Bump version: 0.5.4 → 0.5.5

- Modify load_config function to copy user_config.cfg from user_config.cfg.sample if it doesn't exist
- Update README.md to reflect new features and configuration options, including web-based settings and job queue management

.bumpversion.toml

@@ -1,5 +1,5 @@
 [tool.bumpversion]
-current_version = "0.3.14"
+current_version = "0.5.5"
 parse = "(?P<major>\\d+)\\.(?P<minor>\\d+)\\.(?P<patch>\\d+)"
 serialize = ["{major}.{minor}.{patch}"]
 replace = "{new_version}"

Dockerfile

@@ -4,7 +4,7 @@ FROM python:3.11-slim
 # Set the working directory in the container
 WORKDIR /app
 # Set version label
-ARG VERSION="0.3.14"
+ARG VERSION="0.5.5"
 LABEL version=$VERSION
 
 # Copy project files into the container

README.md

@@ -4,12 +4,19 @@ This project is a Flask-based web server designed to generate and display images
 
 ## Features
 
-*   **Web Interface:** A simple web interface to view generated images.
-*   **Image Generation:** Integrates with ComfyUI to generate images based on given prompts and models.
+*   **Web Interface:** A simple web interface to view generated images, manage favourites, and monitor job queues.
+*   **Image Generation:** Integrates with ComfyUI to generate images using SDXL, FLUX, and Qwen models based on given prompts.
+*   **Prompt Generation:** Automatic prompt generation using OpenWebUI or OpenRouter APIs with topic-based theming.
 *   **Scheduled Generation:** Automatically generates new images at a configurable time.
+*   **Favourites System:** Mark and manage favourite images.
+*   **Job Queue Management:** View and cancel running/pending image generation jobs.
+*   **Thumbnail Generation:** Automatic thumbnail creation for generated images.
+*   **Prompt Logging:** Maintains a log of recent prompts to avoid repetition.
+*   **Settings Management:** Web-based configuration editor for all settings.
 *   **Docker Support:** Comes with a `Dockerfile` and `docker-compose.yml` for easy setup and deployment.
-*   **Configurable:** Most options can be configured through a `user_config.cfg` file.
+*   **Configurable:** Most options can be configured through a `user_config.cfg` file or web interface.
 *   **Authentication:** Optional password protection for image creation.
+*   **Version Management:** Uses bump-my-version for version tracking.
 
 ## Prerequisites
 
@@ -33,8 +40,8 @@ This project is a Flask-based web server designed to generate and display images
     ```
 
 3.  **Configure the application:**
-    *   Copy the `user_config.cfg.sample` to `user_config.cfg`.
-    *   Edit `user_config.cfg` with your settings. See the [Configuration](#configuration) section for more details.
+    *   The `user_config.cfg` file will be automatically created from `user_config.cfg.sample` on first run if it doesn't exist.
+    *   Edit `user_config.cfg` with your settings, or use the web-based settings page accessible by clicking the version number in the bottom right corner of the home page. See the [Configuration](#configuration) section for more details.
 
 4.  **Run the application:**
     ```bash
@@ -51,8 +58,8 @@ This project is a Flask-based web server designed to generate and display images
     ```
 
 2.  **Configure the application:**
-    *   Copy the `user_config.cfg.sample` to `user_config.cfg`.
-    *   Edit `user_config.cfg` with your settings. The `comfyui_url` should be the address of your ComfyUI instance, accessible from within the Docker network (e.g., `http://host.docker.internal:8188` or your server's IP).
+    *   The `user_config.cfg` file will be automatically created from `user_config.cfg.sample` on first run if it doesn't exist.
+    *   Edit `user_config.cfg` with your settings, or use the web-based settings page accessible by clicking the version number in the bottom right corner of any page. The `comfyui_url` should be the address of your ComfyUI instance, accessible from within the Docker network (e.g., `http://host.docker.internal:8188` or your server's IP).
 
 3.  **Build and run with Docker Compose:**
     ```bash
@@ -79,24 +86,40 @@ The application is configured via the `user_config.cfg` file.
 | `[comfyui]` | `width`              | The width of the generated image.                                           | `1568`                |
 | `[comfyui]` | `height`             | The height of the generated image.                                          | `672`                 |
 | `[comfyui]` | `topics`             | A comma-separated list of topics to generate prompts from.                  |                       |
-| `[comfyui]` | `FLUX`               | Enable FLUX models (`True`/`False`).                                        | `False`               |
-| `[comfyui]` | `ONLY_FLUX`          | Only use FLUX models (`True`/`False`).                                      | `False`               |
+| `[comfyui]` | `secondary_topic`    | A secondary topic for prompt generation.                                    |                       |
+| `[comfyui]` | `flux`               | Enable FLUX models (`True`/`False`).                                        | `False`               |
+| `[comfyui]` | `qwen`               | Enable Qwen models (`True`/`False`).                                        | `False`               |
 | `[comfyui:flux]` | `models`       | A comma-separated list of FLUX models.                                      | `flux1-dev-Q4_0.gguf,flux1-schnell-Q4_0.gguf` |
+| `[comfyui:qwen]` | `models`       | A comma-separated list of Qwen models.                                      | `qwen-image-Q4_K_S.gguf, qwen-image-Q2_K.gguf` |
 | `[openwebui]` | `base_url`         | The base URL for OpenWebUI.                                                 | `https://openwebui`   |
 | `[openwebui]` | `api_key`          | The API key for OpenWebUI.                                                  | `sk-`                 |
 | `[openwebui]` | `models`           | A comma-separated list of models for OpenWebUI.                             | `llama3:latest,cogito:14b,gemma3:12b` |
+| `[openrouter]` | `enabled`          | Enable OpenRouter integration (`True`/`False`).                             | `False`               |
+| `[openrouter]` | `api_key`          | The API key for OpenRouter.                                                 |                       |
+| `[openrouter]` | `models`           | A comma-separated list of models for OpenRouter.                            | `mistralai/mistral-7b-instruct:free,google/gemma-7b-it:free,meta-llama/llama-3.1-8b-instruct:free` |
+| `[openrouter]` | `list_all_free_models` | List all free models (`True`/`False`).                                      | `False`               |
 
 ## Usage
 
 *   **Gallery:** Open your browser to `http://<server_ip>:<port>` to see the gallery of generated images.
-*   **Create Image:** Navigate to `/create` to manually trigger image generation.
-
+*   **Create Image:** Navigate to `/create` or `/create_image` to manually trigger image generation with various model options.
+*   **Job Queue:** Monitor and cancel running/pending jobs via the gallery interface.
+*   **API Endpoints:**
+    *   `/api/queue` - Get current job queue details (JSON)
+    *   `/cancel` - Cancel the current running job
+    
 ## Dependencies
 
 *   Flask
 *   comfy_api_simplified
 *   APScheduler
 *   Pillow
+*   tenacity
+*   nest_asyncio
+*   openai
+*   websockets
+*   bump-my-version
+*   openwebui-chat-client
 *   And others, see `requirements.txt`.
 
 ## Contributing

libs/generic.py

@@ -5,6 +5,7 @@ import sys
 import time
 import os
 import random
+import shutil
 from PIL import Image
 import nest_asyncio
 import json
@@ -38,10 +39,21 @@ def save_prompt(prompt):
 
 
 def load_config() -> configparser.ConfigParser:
-    """Loads user configuration from ./user_config.cfg."""
+    """Loads user configuration from ./user_config.cfg. If it doesn't exist, copies from user_config.cfg.sample."""
     user_config = configparser.ConfigParser()
+    config_path = "./user_config.cfg"
+    sample_path = "./user_config.cfg.sample"
+
+    if not os.path.exists(config_path):
+        if os.path.exists(sample_path):
+            shutil.copy(sample_path, config_path)
+            logging.info("Configuration file copied from sample.")
+        else:
+            logging.error("Neither user_config.cfg nor user_config.cfg.sample found.")
+            sys.exit(1)
+
     try:
-        user_config.read("./user_config.cfg")
+        user_config.read(config_path)
         logging.debug("Configuration loaded successfully.")
         return user_config
     except KeyError as e:
@@ -81,17 +93,20 @@ def get_details_from_png(path):
     try:
         date = datetime.fromtimestamp(os.path.getctime(path)).strftime("%d-%m-%Y")
         with Image.open(path) as img:
-            try:
+            data = json.loads(img.info["prompt"])
+            prompt = data['6']['inputs']['text']
+            if '38' in data and 'unet_name' in data['38']['inputs']:
                 # Flux workflow
-                data = json.loads(img.info["prompt"])
-                prompt = data['6']['inputs']['text']
                 model = data['38']['inputs']['unet_name'].split(".")[0]
-            except KeyError:
+            elif '4' in data and 'ckpt_name' in data['4']['inputs']:
                 # SDXL workflow
-                data = json.loads(img.info["prompt"])
-                prompt = data['6']['inputs']['text']
                 model = data['4']['inputs']['ckpt_name']
-            return {"p":prompt,"m":model,"d":date} or {"p":"","m":"","c":""}
+            elif '80' in data and 'unet_name' in data['80']['inputs']:
+                # Qwen workflow
+                model = data['80']['inputs']['unet_name'].split(".")[0]
+            else:
+                model = "unknown"
+            return {"p":prompt,"m":model,"d":date}
     except Exception as e:
         print(f"Error reading metadata from {path}: {e}")
         return ""
@@ -146,8 +161,13 @@ def load_openrouter_models_from_config():
     config = load_config()
     if config["openrouter"].get("enabled", "False").lower() == "true":
         models = config["openrouter"]["models"].split(",")
-        return sorted([model.strip() for model in models if model.strip()], key=str.lower)
-    return []
+        configured_models = sorted([model.strip() for model in models if model.strip()], key=str.lower)
+        free_models = []
+        if config["openrouter"].get("list_all_free_models", "False").lower() == "true":
+            from libs.openrouter import get_free_models
+            free_models = get_free_models()
+        return configured_models, free_models
+    return [], []
 
 def load_openwebui_models_from_config():
     config = load_config()
@@ -160,38 +180,82 @@ def load_prompt_models_from_config():
     """Load and return a list of available prompt generation models (both OpenWebUI and OpenRouter)."""
     config = load_config()
     prompt_models = []
-    
+
     # Add OpenWebUI models if configured
     if "openwebui" in config and "models" in config["openwebui"]:
         openwebui_models = config["openwebui"]["models"].split(",")
         prompt_models.extend([("openwebui", model.strip()) for model in openwebui_models if model.strip()])
-    
+
     # Add OpenRouter models if enabled and configured
     if config["openrouter"].get("enabled", "False").lower() == "true" and "models" in config["openrouter"]:
         openrouter_models = config["openrouter"]["models"].split(",")
         prompt_models.extend([("openrouter", model.strip()) for model in openrouter_models if model.strip()])
-    
+        # Add free models if flag is set
+        if config["openrouter"].get("list_all_free_models", "False").lower() == "true":
+            from libs.openrouter import get_free_models
+            free_models = get_free_models()
+            prompt_models.extend([("openrouter", model) for model in free_models])
+
     return prompt_models
 
 
+def build_user_content(topic: str = "random") -> str:
+    """Build the user content string for prompt generation, including topic instructions and recent prompts avoidance."""
+    config = load_config()
+    topic_instruction = ""
+    selected_topic = ""
+    secondary_topic_instruction = ""
+    # Unique list of recent prompts
+    recent_prompts = list(set(load_recent_prompts()))
+
+    if topic == "random":
+        topics = [t.strip() for t in config["comfyui"]["topics"].split(",") if t.strip()]
+        selected_topic = random.choice(topics) if topics else ""
+    elif topic != "":
+        selected_topic = topic
+    else:
+        # Decide on whether to include a topic (e.g., 30% chance to include)
+        topics = [t.strip() for t in config["comfyui"]["topics"].split(",") if t.strip()]
+        if random.random() < 0.3 and topics:
+            selected_topic = random.choice(topics)
+
+    if selected_topic != "":
+        topic_instruction = f" Incorporate the theme of '{selected_topic}' into the new prompt."
+
+    # Add secondary topic if configured and not empty
+    secondary_topic = config["comfyui"].get("secondary_topic", "").strip()
+    if secondary_topic:
+        secondary_topic_instruction = f" Additionally incorporate the theme of '{secondary_topic}' into the new prompt."
+
+    user_content = (
+        "Can you generate me a really random image idea, Do not exceed 20 words. Use clear language, not poetic metaphors."
+        + topic_instruction
+        + secondary_topic_instruction
+        + "Avoid prompts similar to the following:"
+        + "\n".join(f"{i+1}. {p}" for i, p in enumerate(recent_prompts))
+    )
+
+    return user_content
+
+
 def create_prompt_with_random_model(base_prompt: str, topic: str = "random"):
     """Create a prompt using a randomly selected model from OpenWebUI or OpenRouter.
-    
+
     If OpenWebUI fails, it will retry once. If it fails again, it will fallback to OpenRouter.
     """
     prompt_models = load_prompt_models_from_config()
-    
+
     if not prompt_models:
         logging.warning("No prompt generation models configured.")
         return None
-    
+
     # Randomly select a model
     service, model = random.choice(prompt_models)
-    
+
     # Import here to avoid circular imports
     from libs.openwebui import create_prompt_on_openwebui
     from libs.openrouter import create_prompt_on_openrouter
-    
+
     if service == "openwebui":
         try:
             # First attempt with OpenWebUI
@@ -199,13 +263,13 @@ def create_prompt_with_random_model(base_prompt: str, topic: str = "random"):
             result = create_prompt_on_openwebui(base_prompt, topic, model)
             if result:
                 return result
-            
+
             # If first attempt returns None, try again
             logging.warning("First OpenWebUI attempt failed. Retrying...")
             result = create_prompt_on_openwebui(base_prompt, topic, model)
             if result:
                 return result
-            
+
             # If second attempt fails, fallback to OpenRouter
             logging.warning("Second OpenWebUI attempt failed. Falling back to OpenRouter...")
             openrouter_models = [m for m in prompt_models if m[0] == "openrouter"]
@@ -215,7 +279,7 @@ def create_prompt_with_random_model(base_prompt: str, topic: str = "random"):
             else:
                 logging.error("No OpenRouter models configured for fallback.")
                 return "A colorful abstract composition"  # Default fallback prompt
-                
+
         except Exception as e:
             logging.error(f"Error with OpenWebUI: {e}")
             # Fallback to OpenRouter on exception
@@ -231,7 +295,7 @@ def create_prompt_with_random_model(base_prompt: str, topic: str = "random"):
             else:
                 logging.error("No OpenRouter models configured for fallback.")
                 return "A colorful abstract composition"  # Default fallback prompt
-    
+
     elif service == "openrouter":
         try:
             # Use OpenRouter
@@ -239,7 +303,7 @@ def create_prompt_with_random_model(base_prompt: str, topic: str = "random"):
         except Exception as e:
             logging.error(f"Error with OpenRouter: {e}")
             return "A colorful abstract composition"  # Default fallback prompt
-    
+
 
 user_config = load_config()
 output_folder = user_config["comfyui"]["output_dir"]

libs/openrouter.py

@@ -2,7 +2,7 @@ import random
 import logging
 from openai import OpenAI, RateLimitError
 import nest_asyncio
-from libs.generic import load_recent_prompts, load_config
+from libs.generic import load_recent_prompts, load_config, build_user_content
 from libs.openwebui import create_prompt_on_openwebui
 import re
 nest_asyncio.apply()
@@ -14,49 +14,78 @@ LOG_FILE = "./prompts_log.jsonl"
 user_config = load_config()
 output_folder = user_config["comfyui"]["output_dir"]
 
+def get_free_models():
+    """Fetch all free models from OpenRouter."""
+    if user_config["openrouter"].get("enabled", "False").lower() != "true":
+        return []
+    try:
+        client = OpenAI(
+            base_url="https://openrouter.ai/api/v1",
+            api_key=user_config["openrouter"]["api_key"],
+        )
+        all_models_response = client.models.list()
+        all_models = [m.id for m in all_models_response.data]
+        free_models = [m for m in all_models if "free" in m.lower()]
+        return sorted(free_models, key=str.lower)
+    except Exception as e:
+        logging.warning(f"Failed to fetch free models from OpenRouter: {e}")
+        return []
+
 def create_prompt_on_openrouter(prompt: str, topic: str = "random", model: str = None) -> str:
     """Sends prompt to OpenRouter and returns the generated response."""
+    # Reload config to get latest values
+    config = load_config()
     # Check if OpenRouter is enabled
-    if user_config["openrouter"].get("enabled", "False").lower() != "true":
+    if config["openrouter"].get("enabled", "False").lower() != "true":
         logging.warning("OpenRouter is not enabled in the configuration.")
         return ""
-    
-    topic_instruction = ""
-    selected_topic = ""
-    # Unique list of recent prompts
-    recent_prompts = list(set(load_recent_prompts()))
-    if topic == "random":
-        topics = [t.strip() for t in user_config["comfyui"]["topics"].split(",") if t.strip()]
-        selected_topic = random.choice(topics) if topics else ""
-    elif topic != "":
-        selected_topic = topic
-    else:
-        # Decide on whether to include a topic (e.g., 30% chance to include)
-        topics = [t.strip() for t in user_config["comfyui"]["topics"].split(",") if t.strip()]
-        if random.random() < 0.3 and topics:
-            selected_topic = random.choice(topics)
-    if selected_topic != "":
-        topic_instruction = f" Incorporate the theme of '{selected_topic}' into the new prompt."
 
-    user_content = (
-        "Can you generate me a really random image idea, Do not exceed 10 words. Use clear language, not poetic metaphors."
-        + topic_instruction
-        + "Avoid prompts similar to the following:"
-        + "\n".join(f"{i+1}. {p}" for i, p in enumerate(recent_prompts))
+    user_content = build_user_content(topic)
+
+    # Load configured models
+    configured_models = [m.strip() for m in user_config["openrouter"]["models"].split(",") if m.strip()]
+    if not configured_models:
+        logging.error("No OpenRouter models configured.")
+        return ""
+
+    # Create client early for model checking
+    client = OpenAI(
+        base_url="https://openrouter.ai/api/v1",
+        api_key=user_config["openrouter"]["api_key"],
     )
 
-    # Use the specified model or select a random model from the configured OpenRouter models
+    # Select model
     if model:
-        # Use the specified model
-        model = model
+        original_model = model
+        # Always check if model exists on OpenRouter
+        try:
+            all_models_response = client.models.list()
+            all_models = [m.id for m in all_models_response.data]
+            if model not in all_models:
+                # Fallback to random free model from all OpenRouter models
+                free_models = [m for m in all_models if "free" in m.lower()]
+                if free_models:
+                    model = random.choice(free_models)
+                    logging.info(f"Specified model '{original_model}' not found on OpenRouter, falling back to free model: {model}")
+                else:
+                    # No free models, fallback to random configured model
+                    model = random.choice(configured_models)
+                    logging.warning(f"Specified model '{original_model}' not found, no free models available on OpenRouter, using random configured model: {model}")
+            # else model exists, use it
+        except Exception as e:
+            logging.warning(f"Failed to fetch OpenRouter models for validation: {e}. Falling back to configured models.")
+            if model not in configured_models:
+                # Fallback to random free from configured
+                free_models = [m for m in configured_models if "free" in m.lower()]
+                if free_models:
+                    model = random.choice(free_models)
+                    logging.info(f"Specified model '{original_model}' not found, falling back to free configured model: {model}")
+                else:
+                    model = random.choice(configured_models)
+                    logging.warning(f"Specified model '{original_model}' not found, no free configured models available, using random configured model: {model}")
+            # else use the specified model
     else:
-        # Select a random model from the configured OpenRouter models
-        models = [m.strip() for m in user_config["openrouter"]["models"].split(",") if m.strip()]
-        if not models:
-            logging.error("No OpenRouter models configured.")
-            return ""
-        
-        model = random.choice(models)
+        model = random.choice(configured_models)
     
     try:
         client = OpenAI(
@@ -64,25 +93,50 @@ def create_prompt_on_openrouter(prompt: str, topic: str = "random", model: str =
             api_key=user_config["openrouter"]["api_key"],
         )
 
-        completion = client.chat.completions.create(
-            model=model,
-            messages=[
-                {
-                    "role": "system",
-                    "content": (
-                        "You are a prompt generator for Stable Diffusion. "
-                        "Generate a detailed and imaginative prompt with a strong visual theme. "
-                        "Focus on lighting, atmosphere, and artistic style. "
-                        "Keep the prompt concise, no extra commentary or formatting."
-                    ),
-                },
-                {
-                    "role": "user",
-                    "content": user_content,
-                },
-            ]
+        system_content = (
+            "You are a prompt generator for Stable Diffusion. "
+            "Generate a detailed and imaginative prompt with a strong visual theme. "
+            "Focus on lighting, atmosphere, and artistic style. "
+            "Keep the prompt concise, no extra commentary or formatting."
         )
 
+        # Try the specified model first
+        try:
+            completion = client.chat.completions.create(
+                model=model,
+                messages=[
+                    {
+                        "role": "system",
+                        "content": system_content,
+                    },
+                    {
+                        "role": "user",
+                        "content": user_content,
+                    },
+                ]
+            )
+        except Exception as e:
+            # If system message fails (e.g., model doesn't support developer instructions),
+            # retry with instructions included in user message
+            if "developer instruction" in str(e).lower() or "system message" in str(e).lower():
+                logging.info(f"Model {model} doesn't support system messages, retrying with instructions in user message")
+                combined_content = f"{system_content}\n\n{user_content}"
+                completion = client.chat.completions.create(
+                    model=model,
+                    messages=[
+                        {
+                            "role": "user",
+                            "content": combined_content,
+                        },
+                    ]
+                )
+            else:
+                # If it's another error, try fallback models
+                logging.warning(f"Error with model {model}: {e}. Trying fallback models.")
+                raise e
+
+        # If we get here, the completion was successful
+
         prompt = completion.choices[0].message.content.strip('"')
         match = re.search(r'"([^"]+)"', prompt)
         if not match:
@@ -106,5 +160,45 @@ def create_prompt_on_openrouter(prompt: str, topic: str = "random", model: str =
             logging.error("No OpenWebUI models configured for fallback.")
             return "A colorful abstract composition"  # Final fallback
     except Exception as e:
-        logging.error(f"Error generating prompt with OpenRouter: {e}")
+        # If the specified model fails, try fallback models
+        logging.warning(f"Primary model {model} failed: {e}. Trying fallback models.")
+
+        # Get all available models for fallback
+        configured_models = [m.strip() for m in user_config["openrouter"]["models"].split(",") if m.strip()]
+        free_models = get_free_models()
+
+        # Combine configured and free models, excluding the failed one
+        all_models = configured_models + free_models
+        fallback_models = [m for m in all_models if m != model]
+
+        if not fallback_models:
+            logging.error("No fallback models available.")
+            return ""
+
+        # Try up to 3 fallback models
+        for fallback_model in fallback_models[:3]:
+            try:
+                logging.info(f"Trying fallback model: {fallback_model}")
+                completion = client.chat.completions.create(
+                    model=fallback_model,
+                    messages=[
+                        {
+                            "role": "user",
+                            "content": f"{system_content}\n\n{user_content}",
+                        },
+                    ]
+                )
+                prompt = completion.choices[0].message.content.strip('"')
+                match = re.search(r'"([^"]+)"', prompt)
+                if not match:
+                    match = re.search(r":\s*\n*\s*(.+)", prompt)
+                if match:
+                    prompt = match.group(1)
+                logging.info(f"Successfully generated prompt with fallback model: {fallback_model}")
+                return prompt
+            except Exception as fallback_e:
+                logging.warning(f"Fallback model {fallback_model} also failed: {fallback_e}")
+                continue
+
+        logging.error("All models failed, including fallbacks.")
         return ""

libs/openwebui.py

@@ -1,7 +1,7 @@
 import random
 import logging
 import nest_asyncio
-from libs.generic import load_recent_prompts, load_config
+from libs.generic import load_recent_prompts, load_config, build_user_content
 import re
 from openwebui_chat_client import OpenWebUIClient
 from datetime import datetime
@@ -17,29 +17,9 @@ output_folder = user_config["comfyui"]["output_dir"]
 
 def create_prompt_on_openwebui(prompt: str, topic: str = "random", model: str = None) -> str:
     """Sends prompt to OpenWebui and returns the generated response."""
-    topic_instruction = ""
-    selected_topic = ""
-    # Unique list of recent prompts
-    recent_prompts = list(set(load_recent_prompts()))
-    if topic == "random":
-        topics = [t.strip() for t in user_config["comfyui"]["topics"].split(",") if t.strip()]
-        selected_topic = random.choice(topics)
-    elif topic != "":
-        selected_topic = topic
-    else:
-        # Decide on whether to include a topic (e.g., 30% chance to include)
-        topics = [t.strip() for t in user_config["comfyui"]["topics"].split(",") if t.strip()]
-        if random.random() < 0.3 and topics:
-            selected_topic = random.choice(topics)
-    if selected_topic != "":
-        topic_instruction = f" Incorporate the theme of '{selected_topic}' into the new prompt."
-
-    user_content = (
-        "Can you generate me a really random image idea, Do not exceed 10 words. Use clear language, not poetic metaphors."
-        + topic_instruction
-        + "Avoid prompts similar to the following:"
-        + "\n".join(f"{i+1}. {p}" for i, p in enumerate(recent_prompts))
-    )
+    # Reload config to get latest values
+    config = load_config()
+    user_content = build_user_content(topic)
 
     if model:
         # Use the specified model

routes/create_routes.py

@@ -38,8 +38,8 @@ def create():
     # Load all models (SDXL, FLUX, and Qwen)
     sdxl_models, flux_models, qwen_models = load_models_from_config()
     openwebui_models = load_openwebui_models_from_config()
-    openrouter_models = load_openrouter_models_from_config()
-    
+    openrouter_models, openrouter_free_models = load_openrouter_models_from_config()
+
     queue_count = get_queue_count()
     return render_template("create_image.html",
                          sdxx_models=sdxl_models,
@@ -47,6 +47,7 @@ def create():
                          qwen_models=qwen_models,
                          openwebui_models=openwebui_models,
                          openrouter_models=openrouter_models,
+                         openrouter_free_models=openrouter_free_models,
                          topics=load_topics_from_config(),
                          queue_count=queue_count)
 
@@ -68,8 +69,8 @@ def create_image_page():
     # Load all models (SDXL, FLUX, and Qwen)
     sdxl_models, flux_models, qwen_models = load_models_from_config()
     openwebui_models = load_openwebui_models_from_config()
-    openrouter_models = load_openrouter_models_from_config()
-    
+    openrouter_models, openrouter_free_models = load_openrouter_models_from_config()
+
     queue_count = get_queue_count()
     return render_template("create_image.html",
                          sdxl_models=sdxl_models,
@@ -77,6 +78,7 @@ def create_image_page():
                          qwen_models=qwen_models,
                          openwebui_models=openwebui_models,
                          openrouter_models=openrouter_models,
+                         openrouter_free_models=openrouter_free_models,
                          topics=load_topics_from_config(),
                          queue_count=queue_count)
 

templates/create_image.html

@@ -249,6 +249,13 @@
                 {% endfor %}
             </optgroup>
             {% endif %}
+            {% if openrouter_free_models %}
+            <optgroup label="OpenRouter Free">
+                {% for m in openrouter_free_models %}
+                <option value="openrouter:{{ m }}">{{ m }}</option>
+                {% endfor %}
+            </optgroup>
+            {% endif %}
         </select>
     </div>
     

user_config.cfg.sample

@@ -13,11 +13,15 @@ output_dir = ./output/
 prompt = "Generate a random detailed prompt for stable diffusion."
 width = 1568
 height = 672
-topics = 
+topics =
+secondary_topic =
 
-FLUX = False
-ONLY_FLUX = False
+flux = False
+qwen = False
+only_flux = False
 
+[comfyui:qwen]
+models = qwen-image-Q4_K_S.gguf, qwen-image-Q2_K.gguf
 [comfyui:flux]
 models = flux1-dev-Q4_0.gguf,flux1-schnell-Q4_0.gguf
 
@@ -29,4 +33,5 @@ models = llama3:latest,cogito:14b,gemma3:12b
 [openrouter]
 enabled = False
 api_key =
-models = mistralai/mistral-7b-instruct:free,google/gemma-7b-it:free,meta-llama/llama-3.1-8b-instruct:free
+models = mistralai/mistral-7b-instruct:free,google/gemma-7b-it:free,meta-llama/llama-3.1-8b-instruct:free
+list_all_free_models = False

workflow_flux_original.json

@@ -1,433 +0,0 @@
-{
-  "8": {
-    "inputs": {
-      "samples": [
-        "62",
-        1
-      ],
-      "vae": [
-        "27",
-        0
-      ]
-    },
-    "class_type": "VAEDecode",
-    "_meta": {
-      "title": "VAE Decode"
-    }
-  },
-  "22": {
-    "inputs": {
-      "clip_name1": "t5/t5xxl_fp8_e4m3fn.safetensors",
-      "clip_name2": "clip_l.safetensors",
-      "type": "flux",
-      "device": "default"
-    },
-    "class_type": "DualCLIPLoader",
-    "_meta": {
-      "title": "DualCLIPLoader"
-    }
-  },
-  "27": {
-    "inputs": {
-      "vae_name": "FLUX1/ae.safetensors"
-    },
-    "class_type": "VAELoader",
-    "_meta": {
-      "title": "Load VAE"
-    }
-  },
-  "32": {
-    "inputs": {
-      "upscale_model": [
-        "33",
-        0
-      ],
-      "image": [
-        "8",
-        0
-      ]
-    },
-    "class_type": "ImageUpscaleWithModel",
-    "_meta": {
-      "title": "Upscale Image (using Model)"
-    }
-  },
-  "33": {
-    "inputs": {
-      "model_name": "4x-UltraSharp.pth"
-    },
-    "class_type": "UpscaleModelLoader",
-    "_meta": {
-      "title": "Load Upscale Model"
-    }
-  },
-  "34": {
-    "inputs": {
-      "upscale_method": "lanczos",
-      "scale_by": 0.5,
-      "image": [
-        "32",
-        0
-      ]
-    },
-    "class_type": "ImageScaleBy",
-    "_meta": {
-      "title": "Half size"
-    }
-  },
-  "35": {
-    "inputs": {
-      "unet_name": "flux1-dev-Q4_0.gguf"
-    },
-    "class_type": "UnetLoaderGGUF",
-    "_meta": {
-      "title": "Unet Loader (GGUF)"
-    }
-  },
-  "40": {
-    "inputs": {
-      "int": 20
-    },
-    "class_type": "Int Literal (Image Saver)",
-    "_meta": {
-      "title": "Generation Steps"
-    }
-  },
-  "41": {
-    "inputs": {
-      "width": 720,
-      "height": 1080,
-      "aspect_ratio": "custom",
-      "swap_dimensions": "Off",
-      "upscale_factor": 2,
-      "prescale_factor": 1,
-      "batch_size": 1
-    },
-    "class_type": "CR Aspect Ratio",
-    "_meta": {
-      "title": "CR Aspect Ratio"
-    }
-  },
-  "42": {
-    "inputs": {
-      "filename": "THISFILE",
-      "path": "",
-      "extension": "png",
-      "steps": [
-        "40",
-        0
-      ],
-      "cfg": [
-        "52",
-        0
-      ],
-      "modelname": "flux1-dev-Q4_0.gguf",
-      "sampler_name": [
-        "50",
-        1
-      ],
-      "positive": [
-        "44",
-        0
-      ],
-      "negative": [
-        "45",
-        0
-      ],
-      "seed_value": [
-        "48",
-        0
-      ],
-      "width": [
-        "41",
-        0
-      ],
-      "height": [
-        "41",
-        1
-      ],
-      "lossless_webp": true,
-      "quality_jpeg_or_webp": 100,
-      "optimize_png": false,
-      "counter": 0,
-      "denoise": [
-        "53",
-        0
-      ],
-      "clip_skip": 0,
-      "time_format": "%Y-%m-%d-%H%M%S",
-      "save_workflow_as_json": true,
-      "embed_workflow": true,
-      "additional_hashes": "",
-      "download_civitai_data": true,
-      "easy_remix": true,
-      "speak_and_recognation": {
-        "__value__": [
-          false,
-          true
-        ]
-      },
-      "images": [
-        "34",
-        0
-      ]
-    },
-    "class_type": "Image Saver",
-    "_meta": {
-      "title": "CivitAI Image Saver"
-    }
-  },
-  "44": {
-    "inputs": {
-      "text": "",
-      "speak_and_recognation": {
-        "__value__": [
-          false,
-          true
-        ]
-      }
-    },
-    "class_type": "ttN text",
-    "_meta": {
-      "title": "Positive Prompt T5"
-    }
-  },
-  "45": {
-    "inputs": {
-      "text": "text, watermark, deformed Avoid flat colors, poor lighting, and artificial elements. No unrealistic elements, low resolution, or flat colors. Avoid generic objects, poor lighting, and inconsistent styles, blurry, low-quality, distorted faces, overexposed lighting, extra limbs, bad anatomy, low contrast",
-      "speak_and_recognation": {
-        "__value__": [
-          false,
-          true
-        ]
-      }
-    },
-    "class_type": "ttN text",
-    "_meta": {
-      "title": "Negative Prompt"
-    }
-  },
-  "47": {
-    "inputs": {
-      "text": [
-        "44",
-        0
-      ],
-      "speak_and_recognation": {
-        "__value__": [
-          false,
-          true
-        ]
-      },
-      "clip": [
-        "68",
-        1
-      ]
-    },
-    "class_type": "CLIPTextEncode",
-    "_meta": {
-      "title": "CLIP Text Encode (Prompt)"
-    }
-  },
-  "48": {
-    "inputs": {
-      "seed": 903006749445372,
-      "increment": 1
-    },
-    "class_type": "Seed Generator (Image Saver)",
-    "_meta": {
-      "title": "Seed"
-    }
-  },
-  "49": {
-    "inputs": {
-      "scheduler": "beta"
-    },
-    "class_type": "Scheduler Selector (Comfy) (Image Saver)",
-    "_meta": {
-      "title": "Scheduler Selector"
-    }
-  },
-  "50": {
-    "inputs": {
-      "sampler_name": "euler"
-    },
-    "class_type": "Sampler Selector (Image Saver)",
-    "_meta": {
-      "title": "Sampler Selector (Image Saver)"
-    }
-  },
-  "51": {
-    "inputs": {
-      "images": [
-        "8",
-        0
-      ]
-    },
-    "class_type": "PreviewImage",
-    "_meta": {
-      "title": "Preview Image"
-    }
-  },
-  "52": {
-    "inputs": {
-      "float": 3.5
-    },
-    "class_type": "Float Literal (Image Saver)",
-    "_meta": {
-      "title": "CFG"
-    }
-  },
-  "53": {
-    "inputs": {
-      "float": 1
-    },
-    "class_type": "Float Literal (Image Saver)",
-    "_meta": {
-      "title": "Denoise"
-    }
-  },
-  "60": {
-    "inputs": {
-      "clip_l": "",
-      "t5xxl": [
-        "44",
-        0
-      ],
-      "guidance": [
-        "52",
-        0
-      ],
-      "speak_and_recognation": {
-        "__value__": [
-          false,
-          true
-        ]
-      },
-      "clip": [
-        "68",
-        1
-      ]
-    },
-    "class_type": "CLIPTextEncodeFlux",
-    "_meta": {
-      "title": "CLIPTextEncodeFlux"
-    }
-  },
-  "62": {
-    "inputs": {
-      "noise": [
-        "65",
-        0
-      ],
-      "guider": [
-        "67",
-        0
-      ],
-      "sampler": [
-        "63",
-        0
-      ],
-      "sigmas": [
-        "64",
-        0
-      ],
-      "latent_image": [
-        "41",
-        5
-      ]
-    },
-    "class_type": "SamplerCustomAdvanced",
-    "_meta": {
-      "title": "SamplerCustomAdvanced"
-    }
-  },
-  "63": {
-    "inputs": {
-      "sampler_name": [
-        "50",
-        0
-      ]
-    },
-    "class_type": "KSamplerSelect",
-    "_meta": {
-      "title": "KSamplerSelect"
-    }
-  },
-  "64": {
-    "inputs": {
-      "scheduler": [
-        "49",
-        0
-      ],
-      "steps": [
-        "40",
-        0
-      ],
-      "denoise": [
-        "53",
-        0
-      ],
-      "model": [
-        "68",
-        0
-      ]
-    },
-    "class_type": "BasicScheduler",
-    "_meta": {
-      "title": "BasicScheduler"
-    }
-  },
-  "65": {
-    "inputs": {
-      "noise_seed": [
-        "48",
-        0
-      ]
-    },
-    "class_type": "RandomNoise",
-    "_meta": {
-      "title": "RandomNoise"
-    }
-  },
-  "67": {
-    "inputs": {
-      "model": [
-        "68",
-        0
-      ],
-      "conditioning": [
-        "47",
-        0
-      ]
-    },
-    "class_type": "BasicGuider",
-    "_meta": {
-      "title": "BasicGuider"
-    }
-  },
-  "68": {
-    "inputs": {
-      "lora_01": "None",
-      "strength_01": 1,
-      "lora_02": "None",
-      "strength_02": 1,
-      "lora_03": "None",
-      "strength_03": 1,
-      "lora_04": "None",
-      "strength_04": 1,
-      "model": [
-        "35",
-        0
-      ],
-      "clip": [
-        "22",
-        0
-      ]
-    },
-    "class_type": "Lora Loader Stack (rgthree)",
-    "_meta": {
-      "title": "Lora Loader Stack (rgthree)"
-    }
-  }
-}

workflow_qwen.json

@@ -1,45 +1,26 @@
 {
-  "93": {
+  "3": {
     "inputs": {
-      "text": "jpeg compression",
-      "speak_and_recognation": {
-        "__value__": [
-          false,
-          true
-        ]
-      },
-      "clip": [
-        "126",
-        0
-      ]
-    },
-    "class_type": "CLIPTextEncode",
-    "_meta": {
-      "title": "CLIP Text Encode (Prompt)"
-    }
-  },
-  "95": {
-    "inputs": {
-      "seed": 22,
-      "steps": 10,
-      "cfg": 4.5,
+      "seed": 367723847870487,
+      "steps": 8,
+      "cfg": 2.5,
       "sampler_name": "euler",
-      "scheduler": "normal",
+      "scheduler": "simple",
       "denoise": 1,
       "model": [
-        "127",
+        "66",
         0
       ],
       "positive": [
-        "100",
+        "6",
         0
       ],
       "negative": [
-        "93",
+        "7",
         0
       ],
       "latent_image": [
-        "97",
+        "58",
         0
       ]
     },
@@ -48,26 +29,40 @@
       "title": "KSampler"
     }
   },
-  "97": {
+  "6": {
     "inputs": {
-      "width": 1280,
-      "height": 768,
-      "length": 1,
-      "batch_size": 1
+      "text": "Cat sitting at desk, surrounded by taxidermied dinosaurs",
+      "clip": [
+        "38",
+        0
+      ]
     },
-    "class_type": "EmptyHunyuanLatentVideo",
+    "class_type": "CLIPTextEncode",
     "_meta": {
-      "title": "EmptyHunyuanLatentVideo"
+      "title": "Positive"
     }
   },
-  "98": {
+  "7": {
+    "inputs": {
+      "text": "text, watermark, deformed Avoid flat colors, poor lighting, and artificial elements. No unrealistic elements, low resolution, or flat colors. Avoid generic objects, poor lighting, and inconsistent styles, blurry, low-quality, distorted faces, overexposed lighting, extra limbs, bad anatomy, low contrast",
+      "clip": [
+        "38",
+        0
+      ]
+    },
+    "class_type": "CLIPTextEncode",
+    "_meta": {
+      "title": "Negative"
+    }
+  },
+  "8": {
     "inputs": {
       "samples": [
-        "95",
+        "3",
         0
       ],
       "vae": [
-        "128",
+        "39",
         0
       ]
     },
@@ -76,86 +71,84 @@
       "title": "VAE Decode"
     }
   },
-  "100": {
+  "38": {
     "inputs": {
-      "text": "Terminator riding a push bike",
-      "speak_and_recognation": {
-        "__value__": [
-          false,
-          true
-        ]
-      },
-      "clip": [
-        "126",
-        0
-      ]
-    },
-    "class_type": "CLIPTextEncode",
-    "_meta": {
-      "title": "CLIP Text Encode (Prompt)"
-    }
-  },
-  "102": {
-    "inputs": {
-      "images": [
-        "129",
-        0
-      ]
-    },
-    "class_type": "PreviewImage",
-    "_meta": {
-      "title": "Preview Image"
-    }
-  },
-  "126": {
-    "inputs": {
-      "clip_name": "Qwen2.5-VL-7B-Instruct-Q3_K_M.gguf",
+      "clip_name": "qwen_2.5_vl_7b_fp8_scaled.safetensors",
       "type": "qwen_image",
-      "device": "cuda:1",
-      "virtual_vram_gb": 6,
-      "use_other_vram": true,
-      "expert_mode_allocations": ""
+      "device": "default"
     },
-    "class_type": "CLIPLoaderGGUFDisTorchMultiGPU",
+    "class_type": "CLIPLoader",
     "_meta": {
-      "title": "CLIPLoaderGGUFDisTorchMultiGPU"
+      "title": "Load CLIP"
     }
   },
-  "127": {
+  "39": {
     "inputs": {
-      "unet_name": "qwen-image-Q2_K.gguf",
-      "device": "cuda:0",
-      "virtual_vram_gb": 6,
-      "use_other_vram": true,
-      "expert_mode_allocations": ""
+      "vae_name": "qwen_image_vae.safetensors"
     },
-    "class_type": "UnetLoaderGGUFDisTorchMultiGPU",
+    "class_type": "VAELoader",
     "_meta": {
-      "title": "UnetLoaderGGUFDisTorchMultiGPU"
+      "title": "Load VAE"
     }
   },
-  "128": {
+  "58": {
     "inputs": {
-      "vae_name": "qwen_image_vae.safetensors",
-      "device": "cuda:1"
+      "width": 720,
+      "height": 1088,
+      "batch_size": 1
     },
-    "class_type": "VAELoaderMultiGPU",
+    "class_type": "EmptySD3LatentImage",
     "_meta": {
-      "title": "VAELoaderMultiGPU"
+      "title": "CR Aspect Ratio"
     }
   },
-  "129": {
+  "60": {
     "inputs": {
-      "offload_model": true,
-      "offload_cache": true,
-      "anything": [
-        "98",
+      "filename_prefix": "ComfyUI",
+      "images": [
+        "8",
         0
       ]
     },
-    "class_type": "VRAMCleanup",
+    "class_type": "SaveImage",
     "_meta": {
-      "title": "🎈VRAM-Cleanup"
+      "title": "Save Image"
+    }
+  },
+  "66": {
+    "inputs": {
+      "shift": 3.1000000000000005,
+      "model": [
+        "73",
+        0
+      ]
+    },
+    "class_type": "ModelSamplingAuraFlow",
+    "_meta": {
+      "title": "ModelSamplingAuraFlow"
+    }
+  },
+  "73": {
+    "inputs": {
+      "lora_name": "Qwen-Image-Lightning-8steps-V1.0.safetensors",
+      "strength_model": 1,
+      "model": [
+        "80",
+        0
+      ]
+    },
+    "class_type": "LoraLoaderModelOnly",
+    "_meta": {
+      "title": "LoraLoaderModelOnly"
+    }
+  },
+  "80": {
+    "inputs": {
+      "unet_name": "qwen-image-Q4_K_S.gguf"
+    },
+    "class_type": "UnetLoaderGGUF",
+    "_meta": {
+      "title": "Load Checkpoint"
     }
   }
 }