Guides / AWS Elemental Inference
AWS Elemental Inference and Sceney: live ad context alongside your Elemental workflow
AWS describes Elemental Inference as AI applied in parallel with encoding, inside AWS Elemental MediaLive and MediaConvert: vertical video, highlight clips, smart subtitles and contextual metadata from one pass. Sceney, the vision language model (VLM) from Edge Video AI, sits next to that workflow with one job: telling the ad request what's on screen and what's said going into each break.
Sceney is the vision language model (VLM) from Edge Video AI. It watches live video, reads what's on screen and what's said, and turns it into signals ads and shopping can act on in the moment.
Works alongside your existing playout and ad server.
AI in the encoder
AWS says Elemental Inference works from a single encoding pass to make vertical video, highlight clips and same-language subtitles, and to surface contextual metadata such as IAB content categories, brand safety signals, objects, actions and scene descriptions. It is set up through MediaLive and MediaConvert.
Context at the ad break
Sceney reads the live stream itself: the picture, what's said, and the text and logos on screen. Live enrichment turns that into IAB categories, scene keywords and brands seen on screen, ready for each ad break. If the context isn't ready, the request goes out as it always would.
Who does what
AWS Elemental Inference and Sceney, side by side
| AWS Elemental Inference | Sceney | |
|---|---|---|
| Where it runs | Inside AWS Elemental MediaLive and MediaConvert, in parallel with encoding | Alongside the publisher's existing playout and ad server, on or off AWS |
| Main jobs | Vertical video, highlight clips, smart subtitles, contextual metadata | Live ad context for each break |
| Ad context it returns | IAB content categories, brand safety signals, objects, actions, scene descriptions | IAB categories, scene keywords, brands seen on screen |
| Reads | The video as it is encoded | The live stream: what's on screen and what's said |
| Path to the ad break | MediaLive adds Inference details to the break markers; MediaTailor fetches the signals and passes them to the ad decision server | At the break, the ad server or SSAI can call Edge for the current context, which is added to the request already being sent |
| Brands on screen | AWS describes custom functions, using a Bedrock model you choose, for jobs such as sponsor logo exposure | Brands seen on screen are part of what live enrichment returns |
| Vertical video, clips, subtitles | Yes | Not Sceney's job |
| Channels outside AWS Elemental | Built into AWS Elemental | Yes, the same fields on or off AWS |
| How you get it | From AWS, with consumption-based pricing | From Edge Video AI |
AWS facts are from AWS's own published pages, listed under Sources below. We do not speak for AWS.
Working together
Why run Sceney next to Elemental Inference?
They do different jobs. Elemental Inference earns its place in the encoder: AWS's case for it is processing video once and getting vertical versions, clips, subtitles and metadata from the same pass. Sceney does not make any of those, and it does not replace MediaLive, MediaConvert, MediaTailor or Elemental Inference.
Sceney is built for one moment: the ad break. It reads what's on screen and what's said, and returns IAB categories, scene keywords and brands seen on screen in shapes ad systems already understand. That is the context the buy acts on.
The ad request can carry more than one source. AWS's own MediaTailor guide shows the Elemental Inference call chained with other functions before the request goes to the ad decision server. Sceney's fields are added to the publisher's existing ad request in the same spirit: next to what is already there, not instead of it.
Many channel groups also run a mix: some channels on AWS Elemental, others elsewhere. Sceney returns the same fields for both, so the ad side sees one vocabulary.
How they sit together
What changes, and what doesn't
- Your AWS Elemental workflow keeps doing what it does: encoding, Elemental Inference outputs, and MediaTailor at the break.
- Sceney reads the live stream on its own. It does not need Elemental Inference output to work.
- At the break, the ad server or SSAI can call Edge for the current context. IAB categories, scene keywords and brands seen on screen are added to the request it already sends.
- If the context isn't ready, the request goes out as it always would. Edge can add information; it cannot block a request.
Sits beside you
Works alongside your existing playout and ad server. It does not replace MediaLive, MediaConvert, MediaTailor or Elemental Inference.
Adds context, never blocks
If the context is not ready, the request goes out as it always would. Edge can add information; it cannot block a request.
Describes the programme
Sceney says what is on screen going into each break. It does not identify or profile viewers.
Questions
- What is AWS Elemental Inference?
- AWS describes Elemental Inference as a fully managed service that applies AI in parallel with encoding, inside AWS Elemental MediaLive and MediaConvert. Its features include vertical video, highlight clips, smart subtitles and contextual metadata such as IAB content categories, brand safety signals, objects, actions and scene descriptions. See the AWS Elemental Inference page. That is AWS's description; we do not speak for AWS.
- Is Sceney an alternative to AWS Elemental Inference?
- Not if you already run AWS Elemental. They do different jobs: Elemental Inference works inside the encoder, and Sceney, the vision language model (VLM) from Edge Video AI, adds live ad context at each break, working alongside your playout and ad server.
- Can Sceney and AWS Elemental Inference run on the same channel?
- Yes. Sceney reads the live stream on its own and does not need Elemental Inference output. It does not replace MediaLive, MediaConvert, MediaTailor or Elemental Inference. Works alongside your existing playout and ad server.
- What does Sceney add next to Elemental Inference metadata?
- IAB categories, scene keywords and brands seen on screen, read from what's on screen and what's said going into each ad break. They are added to the publisher's existing ad request. If the context isn't ready, the request goes out as it always would.
- How does AWS Elemental Inference metadata reach the ad break?
- According to AWS, MediaLive adds Elemental Inference details to the break markers. At each break, MediaTailor calls the Elemental Inference GetMetadata API and passes the IAB and GARM signals to the ad decision server. See Contextual ad targeting with Elemental Inference in the AWS MediaTailor user guide.
- Do I need AWS to use Sceney?
- No. Sceney works on or off AWS, and returns the same fields either way. Edge Video AI is not an encoder, a playout system or an ad server.
- Who makes Sceney?
- Edge Video AI, based in The Hague and founded in 2019.
Sources
- AWS, AWS Elemental Inference, product page, read 20 Nov 2026. Source for: AI applied in parallel with encoding; vertical video, clip generation, smart subtitles and contextual metadata (IAB content categories, brand safety signals, objects, actions, scene descriptions) from a single encoding pass; deployed through MediaLive and MediaConvert; consumption-based pricing.
- AWS, AWS Elemental Inference FAQs, read 20 Nov 2026. Source for: the list of capabilities; customisation with functions on an Amazon Bedrock model you choose, with sponsor logo exposure given as an example.
- AWS, What is Elemental Inference?, AWS Elemental Inference User Guide, read 20 Nov 2026. Source for: smart crop, event clipping, smart subtitles, contextual metadata at shot and scene level, customer decisions and functions.
- AWS, Contextual ad targeting with Elemental Inference, AWS Elemental MediaTailor User Guide, read 20 Nov 2026. Source for: MediaLive adds Elemental Inference details to the break markers; MediaTailor calls GetMetadata at each break and passes IAB and GARM signals to the ad decision server; chaining with other functions.
AWS, AWS Elemental, MediaLive, MediaConvert and MediaTailor are trademarks of Amazon.com, Inc. or its affiliates. Edge Video AI is not affiliated with Amazon Web Services.
Running AWS Elemental? Add the moment.
Works alongside your existing playout and ad server.