1. improved system prompt

2. added read time estimation for triggering more image prompts per scene(hopefully)

app.py

@@ -5,7 +5,7 @@ from utils import get_scenes, generate_video_assets, generate_video  # Import th
 # Streamlit app
 st.title("Text to Video Generator")
 
-# Text input box with a max of 1500 characters
+# Text input box with a max of 1500 characters 
 text_script = st.text_area("Enter your text (max 1500 characters):", max_chars=1500)
 
 
@@ -30,7 +30,6 @@ if st.button("Generate Video"):
         # Call the function from utils.py to process the text
         scenes = get_scenes(text_script)
         video_assets_folder = generate_video_assets(scenes, language, selected_speaker)
-        st.write(video_assets_folder)
         generated_video_path = generate_video(video_assets_folder)
         st.video(generated_video_path)
 

utils.py

@@ -14,15 +14,17 @@ import os
 
 def get_scenes(text_script: str):
 
+    read_time = calculate_read_time(text_script)
     prompt = f"""
     ROLE: Story to Scene Generator
     Tasks: For the given story
     1. Read it Completely and Understand the Complete Context
-    2. Rewrite the story in tiny segments(but without even changing a word) and a descriptive image or list of image prompts to visualize each segment
-    3. Make sure each image prompt matches the theme of overall scene and ultimately the story 
-    4. If necessary, a scene can have more than one image prompts
+    2. Rewrite the story in tiny scenes(but without even changing a word) with highly detailed and context aware image or list of image prompts to visualize each scene
+    3. If necessary, a scene can have more than one image prompts
+    4. Make sure there is an image prompt for every 4-5 sec 
 
-    Here is the Complete Story: {text_script}
+    here is the Estimated Read Time of the complete story: {read_time}\n\n
+    and Here is the Complete Story: {text_script}
     """
     
     
@@ -393,6 +395,47 @@ def generate_video_old(audio_file, images, segments):
     except Exception as e:
         print(f"Error generating video: {e}")
         return None
+    
+
+
+def calculate_read_time(text: str, words_per_minute: int = 155) -> str:
+    """
+    Calculate how long it will take to read a given text.
+
+    Args:
+        text (str): The input text to calculate reading time for.
+        words_per_minute (int): Average reading speed in words per minute. Default is 155(an uneducated guess).
+
+    Returns:
+        str: A string describing the reading time in seconds, minutes, or hours.
+    """
+    try:
+        # Validate input
+        if not text or not isinstance(text, str):
+            return "Invalid input: Text must be a non-empty string."
+
+        # Calculate the number of words in the text
+        words = text.split()
+        word_count = len(words)
+
+        # Calculate total reading time in seconds
+        total_seconds = (word_count / words_per_minute) * 60
+
+        # Convert to hours, minutes, and seconds
+        hours = int(total_seconds // 3600)
+        minutes = int((total_seconds % 3600) // 60)
+        seconds = int(total_seconds % 60)
+
+        # Format the output based on the duration
+        if hours > 0:
+            return f"Reading time: {hours} hour(s), {minutes} minute(s), and {seconds} second(s)."
+        elif minutes > 0:
+            return f"Reading time: {minutes} minute(s) and {seconds} second(s)."
+        else:
+            return f"Reading time: {seconds} second(s)."
+
+    except Exception as e:
+        return f"An error occurred: {e}"
 
 
 
@@ -407,13 +450,3 @@ if __name__ == "__main__":
     Lily knew she had found something truly extraordinary.
     """
     generate_audio(short_story, "Urdu", "Asad")
-    # scenes_response = get_scenes(short_story)
-    # scenes = scenes_response.get("scenes")
-    # print("total scenes: ", len(scenes))
-    # for scene in scenes:
-    #     print("image prompts for this scene", len(scene.get("image_prompts")))
-    #     print("\n\n")
-    # for scene_count,scene in enumerate(scenes):
-    #     image_prompts = scene.get("image_prompts")
-    #     for count, prompt in enumerate(image_prompts):
-    #         generate_image(prompt=prompt, path=f"scene_{scene_count+1}_image_{count+1}.png")