Aerial and drone files are bought for where they are and what they can be placed behind, not for the objects visible in frame. A buyer searching for drone footage usually already knows the subject and needs a specific city, coastline, altitude or camera move. So the keywords that earn downloads on aerial work are location, vantage point and use case, and the ones that waste slots are the generic nouns a vision model finds first.
What a buyer is actually searching for in an aerial file
Ground-level stock gets found by subject. Aerial stock gets found by place and purpose. The designer looking for a drone shot of a highway interchange is rarely searching for "road" — they want a particular kind of interchange, from a particular height, in a region their client operates in.
That changes what the first ten keywords have to carry. On a ground-level photo of a cafe, the words cafe, coffee, table and laptop are all load-bearing. On an aerial of the same street, cafe is almost worthless, while city block, rooftops, urban density, top down and the name of the city do the work.
The second difference is that aerial files are often bought as backgrounds and plates. A top-down of a forest canopy is as likely to end up behind a headline as it is to illustrate forestry, so terms describing the file's usability — copy space, negative space, seamless, top down, flat — carry real search weight that they would not carry on a portrait.
Location is the keyword, and agencies have rules about it
Shutterstock keeps the location field optional but tells contributors to be as specific as they can when they fill it, and says neighbourhoods, towns or cities are preferred where possible. Its published metadata standards format the examples as "Paris, France" and "London, UK", and they also set the keyword range at a minimum of 7 and a maximum of 50.
That specificity is not a courtesy. Aerial searches are regional by nature, and a file tagged only "Europe" will never surface for the client who needs that exact stretch of coast. Fill the location field, then repeat the useful part of it in the keywords, because the two are indexed separately.
If the file is editorial, the rules get stricter. Shutterstock requires documentary editorial descriptions to include the day, month and year the content was taken plus the geographic location, and it rejects under Metadata - Editorial Requirements when the year or location is missing or inaccurate. Its example caption uses a dateline format that opens with the city, the country and the date before the description.
Drone work runs into this more often than most categories, because a lot of aerial footage of real places carries real signage, real vehicles and real people, and belongs on the editorial side rather than the commercial one.
Vantage point, altitude and camera move belong in the metadata
Aerial buyers filter by how the camera behaved, and most contributors never write it down. The vocabulary worth carrying falls into four groups:
- Vantage: aerial, aerial view, top down, overhead, bird's eye view, high angle, oblique, low altitude.
- Movement, for footage: orbit, reveal, pull back, push in, fly over, tracking, ascending, descending, parallax.
- Framing and usability: copy space, negative space, panoramic, wide shot, establishing shot, symmetry, pattern.
- Conditions that drive regional and seasonal demand: golden hour, blue hour, fog, snow cover, autumn colour, low tide.
For footage especially, establishing shot and reveal describe the edit the clip will sit inside, which is how video editors search. A clip keyworded only by its subject gets found by nobody who is looking for a sequence.
Shutterstock's range of 7 to 50 keywords sets the outer bounds, and on aerial files the practical risk sits at the top of that range rather than the bottom. Plausible terms are easy to generate for a top-down frame, and padding towards 50 with near-synonyms is exactly the metadata-to-subject mismatch that gets flagged in review.
The release question decides what you can name
Before any of the keywording matters, work out which side of the commercial and editorial line a flight sits on, because a drone sees a great deal more property than a ground camera does.
Adobe requires a property release for any recognisable private property, artwork or location, and its own examples include identifiable exteriors or interiors of private homes and buildings, famous landmarks, historic locations and modern architecture, plus properties that have photography policies such as stadiums, museums, concert venues and amusement parks. The release has to be signed by the property owner, or by a corporate representative where a company owns it. Adobe's submission guidelines put the practical version more bluntly: avoid famous landmarks or buildings that require property releases.
A single aerial pass over a residential neighbourhood can clear several of those categories at once. Adobe's advice when you are unsure is either to obtain the release or to edit the image so the protected element is gone, and on aerial work a crop or a reframe is often genuinely enough, because the protected building is rarely the reason the file sells.
This has a direct keywording consequence. If you cannot release the landmark, do not name it. Naming an unreleased landmark in the title or keywords is the fastest route to a metadata rejection, and a file that would have sold as a coastal city skyline at dusk gets turned down because it was keyworded as the specific tower in the middle of it. The mechanics of attaching releases to an upload batch are worth settling before a flight rather than after it.
Where the generic nouns waste your slots
Automatic tagging is unusually weak on aerial imagery, because vision models were trained predominantly on ground-level framing. Shown a top-down of farmland, a model reliably returns green, field, texture, pattern and background. Every one of those is true of the frame, and not one of them is what a buyer types.
The same aerial frame, keyworded two ways
A top-down farmland frame as a vision model tags it against the same frame keyworded for aerial buyers, with the reason each term earns or wastes a slot. The location rules come from Shutterstock's published metadata standards.
A pre-upload pass for aerial batches
Aerial batches come off a single flight, which means they share a location, a date, a release status and a legal profile. That is an advantage rather than a chore: the decisions get made once per flight instead of once per file, and the per-file work drops to vantage, movement and subject.
Aerial metadata checks before you upload
The once-per-flight decisions plus separate checks for commercial stills, editorial files and footage, drawn from Adobe's property release overview and submission guidelines and Shutterstock's contextual metadata standards.
Which tools actually help on aerial files
Most keywording tools read the frame, and on aerial work the frame is the least informative part of the file. What matters is whether a tool lets you set location, vantage and release status once and apply it across a whole flight. The companion piece on tools for aerial and drone content goes deeper on each of these; the short version follows.
- Xpiks is a desktop tool you drive yourself, with strong batch editing across metadata fields. Its own pricing page lists a free Basic tier capped at 15 files per session, a one-time €49 Pro licence and a €99 yearly Pro+ that adds reverse image search and cloud upload. Best when you want manual control over every field and would rather not hold a subscription.
- StockSubmitter is the long-standing multi-agency submitter and the tool most often named when people ask about distribution. Its real strength is breadth of agency support. Its weakness for this category is that metadata quality remains entirely your job, so the aerial-specific thinking still has to happen somewhere.
- PhotoTag.ai and other vision-first taggers are fast and genuinely good on ordinary subject photography, where naming what is in frame is most of the task. On aerial files they tend to produce the generic-noun sets described above, so they are most useful as a first pass you then edit rather than as the finished metadata.
- Wirestock and Blackbox are managed pipelines: they handle submission in exchange for a share of royalties. Reasonable if you would rather not run a workflow at all, and expensive measured across the full life of a portfolio, since the share does not stop when the work does.
- Agency-native tools, such as Adobe's own keyword suggestions and Shutterstock's uploader, are free and know their own rules better than anything else does. They cover one agency each, though, so a multi-agency aerial portfolio means re-keywording per platform.
- Rastock.ai sits on the metadata-plus-delivery side. It generates titles, descriptions and keywords against each agency's own published rules, holds banned and mandatory keyword lists per platform, and delivers over FTP and SFTP to 10+ agencies with per-agency upload status. There is no revenue share and no ownership claim, and IPTC and CSV export stay available, which matters if you ever want to leave.
When to use which
If you shoot a few hundred aerial files a year and submit to one or two agencies, the agency's own uploader plus a careful manual pass is enough. No tool pays for itself at that volume, and the per-flight decisions are few enough to hold in your head.
If you fly regularly and submit to several agencies, the bottleneck is not writing keywords. It is applying one flight's location, release status and vantage vocabulary consistently across agencies whose field rules all differ. That is where a batch-oriented tool earns its cost, whether it is a desktop application you run yourself or a delivery pipeline that pushes to everything at once.
If you would rather not touch metadata at all and you accept a permanent cut of revenue in exchange, a managed pipeline is the honest answer. For some contributors that trade is the right one, and it is worth saying so plainly rather than pretending otherwise.
If you want the pipeline without the revenue share, that is the gap Rastock.ai is built for. You can see what the tool does or look at the plans — the trial runs 14 days and does not ask for a card.
Frequently asked questions
How many keywords should an aerial stock photo have?
Shutterstock requires a minimum of 7 keywords and allows a maximum of 50. On aerial files the realistic risk is over-tagging rather than under-tagging, because plausible synonyms are easy to generate. Keywords must be relevant and applicable to the content, and metadata that does not match the depicted subject can be rejected, so stop when the terms stop being true.
Do I need a property release for a drone photo of a building?
If the building is recognisable private property, yes. Adobe's examples of content needing a property release include identifiable exteriors and interiors of private homes and buildings, famous landmarks, historic locations and modern architecture. The release must be signed by the property owner or a corporate representative. Adobe's guidance when you are unsure is to obtain the release or edit the protected element out.
Should the location go in the keywords or the location field?
Both, because they are indexed separately. Shutterstock keeps the location field optional but asks contributors to be as specific as they can, preferring neighbourhoods, towns or cities, with examples formatted as Paris, France. Then repeat the useful part of the place name in your keywords so the file also surfaces in ordinary keyword search.
Why do AI taggers do badly on aerial images?
Vision models are trained predominantly on ground-level framing, so a top-down frame returns the safest literal descriptors: green, field, pattern, texture, background. Those are accurate and commercially useless. Aerial buyers search by place, vantage point, altitude and camera move, none of which is visible as an object in the frame.
What has to be in an editorial caption for drone footage?
Shutterstock requires documentary editorial descriptions to include the day, month and year the content was taken plus the geographic location where it was shot. A missing or inaccurate year or location triggers a Metadata - Editorial Requirements rejection. Its own example caption uses a dateline format beginning with the city, country and date.
Is aerial footage keyworded differently from aerial stills?
The location and release logic is identical, but footage adds camera movement, which editors search on directly. Terms like orbit, reveal, pull back, fly over, descending and establishing shot describe the edit the clip will sit in. A clip tagged only by its subject will not be found by an editor looking for a sequence.