Meita.ai packs a lot of tools into a single desktop app. This guide walks through every feature so you know exactly what’s available and when to use it.
AI Metadata Generator
The core feature. Select images, click Generate, and AI analyzes each one to produce titles, descriptions, and keywords optimized for stock platforms.
AI Models
| Model | Best For | Speed | Requires |
|---|---|---|---|
| Super AI Standard | General use, videos | Fast | Included tokens |
| Super AI Premium | Complex scenes, accuracy | Medium | 2 tokens/image |
| Gemini | High volume, cost-effective | Fast | Your API key |
| OpenAI (GPT) | Nuanced descriptions | Medium | Your API key |
| OpenRouter | Access to many models | Varies | Your API key |
Custom Prompts
Leave the prompt field empty to use Meita’s battle-tested default prompt. Or write your own to control the output:
- “Describe this as a food photography image for restaurant marketing”
- “Focus on textures, materials, and interior design elements”
- “Generate keywords suitable for editorial stock photography”
When you write a custom prompt, the “Automatically improve prompt with Meita” option blends your instructions with Meita’s optimized template for the best of both worlds.
Detail Mode (Video)
For video files, enable Detail Mode to upload the entire video to the AI instead of just extracted frames. This produces more accurate metadata but only works with Gemini models and uses more API quota.
Metadata as Context
Enable “Use Metadata as Context” when processing AI-generated images (e.g., from MidJourney). The AI reads the image’s embedded metadata (which often contains the original generation prompt) to produce more accurate keywords.
Filename as Context
Enable “Use Filename as Context” when your files are descriptively named (e.g., sunset-beach-california-2024.jpg). The AI uses the filename to better understand the content.
Only Fill Empty Fields
When re-processing images that already have some metadata, enable this to skip fields that are already filled. The AI only generates content for empty title, description, or tags fields.
Bulk Edit
Access via the Generate dialog → Bulk Edit tab. Edit metadata across multiple images at once:
Add / Remove Tags
Enter comma-separated keywords to add or remove from all selected images. Great for:
- Adding brand terms across your entire batch
- Removing unwanted keywords the AI generated
- Standardizing tags across a collection
Force One Word Tags
Splits multi-word keywords into individual words. “Red sports car” becomes three separate tags: “red”, “sports”, “car”. Useful for platforms that prefer single-word keywords.
Add to Title / Description
Prepend or append text to all titles or descriptions at once. For example, add ” - Stock Photo” to the end of every title.
File Editor
Access via the Generate dialog → File Editor tab. Process your files before or after keywording:
Convert Image Format
Convert between JPEG, PNG, and WebP. Add conversion rules (e.g., PNG → JPEG) and apply to all selected files.
Remove Sound from Video
Strip audio tracks from video files. Useful for stock video platforms that reject content with copyrighted music. Requires FFmpeg (downloaded automatically on first use).
Similar Image Detection
Detect near-duplicate images in your batch. Adjustable threshold from 0% (disabled) to 100% (exact duplicates only). Lower thresholds catch more similar images but take longer to process.
Prompt Generator
A separate tool for generating AI art prompts (MidJourney, Stable Diffusion, DALL-E). Access it from the sidebar.
How It Works
- Optionally add a reference image for visual context
- Enter keywords or a theme (e.g., “cyberpunk city at night”)
- Choose aspect ratio and number of prompts
- Set prefix (
/imagine) and postfix parameters (--style raw) - Click Generate
The AI produces unique, detailed prompts ready to paste into your AI art tool.
Use Cases
- Bulk-generate prompts for stock image creation
- Explore variations on a theme
- Generate prompts from existing images you want to recreate
Remove Background
AI-powered background removal that runs locally on your machine. Access it from the sidebar.
Modes
- Basic (~40MB model) — Faster, good for simple backgrounds
- Medium (~80MB model) — Better quality, handles complex edges
The model downloads once and is cached for future use. All processing happens on your GPU — no data leaves your computer.
Workflow
- Add images (click, drag & drop, or use the empty state button)
- Click Remove Background and select mode/quality
- Review results — click any image to preview full-size
- Revert individual images if the result isn’t satisfactory
- Download All to export, or download individually
Your images persist across app restarts — no need to re-add them each session.
Export System
Meita exports to 8+ stock platforms with platform-specific formatting:
Supported Platforms
- Adobe Stock — CSV with category IDs
- Shutterstock — CSV with category mapping
- Canva — Standard CSV format
- Freepik — Semicolon-delimited CSV
- Dreamstime — CSV format
- Vecteezy — CSV format
- Envato — CSV format
- Pond5 — CSV format
Export Options
- Inject Metadata to Files — Writes EXIF data (title, description, keywords) directly into image files using ExifTool
- Rename Files on Export — Auto-rename based on generated titles
- Grouped Export — Organize output into subfolders (defects, faces/hands, animals)
- File Export — Copy/move files to the export folder alongside CSVs
Settings
Output Rules
| Setting | What It Does |
|---|---|
| Target Title Length | Character count the AI aims for |
| Maximum Title Length | Hard limit — trims if exceeded |
| Separate Title & Description | Generate distinct title and description fields |
| Target/Max Description Length | Same controls for descriptions |
Keywords
| Setting | What It Does |
|---|---|
| Target Count | Number of keywords to generate (default: 50) |
| Force Limit | Remove keywords exceeding the target count |
| Minimum Count | Keep generating until this minimum is reached |
| Force One Word | Split multi-word keywords |
| Banned Keywords | Comma-separated list of keywords to always exclude |
Runner
| Setting | What It Does |
|---|---|
| Workers | Parallel generation tasks (default: 8) |
| Delay | Seconds between API calls — increase for free API tiers |
| Auto Retry | Automatically retry failed generations (up to 3 times) |
| Double Filter | Run keyword filtering twice for cleaner results |
Status Indicators
The metadata generator shows real-time status for each image:
- Idle — Not yet processed
- Generating — Currently being processed by AI
- Scheduled — Queued, waiting for a worker
- Success — Metadata generated successfully
- Failed — Generation failed (click to see error)
- Incomplete — Missing title, description, or tags
- Defect — AI detected possible defects, logos, or watermarks
- Faces/Hands — AI detected faces or hands (may need model releases)
- Animal — AI detected animals (some platforms have special requirements)
Use the filter dropdown or status pills to quickly find and fix issues before exporting.