🚀 Prompt Generator Pro (v5.5)

The All-in-One desktop suite designed to supercharge the workflow of AI Artists (Stable Diffusion, ComfyUI, Midjourney, NovelAI).

👋 If you like what I do and want to support the development, feel free to buy me a coffee:

WHAT IS PROMPT GENERATOR PRO?

Prompt Generator Pro (formerly MGC) has evolved into the ultimate All-in-One desktop companion designed to supercharge the workflow of AI Artists (Stable Diffusion, ComfyUI, Midjourney, NovelAI).

WHAT DOES IT DO?

It bridges the gap between your imagination and the canvas. Whether you need to brainstorm ideas with an integrated AI (Cloud or Local), extract hidden metadata directly from images, or magically sanitize and format chaotic text blocks into pristine prompts, PGP handles it all in a highly customizable, modern workspace.

WHY DO I NEED IT?

• Saves Time: Instantly clean, sort, and weight tags, or extract full generation data from dragged-and-dropped images (even directly from your web browser).

• AI Brainstorming: Connect your favorite AI models (Gemini, Llama 3, ChatGPT, Ollama) to translate, continue stories, or generate prompts on the fly—now with immersive Voice (TTS) responses!

• Absolute Consistency: Enforce high-quality prompt structures with your custom presets and automated blacklists.

• All-in-One Organization: Manage your favorite tag combos (Secret Forge), AI assistants, and image galleries in a single, lightning-fast application.

Hi everyone,

Version 5.5 is a transitional release focused heavily on under-the-hood optimization, core stability, and quality-of-life adjustments. Here is a breakdown of what’s new:

Gallery & Layout Performance

• Gallery Mode Switch Fixed: Resolved the synchronization hitch when toggling back and forth between Base (1-column) and Bulk (Grid) view modes.

• Video Persistence Fix: Patched a minor bug related to right-click context menu behavior and state retention during video playback.

• Format Filtering & Prioritization: Introduced a new segmented file format layout that supports custom prioritization. This is ideal for massive datasets with thousands of images, as you don't need to manually minimize everything—you can jump straight to the format you want. Note: Currently, if you keep a format minimized and then switch the sorting order, it might carry over as hidden in the new view. We are aware of this minor buffer-retention quirk; however, thanks to the new format priority system, it doesn't disrupt the workflow. A complete overhaul for this persistence sync will be added in a future build.

• Grid Spacing: Bulk mode has temporarily reduced padding between thumbnails to guarantee loading stability. This will be refined in upcoming builds.

Metadata Extraction & AI Backend

• Advanced Node Surgery: The ComfyUI node inspector can now parse complex video configurations, nested JSON parameters, and intricate custom workflows.

• Civitai Connection Update: Upgraded the worker engine to allow significantly better detection of active resources, hashes, and embedded models.

• Token Optimization & Translation: Implemented a resource-aware caching strategy in the translation module to minimize unnecessary token expenditures. Google Translation endpoints have also been updated to ensure steady parsing.

File Management & Shortcuts (UX)

• Directory Operations: General performance and stability fixes applied to load_folder backend logic.

• Keybinding Polish: Refined shortcuts like F5 (Quick Refresh) and the Delete key behavior. Deleting files now safely shifts focus to adjacent items instead of snapping the scroll view back to the top of the gallery.

As always, feel free to drop your feedback or feature suggestions in the comments below. Enjoy the update!

What's New & Release Notes for Version 5

Version 5 is finally here! This update focuses on giving you absolute control over your metadata, optimizing your AI workflows, and massively improving the visual and user experience.

New Features & Tools

• New ASCII Studio: A complete suite to transform your images! It's not just a filter, it's a full rendering studio with multiple options:

  • Art Styles: Choose between Basic, Normal, Braille, Manga, Lineart, and the stunning HQ modes (High-Resolution ASCII).

  • Export Formats: Save your artwork as plain text (.txt) or directly rendered as an image (.png).

  • Metadata Control:

    • Mask Mode: Generate your ASCII image while protecting and injecting the original metadata into the new file.

    • Metadata Destruction: Create your art and wipe out any trace of the original information (ideal for extreme privacy).

• New Tagger Studio: You now have granular control over image / video analysis. It allows you to edit the AI extraction thresholds to get more or less precise tags according to your needs.

• Native Windows/PC Integration: The program now supports opening files directly from your system! Double-click or use "Open with..." and the app will load the file instantly.

Metadata, ComfyUI & Datasets

• Improved Node Surgery (ComfyUI): The cleaning of positive prompts has been optimized. The program now intelligently detects and removes disconnected or junk nodes generated during extraction.

• Expanded Metadata Reader: We have added support for new reading formats. (Note: If you have an AI file whose format is not yet detected, don't hesitate to let me know in the comments! It will be incredibly helpful for me to add it in future patches).

• Enhanced Autopsy Mode: Deep scans are now much more precise and robust for stubborn files.

• Quick Copy: New option to copy metadata to your clipboard with a single click.

• Dataset Generator: General stability and speed improvements in batch creation, along with a major update to the Autotagger engine.

Interface & User Experience (UX)

• Remastered Menus: Visual improvements across the entire interface and a new contextual Sub-menu that makes main tools much easier and faster to reach.

• Smart Translation System: Deep improvements to the language engine and fixes for conflicts with file search filters.

• Visual Safety & Fallbacks: New dynamic icons and smart alerts that clearly notify you if the program is activating "Safe Mode" (local engine) due to a missing API Key or lack of internet connection.

Bug Fixes & Performance

• Upscaler (ONNX): Fixed critical bugs that caused ONNX library incompatibilities on certain systems when trying to upscale images.

• Updated Parameters: Internal parameters have been adjusted to ensure maximum compatibility with the latest image generation engines.

• Resolved multiple minor bugs reported in previous versions to guarantee a flawless user experience.

v4.5 GALLERY UPDATE 2

Hi everyone! Here is a "small" update featuring about 5,000 lines of new code. This will likely be the last major update of the month, so if you run into any issues, feel free to leave a comment so I can patch them quickly. This week, I’ll be focusing on finishing a few checkpoints, fixing certain samplers/schedulers, and releasing a couple of LoRAs.

Here is everything new packed into Version 4.5:

🗿 Smart Directory System (Moving away from Onefile)

To allow for much deeper customization, we are moving away from the single-file (onefile) architecture.

• The program now utilizes an internal subdirectory (_internal) where most folders, JSON configs, and models have been relocated, leaving only the .exe outside.

• This provides better file organization, performance, and security. (Tip: Creating a desktop shortcut to the executable and hiding the main folder on your C: drive remains the best way to keep your JSONs safe).

• Re-routing issues regarding Nexus Mode have also been completely resolved.

🤖 Background Autotagger (Waifu Diffusion Engine)

• Integrated a background tagging engine heavily based on Waifu Diffusion.

• You can now tag videos, images, and entire folders to generate training datasets.

• Dataset creation processes now run fully in the background to ensure a seamless workflow.

🎨 Built-in Image Editing Suite

We have added direct editing tools right inside the viewer panel:

• Watermarking: Interactive placement of logos and PNG overlays.

• Transformations: Mirror flipping and canvas rotation tools.

• Freehand Brush: An interactive drawing layer for quick canvas annotations.

• Upscaler & Pixel Art: Resolution upscalers and a custom Pixel Perfect mode to turn any graphic into pixel art.

• 🛡️ Metadata Preservation: Crucially, all these editing tools preserve AI metadata (Prompt DNA) during the saving process, so you will never lose your generation info.

🗔 Multitasking Gallery

• Check out the new icon to the left of the title bar. You can now "unlock" the gallery to turn it into a 100% independent window.

• This allows you to keep the gallery view active on one screen while using Prompt Forge mode or the chat module simultaneously on another.

🌐 Expanded Web Parsing & Link Scraper

We have overhauled the supported sites table for better CDN bypassing and original image extraction. Major updates applied to:

• Pixiv (img-original), Kemono, Aibooru, and most Danbooru image board variants. (Note: Individual downloads for Danbooru are still being linked; other platforms were prioritized since PGP's core focus is creating, editing, and injecting metadata).

🛠️ Technical Fixes & Roadmap

• Fixed GGUF backend support to better facilitate local generation features.

• Sorted out multiple NumPy compilation headaches caused by the new upscalers and tagger libraries.

• Future Plans (v5.0): I am currently evaluating the addition of a built-in custom web browser specialized for asset scraping.

Thanks for supporting the project, and see you in the next update! ~

v4.0: GALLERY UPDATE

We’ve completely overhauled PGP, transforming it from an image utility into a true multimedia workstation. More customization, more power, and deeper analysis than ever before. Here is what's new:

• Native Video Player with Zoom & Fullscreen: Say goodbye to static galleries! PGP now includes a hardware-accelerated video player (powered by MPV) seamlessly integrated into the gallery. It features interactive zoom, fullscreen mode, and a transparent, auto-hiding progress bar that floats elegantly over your animations.

• Universal Metadata & "Autopsy Mode": The extraction engine has been completely rewritten. It now effortlessly reads incredibly complex, nested workflows from ComfyUI, Civitai Video AI, and more. If a file is corrupted or stubborn, the new Autopsy Mode performs a deep binary scan to rescue hidden prompts and generation data from the abyss.

• Community Font Store & Granular UI Scaling: Personalize your workspace to the extreme. Access a massive new in-app store with over 700 typography fonts ready to download. Plus, a new advanced configuration menu lets you individually adjust the text size of every single button, list, and text box in the app to fit your screen perfectly.

• Animated UI Backgrounds & New Themes: Make the app truly yours. You can now set animated GIFs or WebP files as your main application background, alongside several new sleek UI color themes.

• Universal Image Metadata & Web Drag-and-Drop: We've vastly expanded our direct browser integration! You can now drag and drop images directly from even more websites (including PixAI, Pixiv, Civitai, Aibooru, Gelbooru, Tensor.art, r34, and more) straight into the app. PGP will instantly download the media and extract its hidden generation data in milliseconds.

• Video Metadata Injection (Zero RAM): You can now inject, edit, and save metadata directly into .mp4, .webm, and animated .webp files. PGP edits the binary structure instantly without re-encoding the video, keeping your CPU cool and your RAM free.

• Image Metadata Injection: Just like with videos, you can now instantly inject, edit, and restore lost metadata directly into your downloaded PNGs or JPGs (perfect for fixing stripped images from sites like PixAI), saving the exact generation parameters right into the file's structure.

💡 PRO-TIP: Auto-Install FFmpeg for Video Magic! To unlock video metadata injection and media conversion, PGP requires the FFmpeg engine. Since it's not bundled by default to save space, simply right-click any video or GIF in the gallery and select "Convert to Animated WebP". The app will automatically prompt you to download and install the required engine in seconds!

v3.5: THE "NEXUS" UPDATE

We’ve pushed the boundaries of what a prompt tool can do. Here is what's new:

• Voice Nexus (TTS Integration): Your AI Assistant now speaks! Full integration with offline local models (Kokoro v1.0, Piper) and cloud voices (Edge TTS). It even includes a custom Phonetic Acting Editor for emotional responses and stutters.

• Advanced Gallery & Metadata 2.0: A complete built-in image gallery. Drag & drop images straight from your browser (with smart CDN bypass for Civitai, Tensor.art, Aibooru, and PixAI) to instantly extract generation data. It even uses the Civitai API to visually detect the Checkpoints and LoRAs used!

• Secret Forge Sandbox: A new immersive, drag-and-drop workspace. Visually test tools, chain them together in a grid, see real-time explanations, and save your favorite combos into "Favorite Groups".

• "Tools Only" Fast Mode: Need speed? Toggle "Tools Mode" in Appearance settings to completely disable the AI and Audio modules, allowing the app to open instantly with minimal RAM usage.

• 40+ Languages Supported: The smart translation system has been supercharged to automatically translate the app's UI into over 40 languages using AI.

🧙‍♂️ SECRETS, PRO-TIPS & MINI-GUIDES

HIDDEN BUTTONS & EASTER EGGS There are secret features hidden in plain sight:

• The Secret Forge: Click on the main "Prompt Generator Pro" title at the top left of the screen. It will open a hidden full-screen, drag-and-drop sandbox workspace!

• Zen Chat Mode: Hover over or click the Robot "🤖" icon inside the AI Assistant tab. It reveals a hidden menu to toggle "Chat Only Mode", hiding all other tools for deep focus.

• Secondary Menu: Click the AGGA logo in the left sidebar to reveal a hidden secondary menu containing the Gallery and Metadata Reader.

GALLERY & METADATA MINI-GUIDE

• Universal Reader: Drag & drop any image from Automatic1111, ComfyUI, NovelAI, or SwarmUI to instantly read its prompt.

• Web Drag & Drop: You don't need to download images first! You can drag images directly from your browser (Civitai, Tensor.art, Aibooru, pixai, etc) into the app and it will extract the data.

THE PIXAI MAGIC TRICK (Metadata Injection) PixAI strips all metadata from its downloaded images. PGP has a secret trick to fix this:

1. Open your downloaded PixAI image in PGP's Gallery.

2. Go to the PixAI website and simply COPY the prompt text or the whole page (Ctrl+C).

3. Go back to PGP and press Ctrl + V.

4. MAGIC! The app will instantly parse your clipboard, inject the hidden metadata into your image, and save a new, fully restored PNG file automatically.

WORKFLOW EXAMPLE (The Gelbooru Method) Tired of manually cleaning copied tags? Try this:

1. Copy a massive block of text directly from a Gelbooru image page.

2. Paste it in the Input box.

3. Click "Gelbooru" tool -> Click "Sort" -> Click "Professional" -> Click "Copy". You just turned garbage into a masterpiece prompt in 3 seconds!

💾 STORAGE SAVING TIP The local AI voice models (inside the kokoro and piper folders) are amazing but quite large. If you want to free up hard drive space, you can safely delete these folders! The app WILL NOT break. It will automatically detect they are missing and seamlessly switch to using cloud-based voices (Edge TTS) instead.

v2.5: The "Speed & Creativity" Update

• Performance Boost: Significantly increased startup speed and optimized background processes.

• Gelbooru Tool: A brand new formatting tool specifically designed to clean up messy tag blocks copied directly from Gelbooru and Danbooru.

• Novel Writing & Correction: New dedicated AI tools added for continuing stories and correcting grammar/style effortlessly.

• Advanced UI Control: Added a new text color editor, background scaling fixes, and updated visual themes.

• Chat & Localization: Added new chat configuration settings and expanded the translation dictionary to be 98% complete.

v2.0: The "Prompt Forge" Revolution

🆕 New Features:

• "Bring Your Own AI" (BYOAI): Connect Cloud IAs (Gemini, Grok, OpenRouter) or Local SLMs (Ollama, LM Studio, Jan.ai) directly into the app.

• Modes: Generator (Strict Tags), Chat (Brainstorming), Translator (Native language -> Tags), and Uncensored.

• Chat History: Auto-saves your conversations so you never lose an idea.

• Glassmorphism UI: A modern, transparent, and fluid interface that looks great on your desktop.

• Smart Translation Editor: An integrated JSON editor with an "AI Auto-Translate" button to instantly localize the app to any language.

• Advanced Config Menu: Manage API keys, select specific AI models, and customize the UI themes (Rounded/Flat).

v1.5: The "Stability" Update

This version focused on polishing the logic and fixing user-reported annoyances. 🔧 Corrections & Improvements:

• 🧠 Smart "Clean" Logic: Fixed bugs where the cleaning tool would accidentally delete necessary parentheses or mix up spacing. It became context-aware.

• 🚫 Blacklist Fix: Solved an issue where adding/removing words from the blacklist wouldn't update immediately without a restart.

• 🖱️ Drag & Drop V1: Introduced the ability to drag a PNG file into the text box to read simple metadata.

• 👻 Transparency Foundation: Added the first implementation of window opacity sliders.

v1.0: The Core Toolkit

These are the essential tools that started it all and remain the backbone of the application. 🛠️ The Essential Tools (Available since v1):

A. Formatting & Cleanup:

• Clean: Instantly removes double spaces, trash characters, and fixes line breaks.

• Sort: Alphabetizes tags to maintain a consistent prompt structure (helping seed consistency).

• Separate: Ensures all tags are correctly comma-separated.

• Unique: Removes duplicate tags to save token space.

B. Enhancement & Control:

• Professional: Injects a preset list of "quality boosters" (e.g., masterpiece, best quality, highres) with one click.

• Weight (1.3x): Wraps all tags in standard weighting syntax (tag:1.3).

• Triple Emphasis: Wraps tags in (((tag))) for maximum attention.

• Remove Weight: Strips all brackets and weights, returning the prompt to raw text.

• Randomize: Shuffles tags for experimental results.

Thank you so much for your support and happy generating! ♥