Update main.py

main.py

@@ -56,22 +56,19 @@ async def download_file(url, local_path):
         return False
 
 async def create_slideshow(image_paths, audio_path, output_path, duration, zoom=False, zoom_ratio=0.04):
-    """Generate slideshow from images and audio using ffmpeg asynchronously"""
-    # Create temporary file list for ffmpeg concat
-    concat_file = "temp_concat.txt"
-    
+    """Generate slideshow from images and audio using ffmpeg asynchronously, with optional zoom effect"""
     if not zoom:
-        # Original implementation without zoom effect
+        # Create temporary file list for ffmpeg concat
+        concat_file = "temp_concat.txt"
         async with aiofiles.open(concat_file, "w") as f:
             for img in image_paths:
                 await f.write(f"file '{img}'\n")
                 await f.write(f"duration {duration}\n")
-            
             # Add the last image again without duration (required by ffmpeg)
             if image_paths:
                 await f.write(f"file '{image_paths[-1]}'\n")
         
-        # Run ffmpeg command to create slideshow with audio
+        # Run ffmpeg command to create slideshow with audio (no zoom)
         total_duration = len(image_paths) * duration
         cmd = [
             "ffmpeg",
@@ -88,31 +85,39 @@ async def create_slideshow(image_paths, audio_path, output_path, duration, zoom=
             output_path
         ]
     else:
-        # Build separate inputs looping each image with a fixed display time and apply zoom effect
-        cmd_inputs = []
+        # Generate slideshow with zoom effect using ffmpeg zoompan filter.
+        # For each image, loop it for the specified duration and apply zoompan.
+        fps = 25  # frames per second
+        cmd = ["ffmpeg"]
+        # Add each image as an input that loops for 'duration' seconds.
         for img in image_paths:
-            cmd_inputs += ["-loop", "1", "-t", str(duration), "-i", img]
-        # Add the audio input at the end
-        cmd_inputs += ["-i", audio_path]
-
-        # Create filter_complex applying zoompan on each image (assuming 25 fps)
-        num_images = len(image_paths)
-        filter_parts = []
-        for i in range(num_images):
-            filter_parts.append(
-            f"[{i}:v]zoompan=z='min(zoom+{zoom_ratio},1.5)':d={duration * 25}:s=hd720,setsar=1[v{i}]"
+            cmd.extend(["-loop", "1", "-t", str(duration), "-i", img])
+        # Append the audio input. Its index will be len(image_paths)
+        cmd.extend(["-i", audio_path])
+        
+        # Build filter_complex for zoompan on each image.
+        filter_complex = ""
+        for i in range(len(image_paths)):
+            # Each input's video stream is processed with zoompan.
+            # The zoom increases by zoom_ratio per frame (starting at 1.0).
+            filter_complex += (
+                f"[{i}:v]zoompan=z='if(eq(on,0),1,zoom+{zoom_ratio})':"
+                f"d={duration*fps}:s=hd720, setpts=PTS-STARTPTS[v{i}];"
             )
-        # Concatenate all processed clips
-        concat_inputs = "".join(f"[v{i}]" for i in range(num_images))
-        filter_complex = "; ".join(filter_parts) + f"; {concat_inputs}concat=n={num_images}:v=1:a=0,format=yuv420p[v]"
-
-        # Build the ffmpeg command with proper mapping for video and audio (audio is after the images)
-        cmd = ["ffmpeg", "-y"] + cmd_inputs + [
+        # Concatenate all processed video segments.
+        inputs = "".join(f"[v{i}]" for i in range(len(image_paths)))
+        filter_complex += f"{inputs}concat=n={len(image_paths)}:v=1:a=0,format=yuv420p[v]"
+        
+        # Map the concatenated video and the audio stream (audio input is at index len(image_paths))
+        cmd = cmd + [
             "-filter_complex", filter_complex,
             "-map", "[v]",
-            "-map", f"{num_images}:a",
+            "-map", f"{len(image_paths)}:a",
+            "-c:v", "libx264",
+            "-pix_fmt", "yuv420p",
             "-c:a", "aac",
             "-shortest",
+            "-y",
             output_path
         ]
     
@@ -124,8 +129,9 @@ async def create_slideshow(image_paths, audio_path, output_path, duration, zoom=
         )
         stdout, stderr = await process.communicate()
         
-        if os.path.exists(concat_file):
-            os.remove(concat_file)
+        # Remove temporary concat file if it exists
+        if not zoom and os.path.exists("temp_concat.txt"):
+            os.remove("temp_concat.txt")
             
         if process.returncode != 0:
             print(f"FFmpeg error: {stderr.decode()}")
@@ -133,8 +139,8 @@ async def create_slideshow(image_paths, audio_path, output_path, duration, zoom=
         return True
     except Exception as e:
         print(f"FFmpeg error: {str(e)}")
-        if os.path.exists(concat_file):
-            os.remove(concat_file)
+        if not zoom and os.path.exists("temp_concat.txt"):
+            os.remove("temp_concat.txt")
         return False
 
 @app.post("/make_slideshow")