updated app

app.py

@@ -200,22 +200,19 @@ def video_inference(
 
 with gr.Blocks() as demo:
     gr.Markdown("# **<p align='center'>ROME: Realistic one-shot mesh-based head avatars</p>**")
-
+    
     gr.HTML(
         """
         <div style='text-align: center'>
             <img src='file/media/tease.gif' alt='Teaser'>
         </div>
-        """
-    )
-    
-    gr.Markdown(
-        """
+
         <p style='text-align: center'>
-            
             Create a personal avatar from just a single image using ROME.
             <br> <a href='https://arxiv.org/abs/2206.08343' target='_blank'>Paper</a> | <a href='https://samsunglabs.github.io/rome' target='_blank'>Project Page</a> | <a href='https://github.com/SamsungLabs/rome' target='_blank'>Github</a>
+        </p>
 
+        <p style='text-align: center'>
             <blockquote>
                 [The] system creates realistic mesh-based avatars from a single <strong>source</strong>
                 photo. These avatars are rigged, i.e., they can be driven by the animation parameters from a different <strong>driving</strong> frame.