_fovea_r360x80_t0.6"
+videodir="$datadir/eval_video"
+epochs=50
+
+if [ ! -d "$videodir" ]; then
+    echo "make directory for Video"
+    mkdir $videodir
+fi
+
+# nets: 1, 2, 4, 8
+# layers: 2, 4, 8
+# channels: 64 128 256 512 1024
+for n_nets in 1 2 4 8; do
+    for n_layers in 2 4 8; do
+        for nf in 64 128 256 512 1024; do
+            for n_samples in 8 16 32 64 128; do
+                configid="eval@snerffast${n_nets}-rgb_e6_fc${nf}x${n_layers}_d1.00-7.00_s${n_samples}_~p"
+                exportname="eval_${n_nets}x${nf}x${n_layers}_${n_samples}"
+                src_path="../$configid/output_$epochs/helix_color.mp4"
+                dst_path="$videodir/$exportname.mp4"
+                if [ -f "$videodir/$src_path" ]; then
+                    if [ ! -f "$dst_path" ]; then
+                        ln -s $src_path $dst_path
+                    fi
+                fi
+            done
+        done
+    done
+done
\ No newline at end of file

batch_export_net.sh

@@ -21,8 +21,13 @@ fi
 for n_nets in 1 2 4 8; do
     for n_layers in 2 4 8; do
         for nf in 64 128 256 512 1024; do
-            configid="eval@snerffast${n_nets}-rgb_e6_fc${nf}x${n_layers}_d1.00-7.00_s64_~p"
+            for n_samples in 8 16 32 64 128; do
+                configid="eval@snerffast${n_nets}-rgb_e6_fc${nf}x${n_layers}_d1.00-7.00_s${n_samples}_~p"
+                if (( $n_samples == 64 )); then
                     exportname="eval_${n_nets}x${nf}x${n_layers}"
+                else
+                    exportname="eval_${n_nets}x${nf}x${n_layers}_${n_samples}"
+                fi
                 pth_path="$datadir/$configid/model-epoch_$epochs.pth"
                 onnx_path="$onnxdir/$exportname.onnx"
                 trt_path="$trtdir/$exportname.trt"
@@ -39,4 +44,5 @@ for n_nets in 1 2 4 8; do
                 fi
             done
         done
+    done
 done
\ No newline at end of file

batch_infer.sh

@@ -4,6 +4,7 @@ testcase=$1
 datadir='data/__new/classroom_fovea_r360x80_t0.6'
 trainset='data/__new/classroom_fovea_r360x80_t0.6/r120x80.json'
 testset='data/__new/classroom_fovea_r360x80_t0.6/r120x80_test.json'
+videoset='data/__new/classroom_fovea_r360x80_t0.6/helix.json'
 epochs=50
 
 # nets: 1, 2, 4, 8
@@ -30,8 +31,14 @@ for nf in 64 128 256 512 1024; do
             python run_spherical_view_syn.py $trainset -i $configid -e $epochs
         fi
     fi
-    if ! ls $datadir/$configid/output_$epochs/perf_r120x80* >/dev/null 2>&1; then
+    if ! ls $datadir/$configid/output_$epochs/perf* >/dev/null 2>&1; then
         python run_spherical_view_syn.py $trainset -t -m $configid/model-epoch_$epochs.pth -o perf
         python run_spherical_view_syn.py $testset -t -m $configid/model-epoch_$epochs.pth -o perf
     fi
+    if [ ! -d "$datadir/$configid/output_$epochs/r120x80_test_color" ]; then
+        python run_spherical_view_syn.py $testset -t -m $configid/model-epoch_$epochs.pth -o color
+    fi
+    if [ ! -f "$datadir/$configid/output_$epochs/helix_color.mp4" ]; then
+        python run_spherical_view_syn.py $videoset -t -m $configid/model-epoch_$epochs.pth -o color --output-type video
+    fi
 done
\ No newline at end of file

components/fnr.py

@@ -50,6 +50,7 @@ class FoveatedNeuralRenderer(object):
     def render(self, view: Trans, gaze, right_gaze=None, *,
                stereo_disparity=0,
                using_mask=True,
+               mono_periph_mode=0,
                ret_raw=False) -> Union[Mapping[str, torch.Tensor], Tuple[Mapping[str, torch.Tensor]]]:
         if stereo_disparity > TINY_FLOAT:
             left_view = Trans(
@@ -61,24 +62,50 @@ class FoveatedNeuralRenderer(object):
             left_gaze = gaze
             right_gaze = gaze if right_gaze is None else right_gaze
 
-            left_layers_mask = self.foveation.get_layers_mask(left_gaze) \
-                if using_mask else [None] * 3
-            right_layers_mask = self.foveation.get_layers_mask(right_gaze) \
-                if using_mask else [None] * 3
-            res_raw_left = [
+            layers_mask = self.foveation.get_layers_mask() if using_mask else [None] * 3
+            left_shifts = None
+            right_shifts = None
+
+            if using_mask and mono_periph_mode != 0:
+                fovea_left = self._render(self.layers_net[0], self.layers_cam[0], left_view, left_gaze,
+                                          layer_mask=layers_mask[0])['color']
+                fovea_right = self._render(self.layers_net[0], self.layers_cam[0], right_view, right_gaze,
+                                           layer_mask=layers_mask[0])['color']
+                if mono_periph_mode == 3:
+                    mid = self._render(self.layers_net[1], self.layers_cam[1], view,
+                                       ((left_gaze[0] + right_gaze[0]) // 2, left_gaze[1]),
+                                       layer_mask=layers_mask[1])['color']
+                    periph = self._render(self.layers_net[2], self.layers_cam[2], view)['color']
+                    raw_left = [fovea_left, mid, periph]
+                    raw_right = [fovea_right, mid, periph]
+                    shift = int(left_gaze[0] - right_gaze[0]) // 2
+                    left_shifts = [0, 0, shift]
+                    right_shifts = [0, 0, -shift]
+                else:
+                    mid_left = self._render_mid(self.layers_net[1], self.layers_cam[1], left_view, left_gaze,
+                                                layer_mask=layers_mask[1], mono_view=view,
+                                                blend_view=mono_periph_mode == 1)['color']
+                    mid_right = self._render_mid(self.layers_net[1], self.layers_cam[1], right_view, right_gaze,
+                                                 layer_mask=layers_mask[1], mono_view=view,
+                                                 blend_view=mono_periph_mode == 1)['color']
+                    periph = self._render(self.layers_net[2], self.layers_cam[2], view)['color']
+                    raw_left = [fovea_left, mid_left, periph]
+                    raw_right = [fovea_right, mid_right, periph]
+            else:
+                raw_left = [
                     self._render(self.layers_net[i], self.layers_cam[i], left_view,
                                  left_gaze if i < 2 else None,
-                             layer_mask=left_layers_mask[i])['color']
+                                 layer_mask=layers_mask[i])['color']
                     for i in range(3)
                 ]
-            res_raw_right = [
+                raw_right = [
                     self._render(self.layers_net[i], self.layers_cam[i], right_view,
                                  right_gaze if i < 2 else None,
-                             layer_mask=right_layers_mask[i])['color']
+                                 layer_mask=layers_mask[i])['color']
                     for i in range(3)
                 ]
-            return self._gen_output(res_raw_left, left_gaze, ret_raw), \
-                self._gen_output(res_raw_right, right_gaze, ret_raw)
+            return self._gen_output(raw_left, left_gaze, left_shifts, ret_raw=ret_raw), \
+                self._gen_output(raw_right, right_gaze, right_shifts, ret_raw=ret_raw)
         else:
             layers_mask = self.foveation.get_layers_mask(gaze) if using_mask else None
             res_raw = [
@@ -86,34 +113,7 @@ class FoveatedNeuralRenderer(object):
                              layer_mask=layers_mask[i] if layers_mask is not None else None)['color']
                 for i in range(3)
             ]
-            return self._gen_output(res_raw, gaze, ret_raw)
-
-        '''
-        if mono_trans != None and shift == 0:  # do warp
-            fovea_depth[torch.isnan(fovea_depth)] = 50
-            mid_depth[torch.isnan(mid_depth)] = 50
-            periph_depth[torch.isnan(periph_depth)] = 50
-
-            if warp_by_depth:
-                z_list = misc.depth_sample((1, 50), 4, True)
-                mid_inferred = self._warp(trans, mono_trans, mid_cam,
-                                          z_list, mid_inferred, mid_depth)
-                periph_inferred = self._warp(trans, mono_trans, periph_cam,
-                                             z_list, periph_inferred, periph_depth)
-            else:
-                p = torch.tensor([[0, 0, torch.mean(fovea_depth)]],
-                                 device=self.device)
-                p_ = trans.trans_point(mono_trans.trans_point(p), inverse=True)
-                shift = self.full_cam.proj(
-                    p_, center_as_origin=True)[..., 0].item()
-                shift = round(shift)
-
-        blended = self.foveation.synthesis([
-            fovea_refined,
-            mid_refined,
-            periph_refined
-        ], (gaze[0], gaze[1]), [0, shift, shift] if shift != 0 else None)
-        '''
+            return self._gen_output(res_raw, gaze, ret_raw=ret_raw)
 
     def _render(self, net, cam: CameraParam, view: Trans, gaze=None, *,
                 ret_depth=False,
@@ -142,9 +142,54 @@ class FoveatedNeuralRenderer(object):
                 'depth': net_output['depth'].view(1, cam.res[0], cam.res[1]) if ret_depth else None
             }
 
-    def _gen_output(self, layers_img: List[torch.Tensor], gaze: Tuple[float, float], ret_raw=False) -> Mapping[str, torch.Tensor]:
+    def _render_mid(self, net, cam: CameraParam, view: Trans, gaze=None, *,
+                    layer_mask: torch.Tensor,
+                    mono_view: Trans,
+                    blend_view: bool,
+                    ret_depth=False) -> Mapping[str, torch.Tensor]:
+        """
+        [summary]
+
+        :param net: [description]
+        :param cam: [description]
+        :param view: [description]
+        :param layer_mask: [description]
+        :param mono_view: [description]
+        :param gaze: [description], defaults to None
+        :param ret_depth: [description], defaults to False
+        :return: [description]
+        """
+        if gaze is not None:
+            cam = self._adjust_cam(cam, gaze)
+
+        k = layer_mask[None, ..., None].clamp(1 if blend_view else 2, 2) - 1  # (1, H, W, 1)
+        rays_o = (1 - k) * view.t + k * mono_view.t  # (1, H, W, 3)
+        rays_d = view.trans_vector(cam.get_local_rays())  # (1, H, W, 3)
+
+        if layer_mask is not None:
+            infer_mask = layer_mask >= 0
+            rays_o = rays_o[:, infer_mask]
+            rays_d = rays_d[:, infer_mask]
+            net_output = net(rays_o.view(-1, 3), rays_d.view(-1, 3), ret_depth=ret_depth)
+            ret = {
+                'color': torch.zeros(1, cam.res[0], cam.res[1], 3, device=self.device)
+            }
+            ret['color'][:, infer_mask] = net_output['color']
+            ret['color'] = ret['color'].permute(0, 3, 1, 2)
+            if ret_depth:
+                ret['depth'] = torch.zeros(1, cam.res[0], cam.res[1])
+                ret['depth'][:, infer_mask] = net_output['depth']
+            return ret
+        else:
+            net_output = net(rays_o.view(-1, 3), rays_d.view(-1, 3), ret_depth=ret_depth)
+            return {
+                'color': net_output['color'].view(1, cam.res[0], cam.res[1], -1).permute(0, 3, 1, 2),
+                'depth': net_output['depth'].view(1, cam.res[0], cam.res[1]) if ret_depth else None
+            }
+
+    def _gen_output(self, layers_img: List[torch.Tensor], gaze: Tuple[float, float], shifts=None, ret_raw=False) -> Mapping[str, torch.Tensor]:
         refined = self._post_process(layers_img)
-        blended = self.foveation.synthesis(refined, gaze)
+        blended = self.foveation.synthesis(refined, gaze, shifts)
         ret = {
             'layers_img': refined,
             'blended': blended

components/foveation.py

@@ -10,7 +10,7 @@ from utils import misc
 class Foveation(object):
 
     def __init__(self, layers_fov: List[float], layers_res: List[Tuple[float, float]],
-                 out_res: Tuple[int, int], *, blend=0.6, device=None):
+                 out_res: Tuple[int, int], *, blend: float = 0.6, device: torch.device = None):
         self.layers_fov = layers_fov
         self.layers_res = layers_res
         self.out_res = out_res
@@ -23,15 +23,15 @@ class Foveation(object):
         ]  # blend maps of fovea layers
         self.coords = misc.meshgrid(*out_res).to(device=device)
 
-    def to(self, device):
-        self.eye_fovea_blend = [x.to(device=device)
-                                for x in self.eye_fovea_blend]
+    def to(self, device: torch.device):
+        self.eye_fovea_blend = [x.to(device=device) for x in self.eye_fovea_blend]
         self.coords = self.coords.to(device=device)
         return self
 
-    def synthesis(self, layers: List[torch.Tensor],
-                  fovea_center: Tuple[float, float],
-                  shifts: List[int] = None, do_blend=True) -> torch.Tensor:
+    def synthesis(self, layers: List[torch.Tensor], fovea_center: Tuple[float, float],
+                  shifts: List[int] = None,
+                  do_blend: bool = True,
+                  crop_mode: bool = False) -> torch.Tensor:
         """
         Generate foveated retinal image by blending fovea layers
         **Note: current implementation only support two fovea layers**
@@ -41,26 +41,32 @@ class Foveation(object):
         """
         output: torch.Tensor = nn_f.interpolate(layers[-1], self.out_res,
                                                 mode='bilinear', align_corners=False)
-        if shifts != None:
+        if shifts is not None:
             output = img.horizontal_shift(output, shifts[-1])
         c = torch.tensor([
-            fovea_center[0] + self.out_res[1] / 2,
-            fovea_center[1] + self.out_res[0] / 2
-        ], device=self.coords.device)
+            fovea_center[0] + (self.out_res[1] - 1) / 2,
+            fovea_center[1] + (self.out_res[0] - 1) / 2
+        ], device=self.device)
         for i in range(self.n_layers - 2, -1, -1):
-            if layers[i] == None:
+            if layers[i] is None:
                 continue
             R = self.get_layer_size_in_final_image(i) / 2
             grid = ((self.coords - c) / R)[None, ...]
-            if shifts != None:
+            if shifts is not None:
                 grid = img.horizontal_shift(grid, shifts[i], -2)
             # (1, 1, H:out, W:out)
             if do_blend:
-                blend = nn_f.grid_sample(self.eye_fovea_blend[i][None, None], grid, align_corners=False)
-                output.mul_(1 - blend).add_(nn_f.grid_sample(layers[i], grid, align_corners=False) * blend)
+                blend = nn_f.grid_sample(self.eye_fovea_blend[i][None, None], grid,
+                                         align_corners=False)
+            else:
+                blend = nn_f.grid_sample(torch.ones_like(self.eye_fovea_blend[i][None, None]), grid,
+                                         align_corners=False)
+            output.mul_(1 - blend)
+            if crop_mode:
+                output.add_(blend * nn_f.interpolate(layers[i], self.out_res, mode='bilinear',
+                                                     align_corners=False))
             else:    
-                blend = nn_f.grid_sample(torch.ones_like(self.eye_fovea_blend[i][None, None]), grid, align_corners=False)
-                output.mul_(1 - blend).add_(nn_f.grid_sample(layers[i], grid, align_corners=False) * blend)
+                output.add_(blend * nn_f.grid_sample(layers[i], grid, align_corners=False))
         return output
 
     def get_layer_size_in_final_image(self, i: int) -> int:
@@ -98,7 +104,7 @@ class Foveation(object):
         r = torch.norm(p - R, dim=2)  # (size, size, 2)
         return misc.smooth_step(R, R * self.blend, r)
 
-    def get_layers_mask(self, gaze) -> List[torch.Tensor]:
+    def get_layers_mask(self, gaze=None) -> List[torch.Tensor]:
         """
         Generate mask images for layers[:-1]
         the meaning of values in mask images:
@@ -111,15 +117,19 @@ class Foveation(object):
         """
         layers_mask = []
         for i in range(self.n_layers):
-            layers_mask.append(torch.ones(*self.layers_res[i], device=self.device) * -1)
             if i == self.n_layers - 1:
+                if gaze is None:
+                    layers_mask.append(torch.ones(*self.layers_res[i], device=self.device))
+                    continue
                 c = torch.tensor([
                     (gaze[0] + 0.5 * self.out_res[1]) / self.out_res[0],
                     (gaze[1] + 0.5 * self.out_res[0]) / self.out_res[0]
                 ], device=self.device)
             else:
                 c = torch.tensor([0.5, 0.5], device=self.device)
-            coord = misc.meshgrid(*self.layers_res[i]).to(device=self.device) / self.layers_res[i][0]
+            layers_mask.append(torch.ones(*self.layers_res[i], device=self.device) * -1)
+            coord = misc.meshgrid(
+                *self.layers_res[i]).to(device=self.device) / self.layers_res[i][0]
             r = 2 * torch.norm(coord - c, dim=-1)
             inner_radius = self.get_source_layer_cover_size_in_target_layer(
                 self.layers_fov[i - 1], self.layers_fov[i], self.layers_res[i][0]) / self.layers_res[i][0] \
@@ -129,7 +139,7 @@ class Foveation(object):
             else:
                 bounds = [inner_radius * (1 - self.blend), inner_radius, self.blend, 1]
             for bi in range(len(bounds) - 1):
-                region = torch.logical_and(r > bounds[bi], r <= bounds[bi + 1])
+                region = torch.logical_and(r >= bounds[bi], r < bounds[bi + 1])
                 layers_mask[i][region] = bi + \
                     (r[region] - bounds[bi]) / (bounds[bi + 1] - bounds[bi])
         return layers_mask

configs/bgnet.py

 def update_config(config):
     # Net parameters
-    config.NET_TYPE = 'bgnet'
-    config.N_ENCODE_DIM = 10
-    config.FC_PARAMS.update({
+    config.net = 'bgnet'
+    config.n_pos_encode = 10
+    config.fc.update({
         'nf': 128,
         'n_layers': 4
     })

configs/cnerf.py

 def update_config(config):
     # Net parameters
-    config.NET_TYPE = 'cnerf128'
-    config.N_ENCODE_DIM = 10
-    config.DEPTH_REF = True
-    config.FC_PARAMS.update({
+    config.net = 'cnerf128'
+    config.n_pos_encode = 10
+    config.depth_ref = True
+    config.fc.update({
         'nf': 256,
         'n_layers': 8
     })
-    config.SAMPLE_PARAMS.update({
-        'depth_range': (0.4, 6),
+    config.sa.update({
+        'sample_range': (0.4, 6),
         'n_samples': 8,
         'perturb_sample': False
     })

configs/dnerfabins.py

 def update_config(config):
     # Net parameters
-    config.NET_TYPE = 'dnerfa'
-    config.N_ENCODE_DIM = 10
-    config.DEPTH_REF = True
-    config.FC_PARAMS.update({
+    config.net = 'dnerfa'
+    config.n_pos_encode = 10
+    config.depth_ref = True
+    config.fc.update({
         'nf': 256,
         'n_layers': 8
     })
-    config.SAMPLE_PARAMS.update({
-        'depth_range': (0.4, 6),
+    config.sa.update({
+        'sample_range': (0.4, 6),
         'n_samples': 8,
         'perturb_sample': False
     })
\ No newline at end of file

configs/fovea.py

 def update_config(config):
     # Net parameters
-    config.NET_TYPE = 'snerffast4'
-    config.N_ENCODE_DIM = 6
-    #config.N_DIR_ENCODE = 4
-    config.FC_PARAMS.update({
+    config.net = 'snerffast4'
+    config.n_pos_encode = 6
+    #config.n_dir_encode = 4
+    config.fc.update({
         'nf': 512,
         'n_layers': 4
     })
-    config.SAMPLE_PARAMS.update({
-        'depth_range': (1, 30),
+    config.sa.update({
+        'sample_range': (1, 50),
         'n_samples': 64,
         'perturb_sample': False
     })

configs/fovea_small_rot1.py

+def update_config(config):
+    # Net parameters
+    config.net = 'snerffast4'
+    config.n_pos_encode = 6
+    #config.n_dir_encode = 4
+    config.fc.update({
+        'nf': 256,
+        'n_layers': 8
+    })
+    config.sa.update({
+        'sample_range': (1, 50),
+        'n_samples': 64,
+        'perturb_sample': False
+    })

configs/fovea_small_trans.py

 def update_config(config):
     # Net parameters
-    config.NET_TYPE = 'snerffast4'
-    config.N_ENCODE_DIM = 6
-    #config.N_DIR_ENCODE = 4
-    config.FC_PARAMS.update({
+    config.net = 'snerffast4'
+    config.n_pos_encode = 6
+    #config.n_dir_encode = 4
+    config.fc.update({
         'nf': 256,
         'n_layers': 8
     })
-    config.SAMPLE_PARAMS.update({
-        'depth_range': (1, 50),
+    config.sa.update({
+        'sample_range': (1, 50),
         'n_samples': 64,
         'perturb_sample': False
     })

configs/msl2fast.py

 def update_config(config):
     # Net parameters
-    config.NET_TYPE = 'msl2fast'
-    config.FC_PARAMS.update({
+    config.net = 'msl2fast'
+    config.fc.update({
         'nf': 256,
         'n_layers': 8
     })
-    config.SAMPLE_PARAMS.update({
-        'depth_range': (0.5, 5),
+    config.sa.update({
+        'sample_range': (0.5, 5),
         'n_samples': 128,
         'perturb_sample': False
     })

configs/msl_fovea.py

 def update_config(config):
     # Net parameters
-    config.NET_TYPE = 'msl'
-    config.FC_PARAMS.update({
+    config.net = 'msl'
+    config.fc.update({
         'nf': 128,
         'n_layers': 4
     })
-    config.SAMPLE_PARAMS.update({
-        'depth_range': (1, 50),
+    config.sa.update({
+        'sample_range': (1, 50),
         'n_samples': 32
     })

configs/mslfast.py

 def update_config(config):
     # Net parameters
-    config.NET_TYPE = 'mslfast'
-    config.N_ENCODE_DIM = 6
-    config.FC_PARAMS.update({
+    config.net = 'mslfast'
+    config.n_pos_encode = 6
+    config.fc.update({
         'nf': 256,
         'n_layers': 8
     })
-    config.SAMPLE_PARAMS.update({
-        'depth_range': (0.5, 5),
+    config.sa.update({
+        'sample_range': (0.5, 5),
         'n_samples': 128,
         'perturb_sample': False
     })

configs/mslray.py

 def update_config(config):
     # Net parameters
-    config.NET_TYPE = 'mslray'
-    config.FC_PARAMS.update({
+    config.net = 'mslray'
+    config.fc.update({
         'nf': 256,
         'n_layers': 8
     })
-    config.SAMPLE_PARAMS.update({
-        'depth_range': (0.5, 5),
+    config.sa.update({
+        'sample_range': (0.5, 5),
         'n_samples': 128,
         'perturb_sample': False
     })

configs/nerf.py

 def update_config(config):
     # Net parameters
-    config.NET_TYPE = 'nerf'
-    config.N_ENCODE_DIM = 10
-    config.FC_PARAMS.update({
+    config.net = 'nerf'
+    config.n_pos_encode = 10
+    config.fc.update({
         'nf': 256,
         'n_layers': 8
     })
-    config.SAMPLE_PARAMS.update({
-        'depth_range': (1, 50),
+    config.sa.update({
+        'sample_range': (1, 50),
         'n_samples': 64,
         'perturb_sample': True
     })
@@ -15,5 +15,5 @@ def update_config(config):
         'enable': True,
         'nf': 256,
         'n_layers': 8,
-        'additional_samples': 128
+        'n_samples': 128
     })
\ No newline at end of file

configs/nerf_horns.py

+def update_config(config):
+    # Net parameters
+    config.net = 'snerffast4'
+    config.n_pos_encode = 6
+    #config.n_dir_encode = 4
+    config.fc.update({
+        'nf': 256,
+        'n_layers': 8
+    })
+    config.sa.update({
+        'sample_range': (1, 50),
+        'n_samples': 64,
+        'perturb_sample': False
+    })

configs/nerf_horns_4.py

+nerf_horns.py
\ No newline at end of file