AI Studio Pro Documentation
Everything you need to know to get started with AI Studio Pro. Simple, straightforward, no technical jargon.
🚀 Getting Started
What You Need
Before you install AI Studio Pro, make sure your computer has:
- Windows 10 or 11 (64-bit)
- NVIDIA RTX 3060 with 12GB memory (or better)
- 32GB of RAM
- 50GB+ of free hard drive space
⚠️ Important: Check Your GPU
The RTX 3060 12GB is the minimum. If you have an 8GB card, many features won't work. To check your GPU, right-click on your desktop, select "NVIDIA Control Panel," and look under "System Information."
Installation
1Download from Microsoft Store
Click the "Get from Microsoft Store" button on our website or search for "AI Studio Pro" in the Microsoft Store app on your computer.
2Choose Your License
- Free: Try the app with 3 chat models and basic image generation
- Basic: Unlimited chat models and cloud AI providers
- Pro: All 9 studios including advanced image generation
3First Launch
The first time you open AI Studio Pro, it will:
- Set up the necessary files (this takes a few minutes)
- Check your GPU to make sure everything works
- Show you the Main Window
💡 Pro Tip: Start with Free
Download the Free version first. Make sure everything works on your computer, then upgrade to Basic or Pro when you're ready. Your settings and projects will carry over.
📖 Understanding the Basics
What is an AI Model?
Think of an AI model like a brain in a file. Different models are good at different things:
- Chat models can have conversations, answer questions, help with writing
- Image models can create pictures from text descriptions
- Voice models can speak text out loud or understand spoken words
What Do All These File Types Mean?
For Chat (Main Window):
- .gguf files: These are language models that run on your computer. Think of them as the "brain" files for chat.
For Images (Image Studio):
- .onnx files: Fast, efficient image models (Free tier). Good for testing and quick generations.
- .safetensors files: High-quality image models (Pro tier). Slower but much better results.
- .gguf files (for images): Flux models that work on your GPU (Pro tier). Professional quality.
💡 Simple Rule
If it ends in .gguf and is over 1GB, it's for chat. If it ends in .safetensors, it's for images. If it ends in .onnx, it's for fast images.
What Are "Prompts"?
A prompt is what you type to tell the AI what to do:
- For chat: "Explain how a car engine works" or "Write me a poem about cats"
- For images: "A sunset over mountains, oil painting style" or "A cozy coffee shop interior"
The more specific you are, the better the results!
💬 Main Window (Chat)
Your First Conversation
The Main Window is where you chat with AI. Here's how to use it:
1Choose a Model
Click the model selector at the top. You'll see:
- Your Local Models: These run on your computer (no internet needed once downloaded)
- Cloud Providers: OpenAI, Claude, Groq (requires internet and API keys)
💡 New User? Start Here
If you haven't downloaded any models yet, you can use cloud providers (Basic/Pro tier) or check the "Where to Get Models" section below to download your first local model.
2Type Your Message
Just type naturally like you're texting a friend. Ask questions, request help, or have a conversation.
3Generate Images (Free Tier and Up)
Want to create an image? Just type a description:
"Create an image: A peaceful garden with cherry blossoms"
The AI will generate an image for you using ONNX models (512x512, basic quality).
Attaching Files
You can attach documents to your conversation:
- Click the paperclip icon
- Select your file (text, PDF, images work best)
- Ask questions about the file
Projects
Projects help you organize conversations by topic. For example:
- "Work Emails" - conversations about work writing
- "Creative Writing" - story ideas and editing
- "Learning Python" - coding help
Click "New Project" to create one, then all your conversations will stay organized together.
🎨 Image Studio (Pro Tier)
Creating Your First Image
Image Studio is where you make professional-quality images. Here's how:
1Choose Your Model
- SD 1.5 ONNX: Fast, good for testing ideas
- SDXL: High quality, balanced speed (~30-60 seconds)
- Flux GGUF: Best quality, slower (~76 seconds on RTX 3060)
2Write Your Prompt
Describe what you want to see. Be specific!
Example:
A cozy coffee shop interior, warm lighting, people sitting at wooden tables,
barista making coffee, plants on shelves, morning sunlight through windows,
professional photography, depth of field
💡 Prompt Writing Tips
- Include style: "oil painting," "professional photography," "anime style"
- Describe lighting: "golden hour," "soft window light," "dramatic shadows"
- Add details: "wooden furniture," "vintage decorations," "modern minimalist"
- Mention quality: "8k quality," "highly detailed," "sharp focus"
3Adjust Settings
- Size: Start with 512x512, try larger if your GPU can handle it
- Steps: 20-30 is good for most images (more = slower but potentially better)
- CFG Scale: 7-8 is a safe default (how closely to follow your prompt)
4Use Negative Prompts
Tell the AI what you DON'T want to see:
blurry, low quality, distorted, watermark, text, signature,
bad anatomy, ugly, overexposed
5Click Generate
Your image will appear in 15 seconds to 2 minutes depending on your model and settings.
How Long Does It Take?
On an RTX 3060 12GB:
- ONNX (SD 1.5): ~15-20 seconds
- SDXL: ~30-60 seconds
- Flux GGUF: ~76 seconds
Faster GPUs (RTX 4070 Ti, 4080, 4090) are much quicker!
💻 Code Studio (Pro Tier)
What is Code Studio?
Code Studio is an AI-powered coding assistant built into AI Studio Pro. It can help you write code, understand existing code, fix bugs, and learn programming.
Getting Started with Code Studio
1Open Code Studio
From the main menu, click "Code Studio" to launch the coding interface.
2Choose a Folder
You can work with an entire project folder. Click "Open Folder" to select your project directory.
3Ask for Help
Type what you need:
- "Help me write a Python function that sorts a list"
- "Explain what this code does" (with code selected)
- "Find the bug in this function"
- "Add error handling to this code"
💡 What Can Code Studio Do?
- Write new code from descriptions
- Explain existing code line-by-line
- Debug and fix errors
- Suggest improvements
- Convert code between languages
- Add comments and documentation
Working with Files
Code Studio can read and edit files directly:
- View files: Click any file in your project to see it
- Edit files: Ask the AI to make changes, and it will update the file
- Create files: "Create a new Python file called helper.py with these functions..."
⚠️ Tip: Be Specific
The more details you give, the better the code. Instead of "make a function," say "make a Python function that takes a list of numbers and returns the average."
🎤 Audio Studio (Pro Tier)
Text-to-Speech (Making the Computer Talk)
Audio Studio can read text out loud using realistic AI voices.
1Open Audio Studio
From the main menu, select "Audio Studio."
2Choose a Voice
AI Studio Pro includes over 50 voices in different languages, accents, and genders. Listen to samples and pick one you like.
3Type or Paste Your Text
Enter the text you want to hear spoken. This could be:
- A book chapter (for creating audiobooks)
- A script or presentation
- Any text you want to hear
4Click Generate
The AI will create an audio file. You can save it as MP3 or WAV.
💡 Creating Audiobooks
Audio Studio has a batch mode for long texts. It can process an entire book chapter by chapter, then combine everything into one audiobook file.
Speech-to-Text (Understanding Spoken Words)
Audio Studio can also transcribe audio files - turning speech into text.
1Select "Speech to Text" Tab
2Upload Your Audio File
Works with MP3, WAV, or other common audio formats.
3Click Transcribe
The AI will convert the speech to text. This is great for:
- Meeting notes
- Interview transcription
- Converting voice memos to text
🎓 Fine-Tuning Studio (Pro Tier)
What is Fine-Tuning?
Fine-tuning teaches AI models your specific style or subject. Think of it like specialized training - you show the AI many examples, and it learns to create similar things.
Training LoRA for Images
LoRA (Low-Rank Adaptation) is a way to teach image models new styles, characters, or concepts.
Example Uses:
- Train on your art style so the AI can create more in that style
- Train on a specific character so you can generate consistent images of them
- Train on your product photos for marketing images
1Prepare Your Images
You need 15-50 images of what you want to teach. For example:
- Training a character: 20 images of that character from different angles
- Training a style: 30 images in that artistic style
2Open Fine-Tuning Studio
Click "Fine-Tuning Studio" from the main menu.
3Create a New LoRA Project
- Click "New LoRA Training"
- Select your images folder
- Give your LoRA a name (like "MyArtStyle" or "ProductPhotos")
4Start Training
Click "Begin Training." This takes 30 minutes to 2 hours depending on:
- How many images you have
- Your GPU speed
- Training settings
⚠️ Training Uses Your GPU
Your computer will be working hard during training. It's best to let it train overnight or when you're not using other programs.
5Use Your Trained LoRA
Once training finishes, you can use your LoRA in Image Studio to create new images in that style!
🗣️ Voice Training (Pro Tier)
Creating Custom Voices
Voice Training lets you create your own text-to-speech voices. You can train the AI to sound like a specific person (with their permission) or create unique character voices.
Preparing Your Audio
You need audio recordings of the voice you want to clone:
- Amount: 30 minutes to 2 hours of speech
- Quality: Clear audio, no background noise
- Content: Natural speech with varied emotion and intonation
⚠️ Important: Get Permission
Only create voices of people who have given you permission to clone their voice. It's illegal and unethical to clone someone's voice without consent.
1Open Voice Training Studio
2Upload Your Audio Files
Voice Training will process your audio and prepare it for training.
3Start Training
Training a voice takes several hours. The process:
- Analyzes the voice characteristics
- Learns pronunciation patterns
- Captures emotional range
4Test Your Voice
Once training completes, you can test it in Audio Studio with any text you want!
💡 Best Results
For the best voice quality, use clean studio recordings. Audiobook narrations work great. Phone recordings or videos with background music produce lower quality.
📚 Knowledge Distiller (Pro Tier)
What is Knowledge Distiller?
Knowledge Distiller helps you extract and organize information from documents. Perfect for research, studying, or organizing large amounts of text.
Extracting Information from Documents
1Upload Your Documents
Drag and drop files into Knowledge Distiller. It works with:
- PDFs
- Word documents
- Text files
- Web pages (copy and paste)
2Choose What to Extract
Tell the AI what you're looking for:
- People & Organizations: Extract all names mentioned
- Dates & Events: Pull out timeline information
- Concepts: Identify key ideas and themes
- Relationships: How things connect to each other
3Process the Documents
The AI will read through your documents and pull out the information you requested.
4View Results
Knowledge Distiller organizes everything into a searchable database that you can explore in Database Browser.
💡 Great For...
- Research projects (extract key points from papers)
- Story writing (organize character relationships)
- Learning (create study guides from textbooks)
- Business (analyze reports and documents)
🗄️ Database Browser (Pro Tier)
What is Database Browser?
Database Browser lets you explore the knowledge you've extracted with Knowledge Distiller. Think of it like a super-powered search for your information.
Browsing Your Knowledge
1Open Database Browser
From the main menu, click "Database Browser."
2Search for Information
You can search in two ways:
- Keyword Search: Look for specific words or phrases
- Semantic Search: Ask questions in natural language
3View Connections
Database Browser shows how information connects:
- Which documents mention the same people
- How concepts relate to each other
- Timeline of events
AI-Powered Enrichment
You can ask the AI to add more information to your database:
- "Add background information about these people"
- "Summarize connections between these events"
- "Find patterns in this data"
💡 Example Use Case
Writing a historical novel? Extract all people and events from research documents into Knowledge Distiller. Then use Database Browser to quickly find "What was happening in 1920?" or "Which characters were in London?"
🎨 LoRA Library (Pro Tier)
What is LoRA Library?
LoRA Library is where you manage all your trained LoRA models. You can view, test, and combine multiple LoRAs to create unique image styles.
Managing Your LoRAs
1Open LoRA Library
From the main menu, select "LoRA Library."
2View Your Collection
You'll see all LoRAs you've trained or downloaded, with preview images for each.
3Test a LoRA
Click any LoRA to generate a test image. This helps you see what it does before using it in Image Studio.
Merging Multiple LoRAs
One of the powerful features is combining multiple LoRAs:
1Click "Merge LoRAs"
2Select LoRAs to Combine
For example:
- Your art style LoRA (60% strength)
- A character LoRA (80% strength)
- A lighting style LoRA (40% strength)
3Adjust Strengths
Control how much each LoRA influences the final result.
4Generate Preview
Create a test image to see how they work together.
5Save the Combination
If you like it, save this merged LoRA for future use in Image Studio!
💡 Getting More LoRAs
You can also download pre-trained LoRAs from websites like CivitAI and HuggingFace. Just drop them in your LoRA folder and they'll appear in LoRA Library!
📥 Where to Get Models
For Chat (Main Window)
The best place to download chat models is HuggingFace.
1Go to HuggingFace
Visit huggingface.co/models
2Search for Models
Good starting models for RTX 3060 12GB:
- Llama 3.2 3B: Fast, good for everyday chat
- Mistral 7B: Excellent quality, medium speed
- Phi-3 Mini: Very fast, good for simple tasks
3Download the .gguf File
Look for files ending in .gguf. If you see "Q4" or "Q5" in the name, those work well on 12GB cards.
4Put It In Your Models Folder
AI Studio Pro looks for models in:
C:\Users\YourName\Documents\AI Studio Pro\Models
Just drop the .gguf file there, restart AI Studio Pro, and it will appear in your model list!
For Images (Image Studio - Pro Tier)
Image models also come from HuggingFace.
1ONNX Models (Built-in for Free/Basic)
These are already included! SD 1.5 ONNX works right away.
2SDXL Models
Search HuggingFace for "SDXL safetensors" or use the built-in model browser in Image Studio.
3Flux GGUF Models
Look for "Flux-dev Q4_K_S" - the Q4 version works on RTX 3060 12GB.
⚠️ File Size Warning
Image models are BIG. SDXL models are 5-7GB, Flux models are 10-12GB. Make sure you have enough hard drive space!
🔧 Troubleshooting
"Out of Memory" Error
Problem: You see an error about running out of GPU memory.
Solutions:
- Try a smaller model (look for "3B" or "7B" instead of "13B" or larger)
- Close other programs that use your GPU (games, video editors)
- For images: Try 512x512 instead of 768x768 or larger
- For images: Reduce the number of steps (try 20 instead of 30)
Images Taking Too Long / Timing Out
Problem: Image generation never finishes or hits the 5-minute timeout.
Solutions:
- The model might be too large for your GPU. Try SDXL instead of Flux, or ONNX instead of SDXL
- Make sure no other programs are using your GPU
- Reduce image size to 512x512
- Reduce steps to 20
Model Not Showing Up
Problem: You downloaded a model but don't see it in AI Studio Pro.
Solutions:
- Make sure it's in the right folder:
C:\Users\YourName\Documents\AI Studio Pro\Models
- Restart AI Studio Pro
- Check that it's the right file type (.gguf for chat, .safetensors or .onnx for images)
- Make sure the download finished completely (check file size matches what HuggingFace says)
Poor Image Quality
Problem: Images look blurry, distorted, or not what you wanted.
Solutions:
- Improve your prompt: Be more specific, add details about style and lighting
- Use negative prompts: Add "blurry, low quality, distorted" to the negative prompt box
- Try a better model: Upgrade from ONNX to SDXL, or from SDXL to Flux
- Increase steps: Try 25-30 steps instead of 20
- Adjust CFG scale: Try 7.5 or 8 for better prompt following
Still Having Problems?
Contact us at support@aistudio-pro.com or check the FAQ on our main website.