feat: Update app.py to use new Tile-Upscaler client

app.py

@@ -41,7 +41,6 @@ def infer(selected_index, prompt, style_prompt, inf_steps, guidance_scale, width
         trigger_word = selected_lora["trigger_word"]
 
         global client_custom_model
-        global client_tile_upscaler
 
         if client_custom_model is None:
             try:
@@ -52,16 +51,6 @@ def infer(selected_index, prompt, style_prompt, inf_steps, guidance_scale, width
                 client_custom_model = None
                 raise gr.Error("Failed to load client for " + custom_model_url)
 
-        if client_tile_upscaler is None:
-            try:
-                client_tile_upscaler = Client(tile_upscaler_url)
-                print(f"Loaded custom model from {tile_upscaler_url}")
-            except ValueError as e:
-                print(f"Failed to load custom model: {e}")
-                client_custom_model = None
-                raise gr.Error("Failed to load client for " + tile_upscaler_url)
-
-
 
         result = client_custom_model.submit(
             custom_model=custom_model,
@@ -111,18 +100,29 @@ def update_selection(evt: gr.SelectData):
 
 
 def upscale_image(image, resolution, num_inference_steps, strength, hdr, guidance_scale, controlnet_strength, scheduler_name):
-  result = client_tile_upscaler.predict(
-      param_0=handle_file(image),
-      param_1=resolution,
-      param_2=num_inference_steps,
-      param_3=strength,
-      param_4=hdr,
-      param_5=guidance_scale,
-      param_6=controlnet_strength,
-      param_7=scheduler_name,
-      api_name="/wrapper"
-  )
-  return result
+    global client_tile_upscaler
+
+    if client_tile_upscaler is None:
+        try:
+            client_tile_upscaler = Client(tile_upscaler_url)
+            print(f"Loaded custom model from {tile_upscaler_url}")
+        except ValueError as e:
+            print(f"Failed to load custom model: {e}")
+            client_custom_model = None
+            raise gr.Error("Failed to load client for " + tile_upscaler_url)
+
+    result = client_tile_upscaler.predict(
+        param_0=handle_file(image),
+        param_1=resolution,
+        param_2=num_inference_steps,
+        param_3=strength,
+        param_4=hdr,
+        param_5=guidance_scale,
+        param_6=controlnet_strength,
+        param_7=scheduler_name,
+        api_name="/wrapper"
+    )
+    return result