working on the chat system

PROMPT_CONTEXT.md

@@ -1,12 +1,14 @@
-For some context:
+GENERAL CONTEXT:
 
-our app is an AI video generation platform, where people use the frontend app (written in Flutter) to virtually "search" for video (the video synopsis and their content are generated on the fly using AI). 
+This project is an app where users can generate videos using AI. What is interesting is that both search results are generated (so there is no actual search in a DB, instead a LLM hallucinate search result items, simulation a video platform à la YouTube), but also the video streams (a video is composed of an infinite stream of a few seconds long MP4 clips, that are also generated using AI, using a fast generative model that works in nearly real-time, eg it takes 4s to generate 2s of footage).
 
-Note that this uses a custom API written in Python, with a WebSockets communication.
+The architecture is simple: a Flutter frontend UI with two main view (home_screen.dart for search, video_screen.dart for the ifinite video stream player). The frontend UI talks to a Python API (see api.py) using WebSockets, as we have various real-time communication needs (chat, streaming of MP4 chunks etc). This Python API is responsible for performing the actual calls to the generative video model and the LLM model (those are external servers hosted on Hugging Face, but explaining how they work is outside the scope of this documentation).
 
-To go back to the Flutter app, when the user open a thumbnail card after doing such generative AI search, it opens a full view for a video (with a player, title, description, chat section etc and a special search bar in the top header that allows to edit the current video's theme). 
+There is a simulator integrated, which evolves a description (video prompt) over time, using a LLM.
 
+Users can be anonymous, but if they connect using a Hugging Face API key, they get some extra perks.
 
-Task to perform: <fill your demand here>
+TASK:
+<fill your demand here>
 
 Note: For the task to be validated, running the shell command "flutter build web" must succeeed.

api.py

@@ -155,6 +155,8 @@ async def websocket_handler(request: web.Request) -> web.WebSocketResponse:
                         request_type = 'video'
                     elif action == 'search':
                         request_type = 'search'
+                    elif action == 'simulate':
+                        request_type = 'simulation'
                     
                     # Record the request for metrics
                     await metrics_tracker.record_request(user_id, client_ip, request_type, user_role)
@@ -176,6 +178,8 @@ async def websocket_handler(request: web.Request) -> web.WebSocketResponse:
                         await user_session.video_queue.put(data)
                     elif action == 'search':
                         await user_session.search_queue.put(data)
+                    elif action == 'simulate':
+                        await user_session.simulation_queue.put(data)
                     else:
                         await user_session.process_generic_request(data)
                         

api_core.py

@@ -451,6 +451,126 @@ A video can be anything from a tutorial, webcam, trailer, movie, live stream etc
         except Exception as e:
             logger.error(f"Error generating caption: {str(e)}")
             return ""
+            
+    async def simulate(self, original_title: str, original_description: str, 
+                         current_description: str, condensed_history: str, 
+                         evolution_count: int = 0, chat_messages: str = '') -> dict:
+        """
+        Simulate a video by evolving its description to create a dynamic narrative.
+        
+        Args:
+            original_title: The original video title
+            original_description: The original video description
+            current_description: The current description (last evolved or original if first evolution)
+            condensed_history: A condensed summary of previous scene developments
+            evolution_count: How many times the simulation has already evolved
+            chat_messages: Chat messages from users to incorporate into the simulation
+            
+        Returns:
+            A dictionary containing the evolved description and updated condensed history
+        """
+        try:
+            # Determine if this is the first simulation
+            is_first_simulation = evolution_count == 0 or not condensed_history
+            
+            logger.info(f"simulate(): is_first_simulation={is_first_simulation}")
+                
+            # Create an appropriate prompt based on whether this is the first simulation
+            chat_section = ""
+            if chat_messages:
+                chat_section = f"""
+People are watching this content right now and have shared their thoughts. Like a game master, please take their feedback as input to adjust the story and/or the scene. Here are their messages:
+
+{chat_messages}
+"""
+
+            if is_first_simulation:
+                prompt = f"""You are tasked with evolving the narrative for a video titled: "{original_title}"
+
+Original description:
+{original_description}
+{chat_section}
+
+Instructions:
+1. Imagine the next logical scene or development that would follow this description.
+2. Create a compelling new description (200-300 words) that builds on the original but introduces new elements, developments, or perspectives.
+3. Maintain the original style, tone, and setting.
+4. If viewers have shared messages, consider their input and incorporate relevant suggestions or reactions into your narrative evolution.
+5. Also create a brief "scene history" (50-75 words) that summarizes what has happened so far.
+
+Return your response in this format:
+EVOLVED_DESCRIPTION: [your new evolved description here]
+CONDENSED_HISTORY: [your scene history summary]"""
+            else:
+                prompt = f"""You are tasked with continuing to evolve the narrative for a video titled: "{original_title}"
+
+Original description:
+{original_description}
+
+Condensed history of scenes so far:
+{condensed_history}
+
+Current description (most recent scene):
+{current_description}
+{chat_section}
+
+Instructions:
+1. Imagine the next logical scene or development that would follow the current description.
+2. Create a compelling new description (200-300 words) that builds on the narrative but introduces new elements, developments, or perspectives.
+3. Maintain consistency with the previous scenes while advancing the story.
+4. If viewers have shared messages, consider their input and incorporate relevant suggestions or reactions into your narrative evolution.
+5. Also update the condensed history (50-75 words) to include this new development.
+
+Return your response in this format:
+EVOLVED_DESCRIPTION: [your new evolved description here]
+CONDENSED_HISTORY: [your updated scene history summary]"""
+
+            # Generate the evolved description
+            response = await asyncio.get_event_loop().run_in_executor(
+                None,
+                lambda: self.inference_client.text_generation(
+                    prompt,
+                    model=TEXT_MODEL,
+                    max_new_tokens=200,
+                    temperature=0.7
+                )
+            )
+            
+            # Extract the evolved description and condensed history from the response
+            evolved_description = ""
+            new_condensed_history = ""
+            
+            # Parse the response
+            if "EVOLVED_DESCRIPTION:" in response and "CONDENSED_HISTORY:" in response:
+                parts = response.split("CONDENSED_HISTORY:")
+                if len(parts) >= 2:
+                    desc_part = parts[0].strip()
+                    if "EVOLVED_DESCRIPTION:" in desc_part:
+                        evolved_description = desc_part.split("EVOLVED_DESCRIPTION:", 1)[1].strip()
+                    new_condensed_history = parts[1].strip()
+            
+            # If parsing failed, use some fallbacks
+            if not evolved_description:
+                evolved_description = current_description
+                logger.warning(f"Failed to parse evolved description, using current description as fallback")
+            
+            if not new_condensed_history and condensed_history:
+                new_condensed_history = condensed_history
+                logger.warning(f"Failed to parse condensed history, using current history as fallback")
+            elif not new_condensed_history:
+                new_condensed_history = f"The video begins with {original_title}: {original_description[:100]}..."
+            
+            return {
+                "evolved_description": evolved_description,
+                "condensed_history": new_condensed_history
+            }
+            
+        except Exception as e:
+            logger.error(f"Error simulating video: {str(e)}")
+            return {
+                "evolved_description": current_description,
+                "condensed_history": condensed_history or f"The video shows {original_title}."
+            }
 
 
     def get_config_value(self, role: UserRole, field: str, options: dict = None) -> Any:
@@ -696,9 +816,9 @@ Your caption:"""
         elif orientation == 'LANDSCAPE' and height > width:
             # Swap height and width for landscape orientation
             height, width = width, height
-            #logger.info(f"Orientation: {orientation}, swapped dimensions to width={width}, height={height}, steps={num_inference_steps}, fps={frame_rate} | role: {user_role}")
+            logger.info(f"generate_video()  Orientation: {orientation}, swapped dimensions to width={width}, height={height}, steps={num_inference_steps}, fps={frame_rate} | role: {user_role}")
         else:
-            #logger.info(f"Orientation: {orientation}, using original dimensions width={width}, height={height}, steps={num_inference_steps}, fps={frame_rate} | role: {user_role}")
+            logger.info(f"generate_video()  Orientation: {orientation}, using original dimensions width={width}, height={height}, steps={num_inference_steps}, fps={frame_rate} | role: {user_role}")
             pass
         
         # Generate the video with standard settings

api_session.py

@@ -24,19 +24,22 @@ class UserSession:
         self.chat_queue = asyncio.Queue()
         self.video_queue = asyncio.Queue()
         self.search_queue = asyncio.Queue()
+        self.simulation_queue = asyncio.Queue()  # New queue for description evolution
         
         # Track request counts and rate limits
         self.request_counts = {
             'chat': 0,
             'video': 0,
-            'search': 0
+            'search': 0,
+            'simulation': 0  # New counter for simulation requests
         }
         
         # Last request timestamps for rate limiting
         self.last_request_times = {
             'chat': time.time(),
             'video': time.time(),
-            'search': time.time()
+            'search': time.time(),
+            'simulation': time.time()  # New timestamp for simulation requests
         }
         
         # Session creation time
@@ -50,7 +53,8 @@ class UserSession:
         self.background_tasks = [
             asyncio.create_task(self._process_chat_queue()),
             asyncio.create_task(self._process_video_queue()),
-            asyncio.create_task(self._process_search_queue())
+            asyncio.create_task(self._process_search_queue()),
+            asyncio.create_task(self._process_simulation_queue())  # New worker for simulation requests
         ]
         logger.info(f"Started session for user {self.user_id} with role {self.user_role}")
         
@@ -78,9 +82,13 @@ class UserSession:
                     result = await self.shared_api.handle_chat_message(data, self.ws)
                 elif data['action'] == 'leave_chat':
                     result = await self.shared_api.handle_leave_chat(data, self.ws)
-                # Handle thumbnail requests as chat requests for immediate processing
+                # Redirect thumbnail requests to process_generic_request for consistent handling
                 elif data['action'] == 'generate_video_thumbnail':
-                    result = await self._handle_thumbnail_request(data)
+                    # Pass to the generic request handler to maintain consistent logic
+                    await self.process_generic_request(data)
+                    # Skip normal response handling since process_generic_request already sends a response
+                    self.chat_queue.task_done()
+                    continue
                 else:
                     raise ValueError(f"Unknown chat action: {data['action']}")
                     
@@ -262,6 +270,82 @@ class UserSession:
                 if 'search_queue' in self.__dict__:
                     self.search_queue.task_done()
                     
+    async def _process_simulation_queue(self):
+        """Dedicated queue for video simulation requests"""
+        while True:
+            try:
+                data = await self.simulation_queue.get()
+                request_id = data.get('requestId')
+                
+                # Extract parameters from the request
+                video_id = data.get('video_id', '')
+                original_title = data.get('original_title', '')
+                original_description = data.get('original_description', '')
+                current_description = data.get('current_description', '')
+                condensed_history = data.get('condensed_history', '')
+                evolution_count = data.get('evolution_count', 0)
+                chat_messages = data.get('chat_messages', '')
+                
+                logger.info(f"Processing video simulation for user {self.user_id}, video_id={video_id}, evolution_count={evolution_count}")
+                
+                # Validate required parameters
+                if not original_title or not original_description or not current_description:
+                    result = {
+                        'action': 'simulate',
+                        'requestId': request_id,
+                        'success': False,
+                        'error': 'Missing required parameters'
+                    }
+                else:
+                    try:
+                        # Call the simulate method in the API
+                        simulation_result = await self.shared_api.simulate(
+                            original_title=original_title,
+                            original_description=original_description,
+                            current_description=current_description,
+                            condensed_history=condensed_history,
+                            evolution_count=evolution_count,
+                            chat_messages=chat_messages
+                        )
+                        
+                        result = {
+                            'action': 'simulate',
+                            'requestId': request_id,
+                            'success': True,
+                            'evolved_description': simulation_result['evolved_description'],
+                            'condensed_history': simulation_result['condensed_history']
+                        }
+                    except Exception as e:
+                        logger.error(f"Error simulating video for user {self.user_id}, video_id={video_id}: {str(e)}")
+                        result = {
+                            'action': 'simulate',
+                            'requestId': request_id,
+                            'success': False,
+                            'error': f'Simulation error: {str(e)}'
+                        }
+                
+                await self.ws.send_json(result)
+                
+                # Update metrics
+                self.request_counts['simulation'] += 1
+                self.last_request_times['simulation'] = time.time()
+                
+            except Exception as e:
+                logger.error(f"Error in simulation queue processor for user {self.user_id}: {str(e)}")
+                try:
+                    error_response = {
+                        'action': 'simulate',
+                        'requestId': data.get('requestId') if 'data' in locals() else None,
+                        'success': False,
+                        'error': f'Internal server error: {str(e)}'
+                    }
+                    await self.ws.send_json(error_response)
+                except Exception as send_error:
+                    logger.error(f"Error sending error response: {send_error}")
+            finally:
+                if 'simulation_queue' in self.__dict__:
+                    self.simulation_queue.task_done()
+                    
     async def process_generic_request(self, data: dict) -> None:
         """Handle general requests that don't fit into specialized queues"""
         try:
@@ -310,7 +394,9 @@ class UserSession:
                     'caption': caption
                 })
                 
-            elif action == 'generate_thumbnail' or action == 'generate_video_thumbnail':
+            # evolve_description is now handled by the dedicated simulation queue processor
+                
+            elif action == 'generate_video_thumbnail':
                 title = data.get('title', '') or data.get('params', {}).get('title', '')
                 description = data.get('description', '') or data.get('params', {}).get('description', '')
                 video_prompt_prefix = data.get('video_prompt_prefix', '') or data.get('params', {}).get('video_prompt_prefix', '')
@@ -345,61 +431,56 @@ class UserSession:
                         title, description, video_prompt_prefix, options, self.user_role
                     )
                     
-                    await self.ws.send_json({
-                        'action': action,
-                        'requestId': request_id,
-                        'success': True,
-                        'thumbnail': thumbnail_data,
-                    })
-                except Exception as e:
-                    logger.error(f"Error generating thumbnail: {str(e)}")
-                    await self.ws.send_json(error_response(f"Thumbnail generation failed: {str(e)}"))
-                    
-            elif action == 'old_generate_thumbnail' or action == 'generate_thumbnail':
-                # Redirect to video thumbnail generation instead of static image
-                title = data.get('params', {}).get('title')
-                description = data.get('params', {}).get('description')
-                
-                if not title or not description:
-                    await self.ws.send_json(error_response('Missing title or description'))
-                    return
-                
-                # Use the video thumbnail function instead
-                options = {
-                    'width': 512,
-                    'height': 288,
-                    'thumbnail': True,
-                    'video_id': f"thumbnail-{request_id}"
-                }
-                
-                try:
-                    thumbnail = await self.shared_api.generate_video_thumbnail(
-                        title, description, "", options, self.user_role
-                    )
-                    
-                    # Check thumbnail is not empty
-                    if thumbnail is None or thumbnail == "":
+                    # Respond with appropriate format based on the parameter names used in the request
+                    if 'thumbnailUrl' in data or 'thumbnailUrl' in data.get('params', {}):
+                        # Legacy format using thumbnailUrl
                         await self.ws.send_json({
                             'action': action,
                             'requestId': request_id,
                             'success': True,
-                            'thumbnailUrl': ""
+                            'thumbnailUrl': thumbnail_data or "",
                         })
                     else:
+                        # New format using thumbnail
                         await self.ws.send_json({
                             'action': action,
                             'requestId': request_id,
                             'success': True,
-                            'thumbnailUrl': thumbnail
+                            'thumbnail': thumbnail_data,
                         })
                 except Exception as e:
-                    logger.error(f"Error generating video thumbnail: {str(e)}")
-                    await self.ws.send_json({
-                        'action': action,
-                        'requestId': request_id,
-                        'success': True,  # Still return success to avoid client errors
-                        'thumbnailUrl': ""  # But with empty thumbnail
-                    })
+                    logger.error(f"Error generating thumbnail: {str(e)}")
+                    await self.ws.send_json(error_response(f"Thumbnail generation failed: {str(e)}"))
+                
+            # Handle deprecated thumbnail actions
+            elif action == 'generate_thumbnail' or action == 'old_generate_thumbnail':
+                # Redirect to video thumbnail generation
+                logger.warning(f"Deprecated thumbnail action '{action}' used, redirecting to generate_video_thumbnail")
+                
+                # Extract parameters
+                title = data.get('title', '') or data.get('params', {}).get('title', '')
+                description = data.get('description', '') or data.get('params', {}).get('description', '')
+                
+                if not title or not description:
+                    await self.ws.send_json(error_response('Missing title or description'))
+                    return
+                
+                # Create a new request with the correct action
+                new_request = {
+                    'action': 'generate_video_thumbnail',
+                    'requestId': request_id,
+                    'title': title,
+                    'description': description,
+                    'options': {
+                        'width': 512,
+                        'height': 288,
+                        'thumbnail': True,
+                        'video_id': f"thumbnail-{request_id}"
+                    }
+                }
+                
+                # Process with the new action
+                await self.process_generic_request(new_request)
                 
             else:
                 await self.ws.send_json(error_response(f'Unknown action: {action}'))
@@ -473,7 +554,8 @@ class SessionManager:
             'requests': {
                 'chat': 0,
                 'video': 0,
-                'search': 0
+                'search': 0,
+                'simulation': 0
             }
         }
         
@@ -482,5 +564,6 @@ class SessionManager:
             stats['requests']['chat'] += session.request_counts['chat']
             stats['requests']['video'] += session.request_counts['video']
             stats['requests']['search'] += session.request_counts['search']
+            stats['requests']['simulation'] += session.request_counts['simulation']
             
         return stats

assets/config/aitube.yaml

@@ -1,6 +1,6 @@
 ui:
   product_name: "#aitube2"
-  showChatInVideoView: true
+  showChatInVideoView: false
 
 render_queue:
   # how many clips should be stored in advance
@@ -12,6 +12,11 @@ render_queue:
   # start playback as soon as we have 1 video over 3 (25%)
   minimum_buffer_percent_to_start_playback: 5
 
+simulation:
+  # how often the description should evolve (in seconds)
+  # setting to 0 disables description evolution
+  sim_loop_frequency_in_sec: 10
+
 # it's OK to use high values here,
 # because some of those values are limited by the backend config,
 # such as the resoltuion or number of frames

assets/config/default.yaml

@@ -12,6 +12,11 @@ ui:
   # start playback as soon as we have 1 video over 4 (25%)
   minimum_buffer_percent_to_start_playback: 25
 
+simulation:
+  # how often the description should evolve (in seconds)
+  # setting to 0 disables description evolution
+  sim_loop_frequency_in_sec: 8
+
 video:
   # default negative prompt to filter harmful content
   default_negative_prompt: "pixelated, deformed, distorted, disfigured, blurry, text, watermark, low quality, gore, sex, blood, nudity, nude, porn, erotic"

build/web/.last_build_id

@@ -1 +1 @@
-b2c2b70851545cb6ac33ffe1f33d36c9
+00a0a70792b04bcd7ca08cb0a04b422c

build/web/assets/assets/config/aitube.yaml

@@ -1,6 +1,6 @@
 ui:
   product_name: "#aitube2"
-  showChatInVideoView: true
+  showChatInVideoView: false
 
 render_queue:
   # how many clips should be stored in advance
@@ -12,6 +12,11 @@ render_queue:
   # start playback as soon as we have 1 video over 3 (25%)
   minimum_buffer_percent_to_start_playback: 5
 
+simulation:
+  # how often the description should evolve (in seconds)
+  # setting to 0 disables description evolution
+  sim_loop_frequency_in_sec: 10
+
 # it's OK to use high values here,
 # because some of those values are limited by the backend config,
 # such as the resoltuion or number of frames

build/web/assets/assets/config/default.yaml

@@ -12,6 +12,11 @@ ui:
   # start playback as soon as we have 1 video over 4 (25%)
   minimum_buffer_percent_to_start_playback: 25
 
+simulation:
+  # how often the description should evolve (in seconds)
+  # setting to 0 disables description evolution
+  sim_loop_frequency_in_sec: 8
+
 video:
   # default negative prompt to filter harmful content
   default_negative_prompt: "pixelated, deformed, distorted, disfigured, blurry, text, watermark, low quality, gore, sex, blood, nudity, nude, porn, erotic"

build/web/flutter_bootstrap.js

@@ -39,6 +39,6 @@ _flutter.buildConfig = {"engineRevision":"382be0028d370607f76215a9be322e5514b263
 
 _flutter.loader.load({
   serviceWorkerSettings: {
-    serviceWorkerVersion: "2592320512"
+    serviceWorkerVersion: "2056062353"
   }
 });

build/web/flutter_service_worker.js

@@ -3,11 +3,11 @@ const MANIFEST = 'flutter-app-manifest';
 const TEMP = 'flutter-temp-cache';
 const CACHE_NAME = 'flutter-app-cache';
 
-const RESOURCES = {"flutter_bootstrap.js": "66845f0dd2b5279057cda185e2d06e57",
+const RESOURCES = {"flutter_bootstrap.js": "cf2a0ca5fb914aaa2e35d2d3ca0090af",
 "version.json": "b5eaae4fc120710a3c35125322173615",
 "index.html": "2677c99782b3eb75aae7c757748ccacb",
 "/": "2677c99782b3eb75aae7c757748ccacb",
-"main.dart.js": "1abc87c3f21933e9d8ac7095bfd2f625",
+"main.dart.js": "fdb1b459d8d6f04d1b1cc8520b92fbeb",
 "flutter.js": "83d881c1dbb6d6bcd6b42e274605b69c",
 "aitube.svg": "26140ba0d153b213b122bc6ebcc17f6c",
 "favicon.png": "c8a183c516004e648a7bac7497c89b97",
@@ -23,11 +23,11 @@ const RESOURCES = {"flutter_bootstrap.js": "66845f0dd2b5279057cda185e2d06e57",
 "assets/packages/cupertino_icons/assets/CupertinoIcons.ttf": "33b7d9392238c04c131b6ce224e13711",
 "assets/shaders/ink_sparkle.frag": "ecc85a2e95f5e9f53123dcaf8cb9b6ce",
 "assets/AssetManifest.bin": "6c597105edcadb9c676bdc998c88545a",
-"assets/fonts/MaterialIcons-Regular.otf": "a9126745a3792756bbb88c84ed40e354",
+"assets/fonts/MaterialIcons-Regular.otf": "06b86454c633cc9510ad85ddc0523a91",
 "assets/assets/config/README.md": "07a87720dd00dd1ca98c9d6884440e31",
 "assets/assets/config/custom.yaml": "e5c0b238b6f217f1215fbc813f093656",
-"assets/assets/config/aitube.yaml": "c9fa3d1bd8228652881a5409d78a5252",
-"assets/assets/config/default.yaml": "ba11c9ae686f1317a29bce114f0f9fc9",
+"assets/assets/config/aitube.yaml": "447b2def472f3181ce07560f54d374af",
+"assets/assets/config/default.yaml": "d1304586fd15839a754f53dda3dd8a44",
 "canvaskit/skwasm.js": "ea559890a088fe28b4ddf70e17e60052",
 "canvaskit/skwasm.js.symbols": "9fe690d47b904d72c7d020bd303adf16",
 "canvaskit/canvaskit.js.symbols": "27361387bc24144b46a745f1afe92b50",

build/web/index.html

@@ -156,7 +156,7 @@
   </script>
   
   <!-- Add version parameter for cache busting -->
-  <script src="flutter_bootstrap.js?v=BUILD_TIMESTAMP" async></script>
+  <script src="flutter_bootstrap.js?v=1746736495" async></script>
   
   <!-- Add cache busting script -->
   <script>

lib/config/config.dart

@@ -127,6 +127,10 @@ class Configuration {
   // Default negative prompt to avoid harmful content
   String get defaultNegativePrompt =>
       _config['video']['default_negative_prompt'] ?? 'gore, sex, blood, nudity, nude, porn, erotic, worst quality, deformed, distorted, disfigured, blurry, text, watermark';
+      
+  // Simulation settings
+  int get simLoopFrequencyInSec =>
+      _config['simulation']?['sim_loop_frequency_in_sec'] ?? 0;
 
   // Computed properties
 

lib/models/video_result.dart

@@ -15,6 +15,10 @@ class VideoResult {
   final bool useFixedSeed;
   final int seed;
 
+  // Evolved description fields for dynamic narrative
+  final String evolvedDescription;
+  final String condensedHistory;
+
   final int views;
   final String createdAt;
 
@@ -28,6 +32,8 @@ class VideoResult {
     this.isLatent = true,
     this.useFixedSeed = false,
     this.seed = 0,
+    this.evolvedDescription = '',
+    this.condensedHistory = '',
     this.views = 0,
     String? createdAt,
   }) : id = id ?? const Uuid().v4(),
@@ -44,6 +50,8 @@ class VideoResult {
       isLatent: json['isLatent'] as bool? ?? true,
       useFixedSeed: json['useFixedSeed'] as bool? ?? false,
       seed: json['seed'] as int? ?? 0,
+      evolvedDescription: json['evolvedDescription'] as String? ?? '',
+      condensedHistory: json['condensedHistory'] as String? ?? '',
       views: json['views'] as int? ?? 0,
       createdAt: json['createdAt'] as String?,
     );
@@ -59,6 +67,8 @@ class VideoResult {
     'isLatent': isLatent,
     'useFixedSeed': useFixedSeed,
     'seed': seed,
+    'evolvedDescription': evolvedDescription,
+    'condensedHistory': condensedHistory,
     'views': views,
     'createdAt': createdAt,
   };
@@ -74,6 +84,8 @@ class VideoResult {
     bool? isLatent,
     bool? useFixedSeed,
     int? seed,
+    String? evolvedDescription,
+    String? condensedHistory,
     int? views,
     String? createdAt,
   }) {
@@ -87,6 +99,8 @@ class VideoResult {
       isLatent: isLatent ?? this.isLatent,
       useFixedSeed: useFixedSeed ?? this.useFixedSeed,
       seed: seed ?? this.seed,
+      evolvedDescription: evolvedDescription ?? this.evolvedDescription,
+      condensedHistory: condensedHistory ?? this.condensedHistory,
       views: views ?? this.views,
       createdAt: createdAt ?? this.createdAt,
     );

lib/screens/home_screen.dart

@@ -449,9 +449,12 @@ class _HomeScreenState extends State<HomeScreen> {
                 ? Center(
                     child: Text(
                       _isSearching
-                          ? 'Generating videos...'
-                          : 'Start by typing a description of the video you want to generate',
-                      style: const TextStyle(color: AiTubeColors.onSurfaceVariant),
+                          ? 'Hallucinating search results using AI...'
+                          : 'Results are generated on demand, videos rendered on the fly.',
+                      style: const TextStyle(
+                        color: AiTubeColors.onSurfaceVariant,
+                        fontSize: 20
+                      ),
                       textAlign: TextAlign.center,
                     ),
                   )

lib/screens/video_screen.dart

@@ -436,7 +436,7 @@ class _VideoScreenState extends State<VideoScreen> {
                       if (Configuration.instance.showChatInVideoView) ...[
                         const SizedBox(width: 16),
                         Padding(
-                          padding: const EdgeInsets.only(right: 16),
+                          padding: const EdgeInsets.only(left: 0, top: 16, right: 16, bottom: 4),
                           child: ChatWidget(videoId: widget.video.id),
                         ),
                       ],
@@ -451,7 +451,9 @@ class _VideoScreenState extends State<VideoScreen> {
                         const SizedBox(height: 16),
                         Expanded(
                           child: Padding(
-                            padding: const EdgeInsets.symmetric(horizontal: 16),
+                            padding: const EdgeInsets.symmetric(
+                              horizontal: 16
+                            ),
                             child: ChatWidget(
                               videoId: widget.video.id,
                               isCompact: true,
@@ -505,13 +507,14 @@ class _VideoScreenState extends State<VideoScreen> {
               style: Theme.of(context).textTheme.headlineSmall?.copyWith(
                 color: AiTubeColors.onBackground,
                 fontWeight: FontWeight.bold,
+                fontSize: 18
               ),
             ),
           ),
           IconButton(
             icon: const Icon(Icons.share, color: AiTubeColors.primary),
             onPressed: _shareVideo,
-            tooltip: 'Share prompt',
+            tooltip: 'Share this creation',
           ),
         ],
       ),
@@ -523,29 +526,7 @@ class _VideoScreenState extends State<VideoScreen> {
         Column(
           crossAxisAlignment: CrossAxisAlignment.start,
           children: [
-            // Tags
-            if (_videoData.tags.isNotEmpty) ...[
-              Wrap(
-                spacing: 8,
-                runSpacing: 8,
-                children: _videoData.tags.map((tag) => Chip(
-                  label: Text(tag),
-                  backgroundColor: AiTubeColors.surface,
-                  labelStyle: const TextStyle(color: AiTubeColors.onSurface),
-                )).toList(),
-              ),
-              const SizedBox(height: 16),
-            ],
-
             // Description Section
-            const Text(
-              'Description',
-              style: TextStyle(
-                color: AiTubeColors.onBackground,
-                fontWeight: FontWeight.bold,
-                fontSize: 18,
-              ),
-            ),
             const SizedBox(height: 8),
             Text(
               _videoData.description,
@@ -554,6 +535,7 @@ class _VideoScreenState extends State<VideoScreen> {
                 height: 1.5,
               ),
             ),
+            const SizedBox(height: 8),
           ],
         ),
       ],

lib/services/chat_service.dart

@@ -116,6 +116,9 @@ class ChatService {
         color: _userColor,
       );
 
+      // Add message to the local stream before sending to avoid duplicates
+      _chatController.add(message);
+      
       debugPrint('ChatService: Sending message via WebSocket...');
       await _websocketService.sendChatMessage(message);
       debugPrint('ChatService: Message sent successfully');
@@ -146,11 +149,24 @@ class ChatService {
     }
   }
 
-  void dispose() {
+  // This method is only for application shutdown
+  // Individual widgets should use leaveRoom instead
+  Future<void> dispose() async {
+    // Properly leave current room first if connected
     if (_currentRoomId != null) {
-      leaveRoom(_currentRoomId!);
+      try {
+        await leaveRoom(_currentRoomId!);
+      } catch (e) {
+        debugPrint('ChatService: Error leaving room during disposal: $e');
+      }
     }
-    _chatController.close();
+    
+    // Only close the controller if we're truly shutting down
+    if (!_chatController.isClosed) {
+      _chatController.close();
+    }
+    
     _isInitialized = false;
+    debugPrint('ChatService: Successfully disposed');
   }
 }

lib/services/clip_queue/clip_queue_manager.dart

@@ -1,12 +1,15 @@
 // lib/services/clip_queue/clip_queue_manager.dart
 
 import 'dart:async';
+import 'dart:math';
 import 'package:aitube2/config/config.dart';
 import 'package:flutter/foundation.dart';
 import 'package:collection/collection.dart';
 import '../../models/video_result.dart';
 import '../../models/video_orientation.dart';
+import '../../models/chat_message.dart';
 import '../websocket_api_service.dart';
+import '../chat_service.dart';
 import '../../utils/seed.dart';
 import 'clip_states.dart';
 import 'video_clip.dart';
@@ -16,7 +19,7 @@ import 'clip_generation_handler.dart';
 /// Manages a queue of video clips for generation and playback
 class ClipQueueManager {
   /// The video for which clips are being generated
-  final VideoResult video;
+  VideoResult video;
   
   /// WebSocket service for API communication
   final WebSocketApiService _websocketService;
@@ -36,9 +39,18 @@ class ClipQueueManager {
   /// Timer for checking the buffer state
   Timer? _bufferCheckTimer;
   
+  /// Timer for evolving the description
+  Timer? _descriptionEvolutionTimer;
+  
+  /// Last time the description was evolved
+  DateTime _lastDescriptionEvolutionTime = DateTime.now();
+  
   /// Whether the manager is disposed
   bool _isDisposed = false;
   
+  /// Whether the simulation is paused (controlled by video playback)
+  bool _isSimulationPaused = false;
+  
   /// Stats logger
   final QueueStatsLogger _logger = QueueStatsLogger();
   
@@ -47,6 +59,12 @@ class ClipQueueManager {
   
   /// ID of the video being managed
   final String videoId;
+  
+  /// Evolution counter for tracking how many times we've evolved the description
+  int _evolutionCounter = 0;
+  
+  /// Recent chat messages to include in description evolution
+  final List<ChatMessage> _recentChatMessages = [];
 
   /// Constructor
   ClipQueueManager({
@@ -61,6 +79,30 @@ class ClipQueueManager {
       activeGenerations: _activeGenerations,
       onQueueUpdated: onQueueUpdated,
     );
+    
+    // Start listening to chat messages
+    final chatService = ChatService();
+    chatService.initialize().then((_) {
+      chatService.joinRoom(videoId).then((_) {
+        chatService.chatStream.listen(_addChatMessage);
+      }).catchError((e) {
+        debugPrint('ClipQueueManager: Error joining chat room: $e');
+      });
+    }).catchError((e) {
+      debugPrint('ClipQueueManager: Error initializing chat service: $e');
+    });
+  }
+  
+  /// Add a chat message to the recent messages list
+  void _addChatMessage(ChatMessage message) {
+    if (message.videoId == videoId) {
+      _recentChatMessages.add(message);
+      // Keep only the 10 most recent messages
+      if (_recentChatMessages.length > 10) {
+        _recentChatMessages.removeAt(0);
+      }
+      ClipQueueConstants.logEvent('Added chat message: ${message.content.substring(0, min(20, message.content.length))}...');
+    }
   }
 
   /// Whether a new generation can be started
@@ -107,6 +149,10 @@ class ClipQueueManager {
     );
     _clipBuffer.clear();
     
+    // Reset evolution counter and last evolution time
+    _evolutionCounter = 0;
+    _lastDescriptionEvolutionTime = DateTime.now();
+    
     // Set initial orientation
     _currentOrientation = orientation ?? getOrientationFromDimensions(
       Configuration.instance.originalClipWidth, 
@@ -130,6 +176,7 @@ class ClipQueueManager {
       if (_isDisposed) return;
 
       _startBufferCheck();
+      _startDescriptionEvolution();
       await _fillBuffer();
       ClipQueueConstants.logEvent('Initialization complete. Buffer size: ${_clipBuffer.length}');
       printQueueState();
@@ -160,6 +207,127 @@ class ClipQueueManager {
     );
     ClipQueueConstants.logEvent('Started buffer check timer');
   }
+  
+  /// Start the simulation timer
+  void _startDescriptionEvolution() {
+    // Cancel any existing timer
+    _descriptionEvolutionTimer?.cancel();
+    
+    // Only start if simulation frequency is greater than 0
+    if (Configuration.instance.simLoopFrequencyInSec <= 0) {
+      ClipQueueConstants.logEvent('Simulation disabled (frequency is 0)');
+      return;
+    }
+    
+    // Adaptive check interval - less frequent checks to reduce overhead
+    final checkInterval = max(3, Configuration.instance.simLoopFrequencyInSec ~/ 3);
+    
+    ClipQueueConstants.logEvent('Starting simulation with check interval of $checkInterval seconds');
+    
+    // Check periodically if it's time to simulate the video
+    _descriptionEvolutionTimer = Timer.periodic(
+      Duration(seconds: checkInterval),
+      (timer) async {
+        if (_isDisposed) return;
+        
+        // Skip if simulation is paused (due to video playback being paused)
+        if (_isSimulationPaused) {
+          ClipQueueConstants.logEvent('Skipping simulation because it is paused');
+          return;
+        }
+        
+        // Check if we're currently generating a video - if so, delay simulation
+        final isGenerating = _activeGenerations.isNotEmpty;
+        if (isGenerating) {
+          ClipQueueConstants.logEvent('Delaying simulation due to active generations');
+          return;
+        }
+        
+        // Calculate time since last simulation
+        final now = DateTime.now();
+        final duration = now.difference(_lastDescriptionEvolutionTime);
+        
+        // If we've waited long enough, simulate the video
+        if (duration.inSeconds >= Configuration.instance.simLoopFrequencyInSec) {
+          ClipQueueConstants.logEvent('Triggering simulation after ${duration.inSeconds} seconds');
+          await _evolveDescription();
+          _lastDescriptionEvolutionTime = now;
+        }
+      },
+    );
+    ClipQueueConstants.logEvent('Started simulation timer');
+  }
+  
+  /// Simulate the video by evolving the description using the LLM
+  Future<void> _evolveDescription() async {
+    if (!_websocketService.isConnected) {
+      ClipQueueConstants.logEvent('Cannot simulate video: websocket not connected');
+      return;
+    }
+    
+    int retryCount = 0;
+    const maxRetries = 2;
+    
+    // Function to get chat message string
+    String getChatMessagesString() {
+      if (_recentChatMessages.isEmpty) return '';
+      
+      return _recentChatMessages.map((msg) => 
+        "${msg.username}: ${msg.content}"
+      ).join("\n");
+    }
+    
+    while (retryCount <= maxRetries) {
+      try {
+        // Format recent chat messages as a string for the simulation prompt
+        String chatMessagesString = getChatMessagesString();
+        if (chatMessagesString.isNotEmpty) {
+          ClipQueueConstants.logEvent('Including ${_recentChatMessages.length} chat messages in simulation');
+        }
+        
+        // Use the WebSocketService to simulate the video
+        final result = await _websocketService.simulate(
+          videoId: video.id,
+          originalTitle: video.title,
+          originalDescription: video.description,
+          currentDescription: video.evolvedDescription.isEmpty ? video.description : video.evolvedDescription,
+          condensedHistory: video.condensedHistory,
+          evolutionCount: _evolutionCounter,
+          chatMessages: chatMessagesString,
+        );
+        
+        // Update the video with the evolved description
+        video = video.copyWith(
+          evolvedDescription: result['evolved_description'],
+          condensedHistory: result['condensed_history'],
+        );
+        
+        _evolutionCounter++;
+        ClipQueueConstants.logEvent('Video simulated (iteration $_evolutionCounter)');
+        onQueueUpdated?.call();
+        
+        // Success, exit retry loop
+        break;
+      } catch (e) {
+        retryCount++;
+        ClipQueueConstants.logEvent('Error simulating video attempt $retryCount/$maxRetries: $e');
+        
+        if (retryCount <= maxRetries) {
+          // Wait before retrying with exponential backoff
+          final delay = Duration(seconds: 1 << retryCount);
+          ClipQueueConstants.logEvent('Retrying simulation in ${delay.inSeconds} seconds...');
+          await Future.delayed(delay);
+        } else {
+          ClipQueueConstants.logEvent('Failed to simulate video after $maxRetries attempts');
+          
+          // If we've been successful before but failed now, we can continue using the last evolved description
+          if (_evolutionCounter > 0) {
+            ClipQueueConstants.logEvent('Continuing with previous description');
+          }
+        }
+      }
+    }
+  }
 
   /// Mark a specific clip as played
   void markClipAsPlayed(String clipId) {
@@ -193,8 +361,16 @@ class ClipQueueManager {
 
     // First ensure we have the correct buffer size
     while (_clipBuffer.length < Configuration.instance.renderQueueBufferSize) {
+      // Determine which description to use for the prompt
+      String descriptionToUse = video.description;
+      
+      // If we have an evolved description, use that instead
+      if (video.evolvedDescription.isNotEmpty) {
+        descriptionToUse = video.evolvedDescription;
+      }
+      
       final newClip = VideoClip(
-        prompt: "${video.title}\n${video.description}",
+        prompt: "${video.title}\n${descriptionToUse}",
         seed: video.useFixedSeed && video.seed > 0 ? video.seed : generateSeed(),
         orientation: _currentOrientation,
       );
@@ -316,9 +492,18 @@ class ClipQueueManager {
       _clipBuffer.remove(clip);
       _clipHistory.add(clip);
       
+      // Determine which description to use for the prompt
+      String descriptionToUse = video.description;
+      
+      // If we have an evolved description, use that instead
+      if (video.evolvedDescription.isNotEmpty) {
+        descriptionToUse = video.evolvedDescription;
+        ClipQueueConstants.logEvent('Using evolved description for new clip (evolution #$_evolutionCounter)');
+      }
+      
       // Add a new pending clip with current orientation
       final newClip = VideoClip(
-        prompt: "${video.title}\n${video.description}",
+        prompt: "${video.title}\n${descriptionToUse}",
         seed: video.useFixedSeed && video.seed > 0 ? video.seed : generateSeed(),
         orientation: _currentOrientation,
       );
@@ -409,6 +594,24 @@ class ClipQueueManager {
     // Notify listeners
     onQueueUpdated?.call();
   }
+  
+  /// Set the simulation pause state based on video playback
+  void setSimulationPaused(bool isPaused) {
+    if (_isSimulationPaused == isPaused) return;
+    
+    _isSimulationPaused = isPaused;
+    ClipQueueConstants.logEvent(
+      isPaused 
+        ? 'Simulation paused (video playback paused)' 
+        : 'Simulation resumed (video playback resumed)'
+    );
+    
+    // If we're resuming after a pause, update the last evolution time
+    // to avoid immediate evolution after resuming
+    if (!isPaused) {
+      _lastDescriptionEvolutionTime = DateTime.now();
+    }
+  }
 
   /// Print the current state of the queue
   void printQueueState() {
@@ -436,6 +639,7 @@ class ClipQueueManager {
 
     // Cancel all timers first
     _bufferCheckTimer?.cancel();
+    _descriptionEvolutionTimer?.cancel();
     
     // Complete any pending generation completers
     for (var clip in _clipBuffer) {

lib/services/clip_queue/queue_stats_logger.dart

@@ -52,10 +52,7 @@ class QueueStatsLogger {
         !_areStatesEqual(_lastLoggedState!, currentState) ||
         _shouldLogDueToTimeout()) {
       
-      debugPrint('\n=== Queue State Change [$trigger] ===');
-      debugPrint('Ready: ${currentState['readyClips']}');
-      debugPrint('Playing: ${currentState['playingClips']}');
-      debugPrint('Generating: ${currentState['generatingClips']}');
+      // debugPrint('Queue State Change [$trigger] => Ready: ${currentState['readyClips']}, Playing: ${currentState['playingClips']}, Generating: ${currentState['generatingClips']}');
       
       /*
       debugPrint('Pending: ${currentState['pendingClips']}');

lib/services/websocket_api_service.dart

@@ -859,6 +859,11 @@ class WebSocketApiService {
   }
 
   Future<bool> sendChatMessage(ChatMessage message) async {
+    if (_disposed) {
+      debugPrint('WebSocketApiService: Cannot send message, service is disposed');
+      throw Exception('WebSocketApiService is disposed');
+    }
+    
     if (!_initialized) {
       debugPrint('WebSocketApiService: Initializing before sending message...');
       await initialize();
@@ -867,6 +872,12 @@ class WebSocketApiService {
     try {
       debugPrint('WebSocketApiService: Sending chat message...');
       
+      // Check if the chatController is still open to avoid "Cannot add events after calling close" error
+      if (_chatController.isClosed) {
+        debugPrint('WebSocketApiService: Chat controller is closed, cannot process messages');
+        throw Exception('Chat controller is closed');
+      }
+      
       final response = await _sendRequest(
         WebSocketRequest(
           action: 'chat_message',
@@ -897,7 +908,7 @@ class WebSocketApiService {
       final action = data['action'] as String?;
       final requestId = data['requestId'] as String?;
 
-      // debugPrint('WebSocketApiService: Received message for action: $action, requestId: $requestId');
+      debugPrint('WebSocketApiService: Received message for action: $action, requestId: $requestId');
       
       // Update user role if present in response (from heartbeat or get_user_role)
       if (data['user_role'] != null) {
@@ -912,10 +923,8 @@ class WebSocketApiService {
       if (requestId != null && _pendingRequests.containsKey(requestId)) {
         if (action == 'chat_message') {
           debugPrint('WebSocketApiService: Processing chat message response');
-          // Extract the message data for chat messages
-          if (data['success'] == true && data['message'] != null) {
-            _handleChatMessage(data['message'] as Map<String, dynamic>);
-          }
+          // Don't process the message data from our own request since we already added it locally
+          // We only need to complete the request to signal success/failure
           _pendingRequests[requestId]!.complete(data);
         } else if (action == 'join_chat') {
           debugPrint('WebSocketApiService: Processing join chat response');
@@ -948,6 +957,11 @@ class WebSocketApiService {
   }
 
   void _handleChatMessage(Map<String, dynamic> data) {
+    if (_disposed) {
+      debugPrint('WebSocketApiService: Skipping message handling, service is disposed');
+      return;
+    }
+    
     try {
       // Log the exact data we're trying to parse
       debugPrint('Parsing chat message data: ${json.encode(data)}');
@@ -964,7 +978,13 @@ class WebSocketApiService {
       
       final message = ChatMessage.fromJson(data);
       debugPrint('Successfully parsed message: ${message.toString()}');
-      _chatController.add(message);
+      
+      // Only add to stream if it's still open
+      if (!_chatController.isClosed) {
+        _chatController.add(message);
+      } else {
+        debugPrint('WebSocketApiService: Chat controller is closed, cannot add message');
+      }
     } catch (e, stackTrace) {
       debugPrint('Error handling chat message: $e');
       debugPrint('Stack trace: $stackTrace');
@@ -974,11 +994,22 @@ class WebSocketApiService {
 
 
   void _handleChatHistory(Map<String, dynamic> data) {
+    if (_disposed) {
+      debugPrint('WebSocketApiService: Skipping chat history handling, service is disposed');
+      return;
+    }
+    
     try {
       if (data['messages'] == null) {
         debugPrint('No messages found in chat history');
         return;
       }
+      
+      // Check if chat controller is still open
+      if (_chatController.isClosed) {
+        debugPrint('WebSocketApiService: Chat controller is closed, cannot process chat history');
+        return;
+      }
 
       final messages = (data['messages'] as List).map((m) {
         try {
@@ -992,8 +1023,13 @@ class WebSocketApiService {
       
       debugPrint('Processing ${messages.length} historical messages');
       
-      for (final message in messages) {
-        _chatController.add(message);
+      // Check again before adding messages in case it was closed during processing
+      if (!_chatController.isClosed) {
+        for (final message in messages) {
+          _chatController.add(message);
+        }
+      } else {
+        debugPrint('WebSocketApiService: Chat controller was closed during processing');
       }
     } catch (e, stackTrace) {
       debugPrint('Error handling chat history: $e');
@@ -1075,7 +1111,7 @@ class WebSocketApiService {
 
     try {
       final requestData = request.toJson();
-      // debugPrint('WebSocketApiService: Sending request ${request.requestId} (${request.action}): ${json.encode(requestData)}');
+      debugPrint('WebSocketApiService: Sending request ${request.requestId} (${request.action}): ${json.encode(requestData)}');
       _channel!.sink.add(json.encode(requestData));
       
       final response = await completer.future.timeout(
@@ -1191,6 +1227,59 @@ class WebSocketApiService {
     return response['caption'] as String;
   }
 
+  /// Simulate a video by evolving its description to create a dynamic narrative
+  Future<Map<String, String>> simulate({
+    required String videoId,
+    required String originalTitle,
+    required String originalDescription,
+    required String currentDescription,
+    required String condensedHistory,
+    int evolutionCount = 0,
+    String chatMessages = '',
+  }) async {
+    // Skip if the API is not connected
+    if (!isConnected) {
+      debugPrint('WebSocketApiService: Cannot simulate video, not connected');
+      return {
+        'evolved_description': currentDescription,
+        'condensed_history': condensedHistory
+      };
+    }
+
+    try {
+      final response = await _sendRequest(
+        WebSocketRequest(
+          action: 'simulate',
+          params: {
+            'video_id': videoId,
+            'original_title': originalTitle,
+            'original_description': originalDescription,
+            'current_description': currentDescription,
+            'condensed_history': condensedHistory,
+            'evolution_count': evolutionCount,
+            'chat_messages': chatMessages,
+          },
+        ),
+        timeout: const Duration(seconds: 60),
+      );
+
+      if (!response['success']) {
+        throw Exception(response['error'] ?? 'Simulation failed');
+      }
+
+      return {
+        'evolved_description': response['evolved_description'] as String? ?? currentDescription,
+        'condensed_history': response['condensed_history'] as String? ?? condensedHistory
+      };
+    } catch (e) {
+      debugPrint('WebSocketApiService: Error simulating video: $e');
+      return {
+        'evolved_description': currentDescription,
+        'condensed_history': condensedHistory
+      };
+    }
+  }
+
 
   // Additional utility methods
   Future<void> waitForConnection() async {

lib/theme/colors.dart

@@ -2,11 +2,12 @@
 import 'package:flutter/material.dart';
 
 class AiTubeColors {
-  static const background = Color(0xFF0F0F0F);
-  static const surface = Color(0xFF242424);
-  static const surfaceVariant = Color(0xFF2D2D2D);
-  static const primary = Colors.red;
-  static const onBackground = Colors.white;
-  static const onSurface = Colors.white;
+  static const transparent = Color(0x00000000);
+  static const background = Color(0xFF171717);
+  static const surface = Color(0xFF272727);
+  static const surfaceVariant = Color(0xFF2B2B2B);
+  static const primary = Color.fromARGB(236, 214, 225, 4);
+  static const onBackground = Color.fromARGB(239, 255, 255, 255);
+  static const onSurface = Color.fromARGB(226, 255, 255, 255);
   static const onSurfaceVariant = Colors.white70;
 }

lib/widgets/ai_content_disclaimer.dart

@@ -75,7 +75,7 @@ class AiContentDisclaimer extends StatelessWidget {
         child: LayoutBuilder(
           builder: (context, constraints) {
             // Scale text based on container width
-            final baseSize = constraints.maxWidth / 40;
+            final baseSize = constraints.maxWidth / 35;
             final smallTextSize = baseSize * 0.7;
             final mediumTextSize = baseSize;
             final largeTextSize = baseSize * 1.1;
@@ -91,7 +91,7 @@ class AiContentDisclaimer extends StatelessWidget {
                     textBaseline: TextBaseline.alphabetic,
                     children: [
                       Text(
-                        'The following ',
+                        'THE FOLLOWING ',
                         style: GoogleFonts.arimo(
                           fontSize: smallTextSize,
                           color: Colors.white,
@@ -108,7 +108,7 @@ class AiContentDisclaimer extends StatelessWidget {
                         ),
                       ),
                       Text(
-                        'footage',
+                        'CONTENT',
                         style: GoogleFonts.arimo(
                           fontSize: mediumTextSize,
                           color: Colors.white,
@@ -125,7 +125,7 @@ class AiContentDisclaimer extends StatelessWidget {
                         ),
                       ),
                       Text(
-                        isInteractive ? ' will be ' : ' has been ',
+                        isInteractive ? ' WILL BE ' : ' HAS BEEN ',
                         style: GoogleFonts.arimo(
                           fontSize: smallTextSize,
                           color: Colors.white,
@@ -142,7 +142,7 @@ class AiContentDisclaimer extends StatelessWidget {
                         ),
                       ),
                       Text(
-                        'synthesized',
+                        'SYNTHESIZED',
                         style: GoogleFonts.arimo(
                           fontSize: mediumTextSize,
                           color: Colors.white,
@@ -158,16 +158,8 @@ class AiContentDisclaimer extends StatelessWidget {
                           ],
                         ),
                       ),
-                    ],
-                  ),
-                  const SizedBox(height: 18),
-                  Row(
-                    mainAxisAlignment: MainAxisAlignment.center,
-                    crossAxisAlignment: CrossAxisAlignment.baseline,
-                    textBaseline: TextBaseline.alphabetic,
-                    children: [
                       Text(
-                        'on the fly using a',
+                        ' BY A',
                         style: GoogleFonts.arimo(
                           fontSize: smallTextSize,
                           color: Colors.white,
@@ -185,9 +177,10 @@ class AiContentDisclaimer extends StatelessWidget {
                       ),
                     ],
                   ),
+                  
                   const SizedBox(height: 18),
                   Text(
-                    'distilled video model',
+                    'DISTILLED AI VIDEO MODEL',
                     style: GoogleFonts.arimo(
                       fontSize: largeTextSize,
                       color: Colors.white,
@@ -204,23 +197,102 @@ class AiContentDisclaimer extends StatelessWidget {
                     ),
                   ),
                   const SizedBox(height: 18),
-                  Text(
-                    'and may contains visual glitches or hallucinations.',
-                    style: GoogleFonts.arimo(
-                      fontSize: smallTextSize,
-                      color: Colors.white,
-                      fontWeight: FontWeight.w500,
-                      letterSpacing: 1.2,
-                      height: 1.0,
-                      shadows: const [
-                        Shadow(
-                          offset: Offset(0, 2),
-                          blurRadius: 3.0,
-                          color: Color.fromRGBO(0, 0, 0, 0.3),
+                  Row(
+                    mainAxisAlignment: MainAxisAlignment.center,
+                    crossAxisAlignment: CrossAxisAlignment.baseline,
+                    textBaseline: TextBaseline.alphabetic,
+                    children: [
+                      Text(
+                        'AND MAY CONTAIN',
+                        style: GoogleFonts.arimo(
+                          fontSize: smallTextSize,
+                          color: Colors.white,
+                          fontWeight: FontWeight.w500,
+                          letterSpacing: 1.2,
+                          height: 1.0,
+                          shadows: const [
+                            Shadow(
+                              offset: Offset(0, 2),
+                              blurRadius: 3.0,
+                              color: Color.fromRGBO(0, 0, 0, 0.3),
+                            ),
+                          ],
                         ),
-                      ],
-                    ),
-                    textAlign: TextAlign.center,
+                        textAlign: TextAlign.center,
+                      ),
+                      Text(
+                        ' VISUAL GLITCHES',
+                        style: GoogleFonts.arimo(
+                          fontSize: mediumTextSize,
+                          color: Colors.white,
+                          fontWeight: FontWeight.w700,
+                          letterSpacing: 1.2,
+                          height: 1.0,
+                          shadows: const [
+                            Shadow(
+                              offset: Offset(0, 2),
+                              blurRadius: 3.0,
+                              color: Color.fromRGBO(0, 0, 0, 0.3),
+                            ),
+                          ],
+                        ),
+                        textAlign: TextAlign.center,
+                      ),
+                      Text(
+                        ' OR',
+                        style: GoogleFonts.arimo(
+                          fontSize: smallTextSize,
+                          color: Colors.white,
+                          fontWeight: FontWeight.w500,
+                          letterSpacing: 1.2,
+                          height: 1.0,
+                          shadows: const [
+                            Shadow(
+                              offset: Offset(0, 2),
+                              blurRadius: 3.0,
+                              color: Color.fromRGBO(0, 0, 0, 0.3),
+                            ),
+                          ],
+                        ),
+                        textAlign: TextAlign.center,
+                      ),
+                      Text(
+                        ' HALLUCINATIONS',
+                        style: GoogleFonts.arimo(
+                          fontSize: mediumTextSize,
+                          color: Colors.white,
+                          fontWeight: FontWeight.w700,
+                          letterSpacing: 1.2,
+                          height: 1.0,
+                          shadows: const [
+                            Shadow(
+                              offset: Offset(0, 2),
+                              blurRadius: 3.0,
+                              color: Color.fromRGBO(0, 0, 0, 0.3),
+                            ),
+                          ],
+                        ),
+                        textAlign: TextAlign.center,
+                      ),
+                       Text(
+                        '.',
+                        style: GoogleFonts.arimo(
+                          fontSize: smallTextSize,
+                          color: Colors.white,
+                          fontWeight: FontWeight.w500,
+                          letterSpacing: 1.2,
+                          height: 1.0,
+                          shadows: const [
+                            Shadow(
+                              offset: Offset(0, 2),
+                              blurRadius: 3.0,
+                              color: Color.fromRGBO(0, 0, 0, 0.3),
+                            ),
+                          ],
+                        ),
+                        textAlign: TextAlign.center,
+                      ),
+                    ]
                   ),
                 ],
               ),

lib/widgets/chat_widget.dart

@@ -25,6 +25,8 @@ class _ChatWidgetState extends State<ChatWidget> {
   final _messageController = TextEditingController();
   final _scrollController = ScrollController();
   final _messages = <ChatMessage>[];
+  // Track message IDs to prevent duplicates
+  final _messageIds = <String>{};
   bool _isLoading = true;
   bool _isSending = false;
   String? _error;
@@ -120,11 +122,19 @@ class _ChatWidgetState extends State<ChatWidget> {
   void _onNewMessage(ChatMessage message) {
     if (!mounted) return;
     
+    // Check if we already have this message (prevent duplicates)
+    if (_messageIds.contains(message.id)) {
+      debugPrint('ChatWidget: Skipping duplicate message with ID: ${message.id}');
+      return;
+    }
+    
     setState(() {
       _messages.add(message);
+      _messageIds.add(message.id);
       // Keep only last 100 messages
       if (_messages.length > 100) {
-        _messages.removeAt(0);
+        final removedMessage = _messages.removeAt(0);
+        _messageIds.remove(removedMessage.id);
       }
     });
 
@@ -179,7 +189,7 @@ class _ChatWidgetState extends State<ChatWidget> {
             child: Text(
               message.username.substring(0, 1).toUpperCase(),
               style: const TextStyle(
-                color: Colors.white,
+                color: Colors.black,
                 fontWeight: FontWeight.bold,
               ),
             ),
@@ -225,7 +235,7 @@ class _ChatWidgetState extends State<ChatWidget> {
     return Container(
       padding: const EdgeInsets.all(8),
       decoration: const BoxDecoration(
-        color: AiTubeColors.surface,
+        color: AiTubeColors.transparent,
         border: Border(
           top: BorderSide(
             color: AiTubeColors.surfaceVariant,
@@ -239,18 +249,48 @@ class _ChatWidgetState extends State<ChatWidget> {
             child: TextField(
               controller: _messageController,
               style: const TextStyle(color: AiTubeColors.onSurface),
-              maxLength: 256,
+              maxLength: 255,
+              maxLines: 1,
+              onChanged: (value) {
+                // Enforce the character limit by trimming excess characters
+                if (value.length > 255) {
+                  _messageController.text = value.substring(0, 255);
+                  _messageController.selection = TextSelection.fromPosition(
+                    const TextPosition(offset: 255),
+                  );
+                }
+              },
               decoration: InputDecoration(
-                hintText: 'Type a message...',
-                hintStyle: const TextStyle(color: AiTubeColors.onSurfaceVariant),
+                hintText: 'Chat with this aituber..',
+                hintStyle: const TextStyle(color: AiTubeColors.onSurfaceVariant, fontSize: 16),
                 border: OutlineInputBorder(
-                  borderRadius: BorderRadius.circular(24),
+                  borderRadius: BorderRadius.circular(12),
+                  borderSide: const BorderSide(
+                    color: Color(0x20FFFFFF),
+                    width: 1,
+                  ),
+                ),
+                enabledBorder: OutlineInputBorder(
+                  borderRadius: BorderRadius.circular(12),
+                  borderSide: const BorderSide(
+                    color: Color(0x20FFFFFF),
+                    width: 1,
+                  ),
+                ),
+                focusedBorder: OutlineInputBorder(
+                  borderRadius: BorderRadius.circular(12),
+                  borderSide: const BorderSide(
+                    color: AiTubeColors.primary,
+                    width: 1,
+                  ),
                 ),
                 contentPadding: const EdgeInsets.symmetric(
-                  horizontal: 16,
-                  vertical: 8,
+                  horizontal: 8,
+                  vertical: 4,
                 ),
                 counterText: '',
+                filled: true,
+                fillColor: const Color(0x10000000),
               ),
             ),
           ),
@@ -262,7 +302,7 @@ class _ChatWidgetState extends State<ChatWidget> {
                   height: 20,
                   child: CircularProgressIndicator(strokeWidth: 2),
                 )
-              : const Icon(Icons.send),
+              : const Icon(Icons.reply),
             color: AiTubeColors.primary,
             onPressed: _isSending ? null : _sendMessage,
           ),
@@ -377,7 +417,7 @@ class _ChatWidgetState extends State<ChatWidget> {
                 Icon(Icons.chat, color: AiTubeColors.onBackground),
                 SizedBox(width: 8),
                 Text(
-                  'Live Chat',
+                  'Simulation log',
                   style: TextStyle(
                     color: AiTubeColors.onBackground,
                     fontSize: 16,
@@ -402,11 +442,10 @@ class _ChatWidgetState extends State<ChatWidget> {
     _messageController.dispose();
     _scrollController.dispose();
     
-    // Ensure chat room is left before disposal
-    _chatService.leaveRoom(widget.videoId).then((_) {
-      _chatService.dispose();
-    }).catchError((error) {
-      debugPrint('ChatWidget: Error during disposal: $error');
+    // Just leave the chat room, but don't dispose the ChatService
+    // since it's a singleton that may be used by other widgets
+    _chatService.leaveRoom(widget.videoId).catchError((error) {
+      debugPrint('ChatWidget: Error leaving chat room during disposal: $error');
     });
     
     super.dispose();

lib/widgets/search_box.dart

@@ -67,7 +67,7 @@ class _SearchBoxState extends State<SearchBox> {
           FocusScope.of(context).unfocus();
         },
         decoration: InputDecoration(
-          hintText: 'Describe a video you want to generate...',
+          hintText: 'Imagine anything eg "F1 race with farming trucks"',
           hintStyle: const TextStyle(color: AiTubeColors.onSurfaceVariant),
           filled: true,
           fillColor: AiTubeColors.surface,

lib/widgets/video_player/video_player_widget.dart

@@ -126,14 +126,28 @@ class _VideoPlayerWidgetState extends State<VideoPlayerWidget> with WidgetsBindi
   void _pauseVideo() {
     if (_playbackController.isPlaying) {
       _wasPlayingBeforeBackground = true;
-      _togglePlayback();
+      
+      // Manually pause playback and simulation together
+      _playbackController.togglePlayback();
+      _bufferManager.queueManager.setSimulationPaused(true);
+      
+      if (!_isDisposed && mounted) {
+        setState(() {});
+      }
     }
   }
   
   void _resumeVideo() {
     if (!_playbackController.isPlaying && _wasPlayingBeforeBackground) {
       _wasPlayingBeforeBackground = false;
-      _togglePlayback();
+      
+      // Manually resume playback and simulation together
+      _playbackController.togglePlayback();
+      _bufferManager.queueManager.setSimulationPaused(false);
+      
+      if (!_isDisposed && mounted) {
+        setState(() {});
+      }
     }
   }
 
@@ -229,6 +243,9 @@ class _VideoPlayerWidgetState extends State<VideoPlayerWidget> with WidgetsBindi
     }
     
     if (!_isDisposed && mounted) {
+      // Initialize simulation pause state based on initial autoPlay setting
+      _bufferManager.queueManager.setSimulationPaused(!widget.autoPlay);
+      
       setState(() {
         _playbackController.isLoading = false;
         _playbackController.isInitialLoad = false;
@@ -238,6 +255,10 @@ class _VideoPlayerWidgetState extends State<VideoPlayerWidget> with WidgetsBindi
 
   void _togglePlayback() {
     _playbackController.togglePlayback();
+    
+    // Control the simulation based on playback state
+    _bufferManager.queueManager.setSimulationPaused(!_playbackController.isPlaying);
+    
     if (!_isDisposed && mounted) {
       setState(() {});
     }
@@ -391,6 +412,11 @@ class _VideoPlayerWidgetState extends State<VideoPlayerWidget> with WidgetsBindi
   void dispose() {
     _isDisposed = true;
     
+    // Ensure simulation is paused when widget is disposed
+    if (_bufferManager.queueManager != null) {
+      _bufferManager.queueManager.setSimulationPaused(true);
+    }
+    
     // Unregister the observer
     WidgetsBinding.instance.removeObserver(this);