Motion graphics

A motion graphic in Agent28 is a small animated component the AI writes for you: a lower third, a stat callout, a quote, a title card. It lands on the timeline as a normal layer with its own properties and keyframes.

The Motion Graphics tab

Three sub-tabs: Generate, Presets, Settings.

  1. Move the playhead to where the graphic should start.
  2. On Generate, describe it in “Describe a motion graphic…”. Attach a screenshot or logo with Image if it should match something.
  3. Click the generate button. A Director plans the moment (what it says, when, how it moves) and an Engineer writes the component. Stop generating cancels.

One request makes one graphic. Ask for several explicitly if you want a set. Under Settings, Default layer depth decides whether new graphics go Behind or in Front of a person matte.

Presets

Preset groups
GroupWhat it contains
Style presetsCurated templates grouped by editing style. Hover for a looping preview; click, or the + on the preview, to add at the playhead.
GeneralStatic scaffolds with placeholder copy. Add one, then use its Chat tab to replace the text and add motion; the scaffold's colours and fonts are kept.
Your saved graphicsAnything you saved with Save to library. Inserted as-is.
In this projectGraphics already on this timeline, for reuse.

Cards with a Plus pill are part of the full catalog, included with Plus and above; clicking one opens the plans view.

Editing a graphic

Select a graphic on the timeline. The header shows its prompt with Regenerate, Save to library, and Remove, and two tabs:

  • Chat: describe changes to this graphic only. Examples the panel suggests: “Make the text larger and bolder”, “Change the color scheme to blue and white”, “Slow down the entrance animation”, “Add a subtle glow behind the headline”.
  • Properties: X location, Y location, and Scale, each with a keyframe toggle; Start (s) and Duration (s); Blending; and the graphic's own controls, such as its text, colours, and fonts.

You can also drag the graphic on the preview to place it.

Asking in chat

You can ask the main chat for graphics too: “put a lower third with my name when I introduce myself”, “add flying money when I mention earnings”. The agent reads the transcript to find the moment.

  • In Plan mode the agent describes the graphic it intends to build in a message and shows one Approve button with “or tell me what to change”. There is no card or preview at this stage; the description is the plan.
  • To revise, reply. Anything that is not a yes re-plans; “never mind” drops the plan and nothing is added.
  • After Approve the message reads “Done — N graphic(s) on the timeline.” If the build fails, the plan stays and you can Approve again.
  • In Agentmode the graphic is built without the Approve step. Either way it belongs to that turn's Revert.

Reference images attached in chat are named image1, image2… and can be placed inside a graphic by name: “use image2 as the logo in the lower third”.

Overlay, fullscreen, panel

A graphic is one of three kinds, decided from your description:

  • Overlay: transparent, drawn over the footage. Lower thirds, callouts, quotes.
  • Fullscreen: an opaque takeover with its own backdrop. Title cards, section breaks.
  • Panel: a band on the left, right, top, or bottom taking 20 to 50 percent of the frame, with the footage in the rest. Bullet lists beside a speaker.

Style Studio

Style Studio on the home screen turns a reference video into a reusable editing style:

  1. Name the style (“My tutorial style”) and click Choose video. Use one of your own past videos or any video whose editing you admire.
  2. Click Study this video. The AI studies the hook, pacing, structure, captions, and how graphics are used, and captures up to 12 of the video's graphics as reusable components.
  3. Click Save style. It appears in the Create project wizard under Editing style, and its graphics appear in Presets.

Note: Style Studio is included with Pro and above when running on managed AI. If scene analysis is unavailable on your machine, the style is studied from the transcript alone and visual observations are limited.