fewer cloth_loop and use GPU-MC

apps/infer.py

@@ -50,7 +50,7 @@ def generate_model(in_path, model_type):
     torch.cuda.empty_cache()
 
     config_dict = {'loop_smpl': 100,
-                   'loop_cloth': 200,
+                   'loop_cloth': 100,
                    'patience': 5,
                    'out_dir': './results',
                    'hps_type': 'pymaf',

lib/common/seg3d_lossless.py

@@ -584,7 +584,7 @@ class Seg3dLossless(nn.Module):
 
         final = occupancys[1:, 1:, 1:].contiguous()
 
-        if final.shape[0] >= 256:
+        if final.shape[0] > 256:
             # for voxelgrid larger than 256^3, the required GPU memory will be > 9GB
             # thus we use CPU marching_cube to avoid "CUDA out of memory"
             occu_arr = final.detach().cpu().numpy()                 # non-smooth surface