fixed memory leak

Author: deepbeepmeep

README.md

@@ -20,7 +20,7 @@ WanGP supports the Wan (and derived models), Hunyuan Video and LTV Video models
 **Follow DeepBeepMeep on Twitter/X to get the Latest News**: https://x.com/deepbeepmeep
 
 ## 🔥 Latest Updates : 
-### August 29 2025: WanGP v8.2 -  Here Goes Your Weekend
+### August 29 2025: WanGP v8.21 -  Here Goes Your Weekend
 
 - **InfiniteTalk Video to Video**: this feature can be used for Video Dubbing. Keep in mind that it is a *Sparse Video to Video*, that is internally only image is used by Sliding Window. However thanks to the new *Smooth Transition* mode, each new clip is connected to the previous and all the camera work is done by InfiniteTalk. If you dont get any transition, increase the number of frames of a Sliding Window (81 frames recommended)
 

models/wan/any2video.py

@@ -778,13 +778,11 @@ def generate(self,
             from preprocessing.face_preprocessor  import FaceProcessor 
             standin_ref_pos = 1 if "K" in video_prompt_type else 0
             if len(original_input_ref_images) < standin_ref_pos + 1: 
-                if "I" in video_prompt_type:
+                if "I" in video_prompt_type and model_type in ["vace_standin_14B"]:
                     print("Warning: Missing Standin ref image, make sure 'Inject only People / Objets' is selected or if there is 'Landscape and then People or Objects' there are at least two ref images.")
             else: 
                 standin_ref_pos = -1
                 image_ref = original_input_ref_images[standin_ref_pos]
-                image_ref.save("si.png")
-                # face_processor = FaceProcessor(antelopv2_path="ckpts/antelopev2")
                 face_processor = FaceProcessor()
                 standin_ref = face_processor.process(image_ref, remove_bg = model_type in ["vace_standin_14B"])
                 face_processor = None

wgp.py

@@ -81,6 +81,7 @@ def clear_gen_cache():
 
 def release_model():
     global wan_model, offloadobj, reload_needed
+    wan_model = None    
     clear_gen_cache()
     offload.shared_state
     if offloadobj is not None: