Skip to main content
The Image Studio is a dual-mode AI image generation workspace that automatically switches between text-to-image (T2I) and image-to-image (I2I) workflows based on whether you’ve uploaded a reference image.

​
Dual Mode Intelligence

Image Studio features smart mode switching that adapts the entire interface based on your workflow:

Text-to-Image Mode

Generate images from scratch using text descriptions. Access 50+ models including Flux, Nano Banana 2, Seedream 5.0, and Midjourney.

Image-to-Image Mode

Transform existing images with AI. Upload a reference image to automatically switch to 55+ editing models like Kontext, Seedream Edit, and Upscaler.

​
How Mode Switching Works

  1. Start in T2I Mode — The studio defaults to text-to-image generation with 50+ creative models
  2. Upload triggers I2I Mode — Click the upload button and select an image to automatically switch to image-to-image models
  3. Clear returns to T2I — Remove the reference image to return to text-to-image mode
The model selector, aspect ratio options, and quality controls all update automatically when you switch modes.

​
Generating Images from Text

1

Enter Your Prompt

Click into the prompt textarea and describe the image you want to create. The prompt is required for text-to-image generation.
Example: "A futuristic cityscape at sunset with flying cars and neon signs"
2

Select Your Model

Click the model button to browse 50+ text-to-image models:
  • Nano Banana 2 — Google Gemini-powered, supports 1K/2K/4K resolution
  • Flux Dev — High-quality open model with fine detail
  • Seedream 5.0 — ByteDance model with quality presets (basic/high)
  • Ideogram v3 — Excellent typography and composition
  • Midjourney v7 — Professional creative generation
  • GPT-4o — OpenAI’s image generation model
Use the search bar to quickly find models by name.
3

Choose Aspect Ratio

Select from available aspect ratios (varies by model):
  • 1:1 — Square (Instagram, profile pictures)
  • 16:9 — Landscape (desktop wallpapers, YouTube thumbnails)
  • 9:16 — Portrait (mobile wallpapers, stories)
  • 4:3, 3:4, 21:9 — Additional formats
The aspect ratio dropdown automatically shows only formats supported by your selected model.
4

Set Quality/Resolution (Optional)

For models that support quality or resolution settings, a quality button appears:
  • Nano Banana 2: Choose 1K, 2K, or 4K
  • Seedream 5.0: Select basic or high quality
  • Other models: Resolution or quality presets as available
The quality button only appears for models that support these settings. Most models use fixed output quality.
5

Generate

Click the Generate ✨ button to start. The interface will:
  • Hide the hero section
  • Show a loading spinner
  • Poll the API until generation completes
  • Display your result in the canvas area

​
Image-to-Image Transformations

1

Upload a Reference Image

Click the upload button (📤 icon on the left of the prompt bar) to open the reference image picker. You can:
  • Select from upload history — Click any previously uploaded image
  • Upload a new image — Click “Upload new” and choose a file from your device
Once you select an image, the studio automatically switches to I2I mode.
2

Studio Switches to I2I Mode

When you upload an image, the studio automatically:
  • Switches to 55+ image-to-image models
  • Updates the prompt placeholder to “Describe how to transform this image (optional)”
  • Shows quality/resolution controls for compatible I2I models
  • Enables multi-image support if you select a compatible model
3

Enter Transformation Prompt (Optional)

For image-to-image generation, the prompt is optional. Many models can interpret the reference image without text:
Example: "Make this photo look like a watercolor painting"
Example: "Remove the background"
Example: "Upscale to 4K resolution"
Leave the prompt empty to let the model decide how to transform the image.
4

Select I2I Model

Browse 55+ image-to-image models:
  • Nano Banana 2 Edit — Supports up to 14 reference images, 1K/2K/4K output
  • Flux Kontext Pro — High-quality image editing
  • Seedream 5.0 Edit — Natural language style transfer
  • Seededit v3 — Precise editing control
  • Upscaler — Resolution enhancement
  • Background Remover — Clean background removal
5

Generate Transformation

Click Generate ✨ to start the transformation. The API will:
  • Upload your reference image (if not already uploaded)
  • Send the image URL and optional prompt to the model
  • Return the transformed result

​
Supported Models

​
Text-to-Image Models (50+)

ModelKey Features
Nano Banana 2Google Gemini 3.1 Flash · 1K/2K/4K resolution · Google Search enhancement · aspect ratio auto
Flux DevHigh-quality open model · 128-2048px (divisible by 64) · Fine detail
Seedream 5.0ByteDance · Quality basic/high · 8 aspect ratios · up to 4K
Ideogram v3Excellent typography · Precise composition · Text rendering
Midjourney v7Professional creative generation · Artistic styles
GPT-4oOpenAI’s image generation · Natural language understanding

​
Image-to-Image Models (55+)

ModelMax ImagesKey Features
Nano Banana 2 Edit14Google Gemini · 1K/2K/4K · Multi-image support
Flux Kontext Dev I2I10High-quality context-aware editing
Seedream 5.0 Edit1ByteDance · Natural language style transfer · Quality basic/high
GPT-4o Edit10OpenAI’s image editing model
Kling O1 Edit Image10Kuaishou’s editing model

​
Quality and Resolution Controls

When you select a model that supports quality or resolution settings, a quality button appears in the control bar.

​
Resolution Presets

Models like Nano Banana 2 offer resolution presets:
  • 1K — 1024px output (fast, lower cost)
  • 2K — 2048px output (balanced quality/speed)
  • 4K — 4096px output (highest detail, slower)

​
Quality Presets

Models like Seedream 5.0 offer quality levels:
  • basic — Faster generation, lower detail
  • high — Slower generation, enhanced detail
The quality/resolution button is hidden by default and only appears when you select a compatible model. This keeps the interface clean and focused.

​
Canvas & Generation History

When your generation completes, the Image Studio transitions to the canvas view:

Result Display

Your generated image appears in the center with a glassmorphism border and interactive glow effect.

Action Buttons

  • ↻ Regenerate — Run the same generation again
  • ↓ Download — Save image as JPEG
  • + New — Return to prompt bar for a new generation

​
History Sidebar

All generations are saved to a history sidebar on the right side of the screen:
  • Persistent storage — Up to 50 generations saved in browser localStorage
  • Thumbnail previews — Quick visual reference of past generations
  • Click to load — Click any thumbnail to view full resolution
  • Download from history — Hover over thumbnails to download past generations
The history sidebar automatically appears after your first generation and persists across browser sessions.

​
Multi-Image Input

Some image-to-image models support multiple reference images (see Multi-Image Input for details). When you select a multi-image model like Nano Banana 2 Edit, the upload picker switches to multi-select mode:
  1. Click the upload button
  2. Select multiple images (checkboxes appear with order numbers)
  3. Click “Use Selected” to confirm
  4. The trigger button shows a count badge (2, 3, etc.)
Images are sent to the model in the order you select them. The order can affect the result for some models.

​
Keyboard Shortcuts

  • Enter (in textarea) — Generate image
  • Escape — Close dropdowns and panels
  • Tab — Navigate between controls

​
Tips for Best Results

  • Be specific — “A red sports car on a mountain road at sunset” is better than “a car”
  • Include style keywords — “photorealistic”, “oil painting”, “minimalist”, “cinematic”
  • Add detail — Lighting, composition, mood, colors, textures
  • Use commas — Separate concepts for clarity
  • Nano Banana 2 — Best for Google Search-enhanced generation and high resolution
  • Flux Dev — Best for fine detail and photorealism
  • Seedream 5.0 — Best for fast generation with quality presets
  • Ideogram v3 — Best when you need text in the image
  • Midjourney v7 — Best for artistic and creative results
  • Upload high-quality references — Better input = better output
  • Use descriptive prompts — Even optional prompts help guide the transformation
  • Try different models — Each model interprets images differently
  • Experiment with quality settings — “high” quality costs more but produces better results

​
API Integration

Under the hood, Image Studio uses the Muapi.ai unified API: 
// Text-to-Image generation
const result = await muapi.generateImage({
  model: 'nano-banana',
  prompt: 'A futuristic cityscape',
  aspect_ratio: '16:9',
  resolution: '2k'
});

// Image-to-Image generation
const result = await muapi.generateI2I({
  model: 'nano-banana-edit',
  image_url: 'https://...',
  images_list: ['https://...'], // for multi-image models
  prompt: 'Make it look like a painting',
  aspect_ratio: '1:1',
  resolution: '2k'
});
See the API Reference for complete documentation.

Multi-Image Input

Learn how to use multiple reference images

Upload History

Reuse previously uploaded images

Generation History

Browse and download past generations

Build docs developers (and LLMs) love

Get started for freeTalk to us