update: batch mode

app.py

@@ -73,7 +73,15 @@ def _normalize_batch_inputs(source_images, characters, reference_images) -> Tupl
         if isinstance(source_images, Image.Image):
             content_inputs = [source_images]
         elif isinstance(source_images, list):
-            content_inputs = source_images
+            # Handle Gradio Gallery format: list of tuples (image, caption)
+            content_inputs = []
+            for item in source_images:
+                if isinstance(item, tuple) and len(item) >= 1:
+                    # Extract the image from tuple (image, caption)
+                    content_inputs.append(item[0])
+                elif isinstance(item, Image.Image):
+                    # Direct image
+                    content_inputs.append(item)
         else:
             return [], [], [], 0
 
@@ -96,8 +104,9 @@ def _normalize_batch_inputs(source_images, characters, reference_images) -> Tupl
 
 @spaces.GPU()
 def run_fontdiffuer_batch(source_images: Union[List[Image.Image], Image.Image, None],
-                         characters: Union[List[str], str, None],
-                         reference_images: Union[List[Image.Image], Image.Image],
+                        #  characters: Union[List[str], str, None],
+                        #  reference_images: Union[List[Image.Image], Image.Image],
+                        reference_image: Image.Image,
                          sampling_step: int = 50,
                          guidance_scale: float = 7.5,
                          batch_size: int = 4,
@@ -118,6 +127,10 @@ def run_fontdiffuer_batch(source_images: Union[List[Image.Image], Image.Image, N
         List of generated images
     """
 
+    args.adaptive_batch_size = True
+    characters = None
+    reference_images = [reference_image]
+
     # Normalize inputs to lists
     content_inputs, style_inputs, char_inputs, total_samples = _normalize_batch_inputs(
         source_images, characters, reference_images
@@ -284,5 +297,28 @@ if __name__ == '__main__':
                     guidance_scale,
                     batch_size],
             outputs=fontdiffuer_output_image)
+        
+        # Batch Mode
+        gr.Markdown("## Batch Mode")
+        with gr.Row():
+            input_images = gr.Gallery(
+                format='png',
+                file_types=['image'],
+                type='pil',
+            )
+
+            reference_image = gr.Image(label='Reference Image', image_mode='RGB', type='pil')
+
+            output_images = gr.Gallery(
+                format='png',
+                type='pil'
+            )
+
+        RunFontDiffuserBatch = gr.Button('Run FontDiffuser Batch Mode')
+        RunFontDiffuserBatch.click(
+            fn=run_fontdiffuer_batch,
+            inputs=[input_images, reference_image],
+            outputs=output_images
+        )
+
     demo.launch(debug=True)
-    demo.queue(default_concurrency_limit=16)