Uncensored Prompt Generator

A local, privacy-focused prompt generator for image and video generation using Ollama. Supports reference image analysis and generates detailed, uncensored prompts.

Features

• 🔓 Uncensored: Runs locally with no content restrictions
• 🖼️ Image Reference: Analyze and incorporate reference images
• 🎬 Multi-Format: Generate prompts for both images and videos
• 🏠 100% Local: All processing happens on your machine
• 🚀 Interactive & CLI modes: Use interactively or in scripts

Requirements

• Python 3.7+
• Ollama installed and running

Installation

1. Install Ollama

Windows: Download from ollama.com

Linux/Mac:

curl -fsSL https://ollama.com/install.sh | sh

2. Install Python Dependencies

pip install -r requirements.txt

3. Install Recommended Ollama Models

Uncensored text models (choose one or more):

# Dolphin Mistral (Recommended - highly uncensored)
ollama pull dolphin-mistral

# Dolphin Mixtral (Larger, more capable)
ollama pull dolphin-mixtral

# Nous Hermes (Alternative uncensored model)
ollama pull nous-hermes2

# WizardLM Uncensored
ollama pull wizardlm-uncensored

Vision models (for image reference support):

# LLaVA (Recommended)
ollama pull llava

# BakLLaVA (Alternative)
ollama pull bakllava

# LLaVA 34B (Larger, more detailed)
ollama pull llava:34b

Usage

Interactive Mode

Simply run the script:

python prompt_generator.py

Commands:

• Type your request directly to generate prompts
• /image [path] - Set a reference image
• /video - Switch to video prompt mode
• /img - Switch to image prompt mode
• /model [name] - Override the default model
• /clear - Clear reference image
• /quit - Exit

Example session:

[IMAGE] > a cyberpunk city at night

🔄 Generating prompt...

============================================================
GENERATED PROMPT:
============================================================
A sprawling cyberpunk metropolis under a neon-lit night sky...
============================================================

[IMAGE] > /image reference.jpg
✓ Reference image set: reference.jpg

[IMAGE] > create similar but with rain

CLI Mode

Use in scripts or one-off generations:

# Basic usage
python prompt_generator.py "a fantasy landscape"

# Video prompt
python prompt_generator.py "epic battle scene" --type video

# With reference image
python prompt_generator.py "enhance this style" --image reference.jpg

# Override model
python prompt_generator.py "portrait" --model dolphin-mixtral

Recommended Uncensored Models

Best for Prompt Generation:

1. dolphin-mistral - Fast, highly uncensored, great for prompts
2. dolphin-mixtral - More capable but slower
3. nous-hermes2 - Good alternative
4. wizardlm-uncensored - Reliable uncensored option

For Image Analysis:

1. llava - Standard vision model
2. llava:34b - Higher quality analysis (requires more RAM)
3. bakllava - Alternative vision model

Configuration

Edit the script to change default models:

self.text_model = "dolphin-mistral"  # Your preferred text model
self.vision_model = "llava"           # Your preferred vision model

Or use the /model command in interactive mode.

Tips for Best Results

1. Be Specific: The more details you provide, the better the output
2. Use References: Image references help maintain style consistency
3. Iterate: Generate, refine, and regenerate for best results
4. Model Selection: Larger models (mixtral, 34b) give more detailed prompts but are slower

Troubleshooting

"Cannot connect to Ollama"

Make sure Ollama is running:

ollama serve

"Model not found"

Pull the model first:

ollama pull dolphin-mistral
ollama pull llava

Slow generation

• Use smaller models (mistral instead of mixtral)
• Close other applications
• Check if GPU acceleration is working: ollama list

Out of memory

• Use smaller models
• For vision models, try llava:7b instead of llava:34b
• Close other applications

Privacy & Safety

• ✅ 100% local processing
• ✅ No data sent to external servers
• ✅ No logging or tracking
• ✅ Full control over content generation

Advanced Usage

Batch Processing

Create a script to generate multiple prompts:

from prompt_generator import PromptGenerator

generator = PromptGenerator()

prompts = [
    "fantasy castle",
    "sci-fi spaceship",
    "portrait of warrior"
]

for prompt in prompts:
    result = generator.generate_prompt(prompt)
    print(f"{prompt} -> {result}\n")

Custom System Prompts

Modify the system prompts in the script for different output styles.

License

Free to use and modify for any purpose.

Disclaimer

This tool is for creative and artistic purposes. Users are responsible for ensuring their use complies with applicable laws and the terms of service of any platforms where generated content is used.