Make Interactive

Overview#

Make interactive (in the editor's header) takes a video you have not touched and builds interactive features into it: chapters the viewer can jump between, a step checklist that follows the video, highlight markers, info and glossary cards, product cards, a quiz, a call to action, subtitles. You do not place anything by hand. MP4E first watches and listens to the video, then offers the features that fit what it found, and builds the ones you pick.

The wizard has four steps — Understand, Choose, Build and What was added. Click a step at the top to go back to it. Close the wizard at any time: an analysis or a build in progress keeps going, and reopening the wizard takes you to where it stands (a running build opens on Build, a project that already has built features opens on What was added).

1. Understand#

The analysis is what the later steps are built from: summaries, scenes described in prose, chapters, the people in the video, key moments, and the things on screen or mentioned. This step starts the analysis (or picks up one already running), shows how it is going, and, when it is done, summarises what was found — your own corrections count and items you hid are left out. Analyse again replaces the analysis after you confirm. Above the button, What to look for lets you name the kinds of things to look for (for example animals, or the tools in a workshop) before you start. Continue moves on to Choose.

An analysis is paid work. You can read and correct it afterwards in the editor's Intelligence tab.

2. Choose#

Pick an experience — a ready-made combination of features for a kind of video. The one that fits best is marked Suggested and selected for you:

• How-to / Recipe — chapters to jump between, a step checklist that ticks itself off, tips as they come up, subtitles.
• Lesson / Training — chapters, highlights, glossary cards for the terms explained, quiz questions after each chapter (the video pauses for each, and the score shows at the end), subtitles.
• Product / Shopping — chapters, product cards while each product is on screen, a shopping cart, facts about them, a call to action. Viewers can tap a product card's button or the call to action while the video plays. A build adds a link only when the video states a web address and never makes one up: a product card without one has no button, and a call to action without one only reports the tap to your rules, so add the links you want afterwards.
• Tour / Showcase — chapters for each stop, highlight markers, info cards, a call to action.

Below the experience is its list of features, each with the reason it is offered ("Steps found: 9"). A feature is only offered when the analysis holds what it needs, so a video with no products never gets product cards; features the video cannot have are not shown. Subtitles are offered only when the whole transcript is short enough for a build to write out; on a longer video, add them from the Localization panel. When the video's picture already shows captions, Subtitles starts unticked with the note "This video already shows captions"; tick it if you want your own anyway. Untick any you do not want. Add more lists the features the video can have that the chosen experience does not include — a quiz on a how-to, say. A feature already built is shown as Already added and cannot be chosen again until you remove it (see What was added).

When you pick product cards and the analysis found products that show on screen, a line under the features reads Track N products · about M min, ticked. With it ticked, the build first finds those products in the picture and follows them through the video. Each tracked product then opens its own card when a viewer taps it, beside the product rather than over it, and shows it by itself while it is the only product on screen. The card shows the product's name, description and price, with no button, so the viewer stays in the video. Products that could not be tracked get the usual timed card. Untick the line to build timed cards only. If your account cannot track products, the line is unavailable and says why, and the build runs without tracking.

Under the line, Cue sets how viewers can tell the products are tappable: None, Subtle (the default: a soft light around each product as it appears, and an outline under the pointer) or Markers (a small pulsing dot on each). You can change it later in the products group's settings (see Objects). If you have already set up your own tap card for that group, the build keeps yours.

Shopping cart is offered when the analysis read a price for at least one product ("Products with a price: 3"). It sits under Product cards and needs them: unticking Product cards unticks it, and it reads Needs Product cards while Product cards is neither ticked nor already in the project. When an earlier build added the product cards, you can add the cart on its own. The build sets the cart up itself, before the agent starts: the Cart on a layer of its own, and Add to cart on the product cards of every product with a price — the tracked products' own cards and the timed cards. Pressing it puts the product in the cart, and the video keeps playing. Checkout happens on your site: see In-video shopping for what your page receives. If the video already has a Cart you added yourself, the build adds no second one and leaves the cart to you.

Then choose a theme: Clean, Bold, Cinematic, Playful or Studio. Each tile previews the theme's colours, and the theme is applied to every plugin the build adds that has a theme setting. The step also shows how many agent credits you have left. A quiz, like the other features, is written for you — from what is said in the video — and you edit its questions afterwards in the plugin's settings.

3. Build#

The start screen summarises the experience, the features and the theme. Start does three things before the agent begins: it saves any edits you have pending, it turns on version history for the project if it was off, and it saves a version of the project as it is now if its current state has none. That version is what Undo the whole build returns to.

When tracking is ticked, the products are tracked first. The screen shows how far it has got, which can take several minutes, and the agent starts when tracking ends. Tracking keeps going if you close the wizard. If it fails, the build still runs, with timed cards for every product.

A build is an agent run, so it spends agent credits like one and is refused for the same reasons (for example, another run is already in progress); the screen says why in words. While it runs you see what it is doing (reading the plugins, adding the features, checking the project), its latest progress notes and how many changes are saved so far. You can keep working, or close the wizard: the run also shows in the Agent panel. If the agent needs to ask something, the question appears here and you answer it in the Agent panel. Stop (after a confirmation) ends the build where it is; whatever was already saved stays.

Each feature is built on a layer of its own, named after the feature, so the result is easy to find and change. When the project already has a scene group of the analysis' chapters (for example one made with Create scenes from analysis), chapter navigation uses that group instead of adding a second one. When the run finishes, the wizard checks that every feature you chose really ended up on its layer and that the project is still valid, and shows the agent's summary.

4. What was added#

One row per feature in the project: what it is, why it was offered, its layer and the plugin that runs it. The same list is available any time from the Added tab in the editor's right panel. Per feature:

• Switch — turn the feature off and on without deleting anything. A feature that is off is marked Off and does not play.
• Remove — after a confirmation, deletes that feature's layer only. Plugins, scene groups and variables the build added for it stay in the project.

Product cards built with tracking has two parts in its row: On tap (the products group, how many tracked objects it holds, and the cue) and, when some products were not tracked, Timed (the layer of timed cards). Its switch, Timed cards on, turns only the timed cards on and off — the tap cards keep working — and Off then marks the Timed line. Remove takes the tap cards and the cue off the products group, even if you changed the cue since, and deletes the timed layer, as one step you can undo. The group, its tracked objects and any card you set up yourself stay. When the build kept a card you had set up yourself, or could not give the tracked products their own card, the panel says so under what was tracked.

The Shopping cart row names its layer and, while the cart is connected, Add to cart on product cards. Its Remove takes away the cart's layer and the connection and turns the product cards' Add to cart buttons off, as one step you can undo; the product cards stay. Removing Product cards removes the cart too, since it would be left with nothing to add, and the confirmation says so.

The panel also shows the build's check: Checked when all is well, or a list of problems ("Quiz was not added", or a problem the build introduced into the project) with a pointer to fix them. It also notes, without counting it as a problem, a timed card added for a product that already opens its own card on tap. Refine in Agent panel opens the Agent panel so you can ask for changes in plain language — "make the quiz shorter", "move the call to action to the last ten seconds".

Undoing a build#

Undo the whole build (on What was added, after a confirmation that names the version) restores the project to the version saved when the build started, adding a new revision that brings it back; nothing is lost from the history. If you have unsaved edits they are replaced. The tap cards and cue a tracked build set up go with it, and so do the objects tracking added. It is unavailable while the build is running or another agent run is active, when no build has run, when the build was already undone, and when the saved version is no longer in the history.

From code#

Everything the wizard does after the analysis is also available over the REST API and to MCP clients — see Agent API → Make a video interactive.