Update inference_unianimate_entrance.py

tools/inferences/inference_unianimate_entrance.py

@@ -406,6 +406,9 @@ def worker(gpu, seed, steps, useFirstFrame, reference_image, ref_pose, pose_sequ
         noise = torch.randn([1, 4, frames_num, int(cfg.resolution[1]/cfg.scale), int(cfg.resolution[0]/cfg.scale)])
         noise = noise.to(gpu)      
         # print(f"noise: {noise.shape}")
+
+        # add a noise prior
+        noise = diffusion.q_sample(random_ref_frame.clone(), getattr(cfg, "noise_prior_value", 949), noise=noise)
         
         
         if hasattr(cfg.Diffusion, "noise_strength"):
@@ -417,8 +420,7 @@ def worker(gpu, seed, steps, useFirstFrame, reference_image, ref_pose, pose_sequ
 
     
         
-        # add a noise prior
-        noise = diffusion.q_sample(random_ref_frame.clone(), getattr(cfg, "noise_prior_value", 949), noise=noise)
+        
 
         # construct model inputs (CFG)
         full_model_kwargs=[{
@@ -489,6 +491,8 @@ def worker(gpu, seed, steps, useFirstFrame, reference_image, ref_pose, pose_sequ
                 ddim_timesteps=steps,
                 eta=0.0)
             
+            # print(f"video_data dtype: {video_data.dtype}")
+            
             if hasattr(cfg, "CPU_CLIP_VAE") and cfg.CPU_CLIP_VAE:
                 # if run forward of  autoencoder or clip_encoder second times, load them again
                 clip_encoder.cuda()
@@ -504,15 +508,12 @@ def worker(gpu, seed, steps, useFirstFrame, reference_image, ref_pose, pose_sequ
             video_data = torch.cat(decode_data, dim=0)
             video_data = rearrange(video_data, '(b f) c h w -> b c f h w', b = cfg.batch_size).float()
 
-            # Check sth
-                
-            # print(f' video_data is of shape ({video_data.shape})')
-            # print(f' video_data is ({video_data})')
 
             del model_kwargs_one_vis[0][list(model_kwargs_one_vis[0].keys())[0]]
             del model_kwargs_one_vis[1][list(model_kwargs_one_vis[1].keys())[0]]
 
             video_data = extract_image_tensors(video_data.cpu(), cfg.mean, cfg.std)
+
             
             # synchronize to finish some processes
             if not cfg.debug: