Green screen is the easy case for an editor. The background can be removed, so the presenter can be placed anywhere: beside a slide, inside a diagram, on a set the course chose. Most of what a university has on file was not shot that way. It is a lecturer standing in front of a classroom display, a whiteboard or a bookshelf, with the camera on a tripod.
Until this month that footage stopped our engine at the first step: it looked for green to remove, found none, and halted. Now it recognises the recording for what it is and switches to a different way of editing, which we call In-Room. The presenter stays in their own room. Everything the lecture needs is added on top of that room, or briefly in front of it.
The clips below come from a two-minute postgraduate management lecture, recorded in front of a classroom display. On the left is the camera file exactly as it was recorded. On the right is the finished lecture at the same moment.
1. The opening
The raw file starts the way most do: the lecturer, mid-shot, already talking. The finished lecture opens on a title card for the topic, then cuts in a little closer on the lecturer for the sentence that sets up the session.
“Now we come to the most important part, that is the nature of managerial economics …”
The same words, under a title card and then a closer shot.
A fixed camera starts to feel static after twenty seconds or so. A human editor would cut to a closer shot now and then. The engine does the same, but only in the gaps between overlays, only for a single sentence, and always scaled about the lecturer's head, so the face stays exactly where it was and nothing on screen jumps.
2. Words on the wall
When the lecturer names the first characteristic, a chapter card marks it. When she lists what managerial economics draws on, each item appears as a tag on the part of the room she is not standing in, on the word that introduces it.
“Here, the first is applied science. Managerial economics applies economic theory and tools … statistical and mathematical tools …”
A chapter card, then “Economic theory” and “Statistical tools” arrive as she says them.
Which side counts as free is measured, not assumed. Before anything is placed, the engine watches the whole recording and maps where the lecturer actually moves, including the full sweep of every gesture. Overlays only go on the wall outside that map. If the free wall is too narrow for a card, the card becomes a bar along the bottom instead.
3. A worked example, in a side panel
The heart of this lecture is an example: a bakery pricing a new pastry by checking its costs, the demand, and what competitors charge. That is a list a student should see build up, so it gets a side panel, and the camera makes room for it.
“What they do is they evaluate the cost … They are checking the demand pattern …”
A panel slides in with the example; the frame eases over so the lecturer stays fully in view beside it.
Side panels are the strongest effect, so they are rationed: at most one every forty seconds. A list that arrives sooner than that is shown as tags instead. Here the second half of the example, competitors and the final price, comes back as tags a few seconds later rather than as a second panel.
4. The close
The last thirty seconds sum up and move on. A one-line bar carries the takeaway, a chapter card announces the second characteristic, and the question the lecturer asks the class gets a card of its own.
“… the concept of applied is equal to a practical sense … The second point is decision oriented. Okay, what is decision?”
“Applied = practical sense” as a bar, a chapter card, then the question on a card of its own.
How the engine decides
Recognise the recording. At ingest the engine looks for a backing it could key out. If there is none, the lecture is marked as In-Room automatically — nobody has to sort the footage first.
Map the room. The lecturer's movement over the whole recording gives the space they use and where their head is. Everything outside it is wall the overlays may use.
Turn the lecture plan into moments. The same plan that would drive an animated lecture — headlines, lists, definitions, questions, each anchored to the words that introduce them — becomes title and chapter cards, tags, note cards, side panels and bars.
Keep the rhythm. One thing on the wall at a time, a clean shot of at least three seconds between them, a side panel at most once in forty seconds, full-screen cards kept short. In this lecture the lecturer is on screen for 92% of the running time.
Move the camera like an editor. Punch-ins on single sentences in quiet stretches; slow reframes for side panels; cuts only at the ends of sentences.
Dress it in the course's style. The overlays wear the course's own template style — its typefaces, palette, corners and motion. This lecture is in Sticker. Any of the twenty-five styles works.
The welcome and closing cards still show the lecturer as a cut-out, as they do for green-screen lectures. Without a green backing, the cut-out is made by a model that separates a person from the room behind them, from a single well-chosen frame.
What it does not do
Edit the lecture as recorded
- Keep the lecturer in their own room, lit as they were
- Add cards, tags, panels and bars on the words that introduce them
- Punch in and reframe like a camera operator would
- Run the same cleanup, captions and checks as every other lecture
Pretend there was a green screen
- Remove or replace the room behind the lecturer
- Fix a camera that moves or is hand-held: it assumes a tripod
- Read what is on a display behind the lecturer — an overlay may sit over it
- Add detail a punch-in cannot find: a 1080p source softens slightly at 1.2×
What it came to on this lecture
| Added | Count | On screen for |
|---|---|---|
| Title and chapter cards | 3 | 11 s in front of the lecturer |
| Question card | 1 | |
| Tags, note card and bar on the wall | 4 | 64 s on the room, beside the lecturer |
| Side panel with reframe | 1 | |
| Punch-ins | 3 | one sentence each |
| Clean shot of the lecturer | — | 57 s, untouched |
| Time to render | — | about two minutes for 2:12 of lecture |
A green screen gives an editor more options. It is not the only way to make a lecture look produced, and most of the lectures a university already holds were never shot on one.
In-Room is now the default for any recording without a green backing, and it sits alongside the rest of the pipeline: cleanup, captions, Quick Checks and the welcome and closing cards. For the cleanup that runs first on every lecture, see the part of the recording no student should hear.
Send us a lecture recorded in an ordinary classroom — no green screen, no studio — and we will send it back edited In-Room, in your course’s style. Send a lecture →