Update app.py

app.py

@@ -319,6 +319,8 @@ def auto_detect_face_area(image):
 pipe_txt2img = None
 current_pipe_model_id = None
 pipe_img2img = None
+pipe_img2img_pose = None
+pipe_img2img_depth = None
 
 
 #Das Laden des Modells bedeutet, die trainierten Gewichte (Parameter) von der Festplatte zu lesen und 
@@ -463,49 +465,112 @@ def load_txt2img(model_id):
             print(f"❌ Auch Fallback fehlgeschlagen: {fallback_error}")
             raise
 
-def load_img2img():
-    """Lädt Multi-ControlNet-Inpainting-Pipeline (Pose + Canny)"""
-    global pipe_img2img
-    if pipe_img2img is None:
-        print("🔄 Lade Multi-ControlNet-Inpainting-Modell (OpenPose + Canny)...")
-        try:
-            # LADE BEIDE ControlNet-Modelle
-            controlnet_openpose = ControlNetModel.from_pretrained(
-                "lllyasviel/sd-controlnet-openpose",
-                torch_dtype=torch_dtype
-            )
-            controlnet_canny = ControlNetModel.from_pretrained(
-                "lllyasviel/sd-controlnet-canny",
-                torch_dtype=torch_dtype
-            )
-            
-            # ÜBERGEBE EINE LISTE an die Pipeline
-            pipe_img2img = StableDiffusionControlNetInpaintPipeline.from_pretrained(
-                "runwayml/stable-diffusion-v1-5",
-                controlnet=[controlnet_openpose, controlnet_canny],  # <-- Liste
-                torch_dtype=torch_dtype,
-                safety_checker=None,
-                requires_safety_checker=False
-            ).to(device)
-            
-            # Scheduler konfigurieren (Ihre bestehende Logik)
-            pipe_img2img.scheduler = DPMSolverMultistepScheduler.from_config(
-                pipe_img2img.scheduler.config,
-                algorithm_type="sde-dpmsolver++",
-                use_karras_sigmas=True,
-                timestep_spacing="trailing"
-            )
-            print("✅ Multi-ControlNet-Inpainting-Pipeline geladen (OpenPose + Canny)")
-            
-        except Exception as e:
-            print(f"❌ Fehler beim Laden der Multi-ControlNet-Inpainting-Pipeline: {e}")
-            raise
-
-        # Optimierungen
-        pipe_img2img.enable_attention_slicing()
-        print("✅ Multi-ControlNet-Inpainting-Pipeline optimiert")
 
-    return pipe_img2img
+def load_img2img(keep_environment=False):
+    """
+    Lädt die Multi-ControlNet-Inpainting-Pipeline dynamisch basierend auf dem Modus.
+    
+    Args:
+        keep_environment (bool): 
+            - False: Lädt OpenPose + Canny (für 'Focus verändern')
+            - True:  Lädt Depth + Canny (für 'Umgebung ändern' oder 'Ausschließlich Gesicht')
+    
+    Returns:
+        Die korrekt konfigurierte Pipeline
+    """
+    global pipe_img2img_pose, pipe_img2img_depth
+    
+    # Initialisiere globale Variablen, falls noch nicht geschehen
+    if 'pipe_img2img_pose' not in globals():
+        pipe_img2img_pose = None
+    if 'pipe_img2img_depth' not in globals():
+        pipe_img2img_depth = None
+    
+    if keep_environment:
+        # ===== MODUS: Depth + Canny =====
+        if pipe_img2img_depth is None:
+            print("🔄 Lade Multi-ControlNet-Inpainting-Modell (Depth + Canny)...")
+            try:
+                # LADE BEIDE ControlNet-Modelle für Depth-Modus
+                controlnet_depth = ControlNetModel.from_pretrained(
+                    "lllyasviel/sd-controlnet-depth",
+                    torch_dtype=torch_dtype
+                )
+                controlnet_canny = ControlNetModel.from_pretrained(
+                    "lllyasviel/sd-controlnet-canny",
+                    torch_dtype=torch_dtype
+                )
+                
+                # WICHTIG: Reihenfolge muss mit prepare_controlnet_maps übereinstimmen!
+                # [Depth, Canny]
+                pipe_img2img_depth = StableDiffusionControlNetInpaintPipeline.from_pretrained(
+                    "runwayml/stable-diffusion-v1-5",
+                    controlnet=[controlnet_depth, controlnet_canny],  # Depth zuerst!
+                    torch_dtype=torch_dtype,
+                    safety_checker=None,
+                    requires_safety_checker=False
+                ).to(device)
+                
+                # Scheduler konfigurieren
+                pipe_img2img_depth.scheduler = DPMSolverMultistepScheduler.from_config(
+                    pipe_img2img_depth.scheduler.config,
+                    algorithm_type="sde-dpmsolver++",
+                    use_karras_sigmas=True,
+                    timestep_spacing="trailing"
+                )
+                
+                # Optimierungen
+                pipe_img2img_depth.enable_attention_slicing()
+                print("✅ Multi-ControlNet-Inpainting-Pipeline geladen (Depth + Canny)")
+                
+            except Exception as e:
+                print(f"❌ Fehler beim Laden der Depth+Canny Pipeline: {e}")
+                raise
+        
+        return pipe_img2img_depth
+    
+    else:
+        # ===== MODUS: OpenPose + Canny =====
+        if pipe_img2img_pose is None:
+            print("���� Lade Multi-ControlNet-Inpainting-Modell (OpenPose + Canny)...")
+            try:
+                # LADE BEIDE ControlNet-Modelle für Pose-Modus
+                controlnet_openpose = ControlNetModel.from_pretrained(
+                    "lllyasviel/sd-controlnet-openpose",
+                    torch_dtype=torch_dtype
+                )
+                controlnet_canny = ControlNetModel.from_pretrained(
+                    "lllyasviel/sd-controlnet-canny",
+                    torch_dtype=torch_dtype
+                )
+                
+                # WICHTIG: Reihenfolge muss mit prepare_controlnet_maps übereinstimmen!
+                # [OpenPose, Canny]
+                pipe_img2img_pose = StableDiffusionControlNetInpaintPipeline.from_pretrained(
+                    "runwayml/stable-diffusion-v1-5",
+                    controlnet=[controlnet_openpose, controlnet_canny],  # OpenPose zuerst!
+                    torch_dtype=torch_dtype,
+                    safety_checker=None,
+                    requires_safety_checker=False
+                ).to(device)
+                
+                # Scheduler konfigurieren
+                pipe_img2img_pose.scheduler = DPMSolverMultistepScheduler.from_config(
+                    pipe_img2img_pose.scheduler.config,
+                    algorithm_type="sde-dpmsolver++",
+                    use_karras_sigmas=True,
+                    timestep_spacing="trailing"
+                )
+                
+                # Optimierungen
+                pipe_img2img_pose.enable_attention_slicing()
+                print("✅ Multi-ControlNet-Inpainting-Pipeline geladen (OpenPose + Canny)")
+                
+            except Exception as e:
+                print(f"❌ Fehler beim Laden der OpenPose+Canny Pipeline: {e}")
+                raise
+        
+        return pipe_img2img_pose