Tools
Metadata Studio is a single page that gives you five working tools for stock metadata. They share your provider key, your uploaded images and your editable results, so you can move from generation to export without leaving the tab.
How the five tools fit together
They run against the same uploaded images and the same provider key, so a typical session moves in one direction: generate metadata, classify Shutterstock categories where you need them, refresh the model list when a provider changes, then export the CSV for the marketplace you sell on. Nothing is hidden behind a second account, and every intermediate result can be corrected by hand before it leaves your browser.
Metadata Generator
The core tool. Upload raster images and a vision model writes a title of up to 70 characters, a description of up to 150 characters and 20 to 49 keywords, using only what is visibly in the picture. Results are normalised — smart quotes fixed, filler phrases removed, singular and plural duplicates merged — and every field stays editable inline.
Run it as a batch with two or three images at a time, stop or cancel at any point, and retry only the failures. The AI metadata generator guide covers what each field contains, and the features page lists every control.
Shutterstock Category Classifier
On the Shutterstock tab, each image is classified against the official 26-name Shutterstock taxonomy with a dedicated vision call. It returns one or two canonical category names that are ready for the export, and they remain editable per item. This is a separate call from metadata generation, so a classification failure never blocks or alters the title, description or keywords.
If an item still lacks a category, the classifier back-fills it, and a row with no valid category is rejected at export instead of being written silently. See the Shutterstock metadata guide for the full taxonomy and flag rules.
Live Model Discovery
Model lists change, so Metadata Studio can pull the live list for your key with a Refresh models button instead of trusting a hardcoded set. Each result marks which models your key can use, and vision-capable models are labelled. You can also type any model id your account exposes.
This is how you recover from a retired model or a provider error without editing code. The features page explains discovery alongside key rotation and cooldowns.
Adobe Stock CSV Export
Export the Adobe Stock format with the exact columns
Filename,Title,Keywords,Category,Releases, up to 2,000 rows. The download is named
Adobe_Stock_CSV_<DD-MM-YYYY>.csv, with RFC4180 quoting so commas inside a keyword
list survive the round trip.
The Adobe Stock metadata guide covers the columns and the Category field, and the templates page has a downloadable sample CSV header with one example row.
Shutterstock CSV Export
Export the Shutterstock format with the columns
Filename,Description,Keywords,Categories,Editorial,Mature Content,Illustration, up to
2,000 rows, as Shutterstock_CSV_<DD-MM-YYYY>.csv. Keywords are deduped
case-insensitively, categories are the canonical names joined by a comma, and the three flags emit
literal Yes or No values in header order.
At least one valid category per row is required; otherwise the export returns an error that names the offending files. The client also warns, without blocking, when a row has fewer than 20 keywords or is marked Editorial, because editorial assets need a capture date and location in the description. See the Shutterstock metadata guide and the templates page.
Open the Shutterstock export Shutterstock guide CSV templates
What you need
A modern browser and one API key from Gemini, OpenRouter or Groq. There is no account, no pricing and no install. Your key stays in this browser, so the tools work the same whether you are on your own machine or a shared one, and clearing browser storage removes the key. For the full behavior list, see the AI generator guide and the frequently asked questions.
Use all five tools with one key
No account, no pricing. Add a provider key and the tools are ready in your browser.