Adobe Stock Metadata Generator
Upload a folder of images and Metadata Studio writes the Adobe Stock fields for you: a tight title, a plain description and an ordered keyword list. Everything stays editable before the CSV leaves your browser.
The five Adobe Stock columns
The Adobe export is deliberately small. Every row carries the same five fields Adobe Stock expects, in this order:
| Column | What goes in it |
|---|---|
| Filename | The original name of the uploaded image. |
| Title | A single descriptive line of up to 70 characters. |
| Keywords | An ordered list of 20 to 49 keywords, most relevant first. |
| Category | The Adobe Stock category assigned to the image. |
| Releases | The release reference for assets that need a model or property release. |
The download is named Adobe_Stock_CSV_<DD-MM-YYYY>.csv and holds up to 2,000
rows, so a single export can cover a full day of shooting. The exact header spelling and column
order are documented on the templates and column reference, where you can
also download a sample CSV to test your own importer against.
Character limits that shape the copy
Adobe Stock enforces two hard ceilings, and Metadata Studio builds both into generation rather than trimming the text afterwards. A title is capped at 70 characters and a description at 150. The title is the field buyers and the search index lean on most, so the model is told to lead with the subject — what the image visibly shows and the setting it plausibly belongs to — without keyword stuffing.
Keywords are capped at 49 entries, with 20 to 40 the range that genuinely helps. Fewer than 20 leaves relevant searches uncovered, while a padded list of near-synonyms dilutes the terms that matter. The generator returns the list already ordered by relevance and reports the keyword count for each image, so thin rows are easy to spot before you export.
What makes the Adobe prompt different
Metadata Studio runs an Adobe-specialist prompt on the Adobe tab. Instead of a generic caption, the model is asked for the shape Adobe Stock reviewers expect: literal, commercial and free of editorial commentary on commercial assets. It is instructed to describe only what is visible and to refuse guesses about locations, names or events it cannot actually see.
The reply is normalised before it reaches your table. Typographic smart quotes are folded to plain quotes, banned and hype words are stripped out, and singular and plural duplicates are merged into a single keyword. What is left reads cleanly and imports without surprises.
You can read more about the model and its honest limits on the AI generator overview, or walk through the six-step workflow on how it works.
Category and releases
Adobe Stock groups every asset into one category and, where relevant, one or more releases. The Category and Releases values are yours to set: the generator fills the creative fields, while the category you choose and any release reference you hold stay under your control. Because a release is a legal document tied to a specific shoot, the tool never invents one.
If you are unsure how a category maps to your work, compare a few of your existing accepted assets and reuse the same value. Consistency across a series usually matters more than finding a technically perfect label for a single frame.
Getting the keyword order right
Keywords do the heavy lifting on Adobe Stock, and the generator returns them ranked from the most relevant term down to the loosest. Treat that order as a draft: keep the specific subject and action words at the top, and cut anything that only repeats a term you already have. If you want a deeper method, the stock keyword generator page covers pruning and normalization in detail.
Review, edit and export
Generation is only the first pass. Every field is editable inline, and each row shows the provider and model that produced it. If one image comes back weak, use Regenerate on that item instead of rerunning the batch. The batch controls let you stop a run, retry only the failures and clear the queue when you are done.
When the table looks right, export from the toolbar and the Adobe_Stock_CSV file is
written in your browser. Nothing is uploaded to us and nothing is stored. The same
bring-your-own-key model that keeps your API key private also means your images and metadata stay on
your machine.
A repeatable Adobe Stock routine
- Shoot and cull as usual, then keep only the frames you would submit.
- Add an API key, choose a vision model and queue the folder.
- Let the batch run two or three images at a time, then retry any failures.
- Read each title as a buyer would, shorten the weak ones and fix the keyword order.
- Set the Category, attach Releases where needed and export the CSV.
Used this way, the tool removes the blank-page problem without taking the editorial decisions away from you. Metadata Studio is independent and not affiliated with Adobe, so the final call on what an asset says always rests with you.
Adobe Stock metadata questions
What columns does the Adobe Stock CSV contain?
The export uses the five Adobe Stock columns in order: Filename, Title, Keywords, Category and Releases.
How many keywords should I use for Adobe Stock?
Aim for 20 to 40. The tool allows up to 49, and fewer than 20 leaves obvious searches uncovered, so 20 to 40 is the practical sweet spot.
What is the longest title I can export?
Titles are capped at 70 characters and descriptions at 150. Both limits are enforced while the metadata is generated, so you rarely need to shorten anything by hand.
Does the generator also work for Shutterstock?
Yes. Switch the marketplace toggle to Shutterstock for its own column set, auto-classified categories and Editorial, Mature Content and Illustration flags. See the Shutterstock metadata generator page for details.
Write your next Adobe Stock CSV with AI
Add your own provider key, queue a folder and edit the results before you export. No account.