Short answer: Xpiks, StockSubmitter, Wirestock, Blackbox and Rastock AI all handle photos and video, but they solve different halves of the problem. Xpiks and StockSubmitter are desktop tools you drive yourself. Wirestock and Blackbox are managed pipelines that take a share of royalties. Rastock AI generates metadata and delivers over FTP without a revenue share.
The wider problem is that most AI keywording tools were built for still images and had video added later. That distinction is invisible on a pricing page and obvious the first time you run a clip through one. This piece covers what actually differs, which tools cover both media types, and how to test a tool before you trust a footage batch to it.
Why video metadata is a different job
The cheapest way to add "video support" to an image tagger is to extract one frame and run the existing model on it. The output is technically a description — it is just a description of a photograph that happens to have been taken from a clip.
That matters because footage buyers search for things a single frame cannot show:
- Motion. A still frame shows a woman on a beach. The clip shows her walking. "Walking" is the search term.
- Camera technique. Drone, handheld, gimbal, dolly, static tripod, slow motion, timelapse. Editors search by these constantly and a frame contains none of them.
- Pacing and usability. Whether a clip loops cleanly, whether it has space for titles, whether the action completes inside the duration.
- Technical attributes. Resolution, frame rate and codec are part of the buying decision in a way that megapixels rarely are for stills.
A tool that samples several frames across a clip can infer at least the first two. A tool that samples one cannot, no matter how good the underlying model is. This is the single most useful thing to test, and it takes one clip and about two minutes.
The tools that cover both photos and video
Five tools come up repeatedly in contributor discussions. They are not interchangeable — each is strong at a different stage of the pipeline.
Xpiks
Best for: contributors who want full manual control and pay nothing. Xpiks is free and open source, runs on the desktop, and handles photos, vectors and video.
Limitation: it is a keywording and upload tool rather than an AI-first one. You get precise control and no subscription; you do more of the thinking yourself. Development pace depends on volunteer time, which is the honest trade-off with any open-source tool.
StockSubmitter
Best for: submission automation across many agencies at once. It is a Windows desktop application that fills metadata, translates it, stores it and submits to dozens of destinations, for both stills and footage. Contributors who upload to a long list of agencies tend to rate its automation highest.
Limitation: Windows-only, and the interface reflects its age. It is a submission engine first; metadata quality is largely what you bring to it.
Wirestock
Best for: contributors who would rather not run a pipeline at all. You submit once, Wirestock handles keywording, distribution and agency relationships inside its own managed system, for photos and video alike.
Limitation: it keeps a percentage of your royalties, and you give up direct control of how each agency sees your work. For some contributors that is a fair trade for the hours saved; for high-volume sellers the cumulative cost is the deciding factor.
Blackbox
Best for: footage-first contributors. Blackbox grew up around stock video rather than stills, and its conventions reflect that. If video is the majority of what you shoot, it speaks your language by default.
Limitation: it is also a managed model with a revenue share, and it is less useful if your library is mostly photographs.
Rastock AI
Best for: mixed photo and video libraries where you want the metadata and the delivery in one pass but do not want a middleman on your royalties. It generates titles, descriptions and keywords for both media types, embeds IPTC and XMP, and delivers over FTP or SFTP to 10+ agencies with per-destination status tracking. Banned and mandatory keyword rules apply across a whole batch, and a run can be rolled back if it comes out wrong. There is no revenue share, no ownership claim, and metadata exports as IPTC, XMP or CSV.
Limitation: it is newer than the incumbents, so the community around it is smaller and there are fewer third-party guides to lean on. It also assumes you want to keep your own agency accounts — if you specifically want someone else to manage those relationships, a managed pipeline fits better.
When to use which
- Mostly video, low volume, no budget → Xpiks. Free, handles footage, and at low volume the manual effort is affordable.
- Mostly video, high volume, want it handled → Blackbox or Wirestock. You pay in royalty share instead of time.
- Long agency list, metadata already sorted → StockSubmitter. Its strength is breadth of destinations, not authoring.
- Mixed photo and video, want to keep 100% of royalties → Rastock AI, or Xpiks plus your own FTP discipline if you would rather not pay a subscription.
- One agency, a handful of files a month → none of the above. The agency's own uploader and keyword suggestions are enough, and any of these tools is more infrastructure than the problem needs.
The agency side: where footage metadata actually lands
A tool can only be as good as the destination's rules allow. Video destinations differ from image destinations in ways that catch people out, and our agency comparison hub goes through each platform in detail. The short version for footage:
- Several agencies accept video through the same contributor FTP or SFTP as images, but not necessarily into the same folder. Alamy, for example, separates uploads into named folders by content type, so a root-level upload silently goes nowhere useful.
- Some video-first platforms qualify contributors before enabling bulk upload at all, so an FTP path that exists in the documentation may not yet exist for your account.
- Subscription-pool platforms pay on usage rather than per licence, which changes which clips are worth uploading — long, versatile, loopable footage earns differently there than a single dramatic five-second shot.
None of this is metadata work exactly, but it determines whether good metadata ever gets read.
How to test a tool on footage in ten minutes
Marketing pages will not tell you whether a tool understands motion. This will:
- Pick one clip with obvious movement. A drone push-in, a slow-motion pour, someone walking through frame.
- Read the generated keywords for any verb. If every term is a noun or an adjective, the tool read one frame.
- Look for camera technique. Does "aerial", "slow motion" or "handheld" appear when it should?
- Interrupt an upload deliberately. Video files are large enough that transfers fail. Does it resume, or start over?
- Check what "done" means. A single success message across six agencies is not status tracking. You want to know which destination accepted what.
- Export the metadata. If it will not come out as CSV or XMP, your library is hostage to the tool.
A workable setup for a mixed library
If you shoot both, the mistake is running two disconnected pipelines and reconciling them by hand. Our workflow hub covers the full sequence, but the footage-specific version is short: keep one metadata source of truth, write it into the files as IPTC and XMP so it travels with them, and treat delivery as a tracked job rather than a drag-and-drop.
Embedding matters more for video than for stills, because footage often changes hands — an editor pulls a clip into a project, exports a proxy, hands it on. Metadata written inside the file survives that; metadata that only lives in a spreadsheet does not. If you are still choosing between tools, our wider keywording tool comparison covers the stills side in more depth, and Rastock's plans are priced by upload volume rather than by media type, so mixed libraries are not penalised.
The honest summary: there is no tool that is best at both halves for everyone. If footage dominates, use something built around footage. If your library is mixed and you want to keep your royalties and your agency relationships, a metadata-plus-delivery tool without a revenue share is the shorter path. Either way, test with one real clip before you commit a batch — the difference between a tool that watches a video and one that photographs it shows up immediately.
Frequently asked questions
Can AI metadata tools tag video as well as photos?
Some can, but fewer than the marketing suggests. Most AI taggers analyse a single extracted frame, which produces a description of a still image rather than of motion. Tools built for footage sample several frames across the clip and add movement, camera technique and pacing to the keyword set. Before committing, run one real clip through a trial and check whether the output mentions any motion at all.
Why does video need different keywords from photos?
Because buyers search differently. Someone licensing a still searches for a subject and a mood. Someone licensing footage searches for a subject plus an action and often a camera technique — "drone shot coastline", "slow motion pouring coffee", "timelapse city traffic". If your keyword set only describes what is in the frame, you miss the half of the query that describes what the frame is doing.
Which tools support both photos and video?
Xpiks handles photos, vectors and video and is free. StockSubmitter automates submission for both across many agencies from a Windows desktop app. Wirestock and Blackbox accept both but operate as managed pipelines that take a share of royalties. Rastock AI generates metadata for both and delivers over FTP or SFTP without a revenue share. Each has a different trade-off — see the comparison above.
Do stock agencies accept video over the same FTP as photos?
Often yes, but not always to the same folder or under the same rules. Adobe Stock accepts video through its contributor SFTP alongside images. Alamy separates uploads into named folders by content type. Some video-first platforms qualify contributors before enabling FTP at all. Check each destination's folder structure before assuming a single upload path covers everything.
Is it worth using one tool for both, or two specialists?
It depends on your ratio. If footage is an occasional side output next to a large stills library, one tool that covers both saves a workflow you would otherwise maintain twice. If video is the majority of your output, a video-first pipeline will understand pacing and camera work better than a general tool. The switching cost is mostly in re-learning conventions, not in the files themselves.
What should I check before trusting a tool with a large footage batch?
Four things: whether it samples multiple frames or just one, whether uploads resume after an interrupted transfer, whether it reports per-agency delivery status rather than a single success message, and whether metadata exports as CSV or XMP so nothing is trapped in the tool. Large video files make a silent failure expensive, so status reporting matters more than it does for stills.