AI Actions
An AI action is a reusable performance for one character. You describe how the character should act, such as “excited product pitch with big hand gestures”, and save it. Each time you run it, you give it a new line to say, and the AI makes the character speak that line while performing it. Use AI actions for recurring presenter moments, like an intro, a pitch or a sign-off, across many scenes.
Two types of AI action
Section titled “Two types of AI action”| Editable action | Video action | |
|---|---|---|
| What you get | A voice clip plus rig actions placed along the speech | An AI video clip of your character performing and speaking |
| Editable afterwards | Yes, every clip can be moved, trimmed or removed | No, it’s a video |
| AI credits | The performance is free; a typed line is voiced with text-to-speech, which uses credits | Uses AI credits for the video, based on its length and model |
Editable action (“Free: the AI directs rig actions along your speech. Editable in the studio.”) picks matching actions from the library and can create up to four new ones for moments the library doesn’t cover. Video action (“AI video clip of your character performing and speaking. Uses AI credits.”) generates a clip from a still picture of your character and places it over the character.
Create an AI action
Section titled “Create an AI action”- Select the character on the canvas. It must be facing Front
- Click Actions in its toolbar
- Under My animations, click Create AI action
- Enter an Action name (up to 60 characters), for example “Excited product pitch”
- Under Describe the action, describe what the character does while speaking: gestures, mood, energy (up to 500 characters)
- Under Action type, choose Editable action or Video action. For a video action, also pick a Video model
- Under Speech, type What should the character say? and choose a Voice, or pick a recording under Or use one of your audio files. With the Vidu Q3 Turbo or Grok Imagine models you only type the line, because those models speak with their own generated voice
- Click Save to keep the AI action for later, or Save & run to save it and perform it right away at the playhead
The AI action appears as a tile under My animations, with an Editable or AI video badge.
Run an AI action
Section titled “Run an AI action”- Select the character and move the playhead to where it should start speaking
- Open Actions and click the AI action’s tile
- Type What should the character say?, then choose a Voice or one of your audio files (the voice you saved with the AI action is pre-selected)
- For Vidu Q3 Turbo or Grok Imagine, set Clip length (seconds)
- Click Add to character
The dialog shows the progress: “Generating the voiceover…”, “Directing the performance…” and, for video actions, “Rendering the video…”. The run keeps going if you close the dialog. When it’s done you’ll see “Placed on your character.”
What an editable action adds
Section titled “What an editable action adds”- A voice clip at the playhead, so the character lip-syncs to the line. If the line runs past the end of the scene or the character, both are lengthened to fit
- Library actions placed along the speech, blending into each other
- Up to four new actions created for this run, saved under My animations so you can reuse or edit them
Everything lands on the character’s timeline as normal clips, so you can move, trim or remove any of them. If no library action fits, the character just talks and keeps its default action.
What a video action adds
Section titled “What a video action adds”The finished clip is added as a video on top of your character, with the same position and size, starting at the playhead and running for the length of the clip. The clip carries its own sound, which is the character’s speech. If the clip runs past the end of the scene, the scene is lengthened.
Video models
Section titled “Video models”| Model | Voice | Length |
|---|---|---|
| P-Video Avatar | True lip sync to your chosen voice or audio file | Follows the speech, up to 20 seconds |
| Vidu Q3 Turbo | Speaks the line with its own generated voice | 1 to 16 seconds |
| Grok Imagine | Speaks the line with its own generated voice | 1 to 15 seconds |
Video actions animate a still picture of the character. If Glintia can’t find one, you’ll see “This character has no still image to animate.”
Edit, fine-tune or delete an AI action
Section titled “Edit, fine-tune or delete an AI action”Hover the AI action’s tile in the Actions panel:
- Edit AI action settings (gear): change the name, description, type, model or voice, then click Save Changes
- Edit action in studio (pencil, editable actions only): opens the AI action’s own animation in the Animation studio, where you can pose it and save it. See Custom Actions and Motion Capture
- Delete this AI action (trash): confirm to delete it. Results already on your scenes are kept
Credits and limits
Section titled “Credits and limits”- Editable actions: directing the performance and creating new actions doesn’t use AI credits. A line you type is voiced with text-to-speech, which uses AI credits like any voiceover (see Text-to-Speech). Choosing one of your audio files avoids a new voiceover; if that file has no mouth shapes yet, it’s transcribed for 0.25 AI credits per minute
- Video actions: use AI credits based on the clip’s length and the model, plus a text-to-speech voiceover when P-Video Avatar voices a typed line
- You can keep up to 30 AI actions per account. New actions created by editable runs count toward your limit of 30 custom animations
- AI actions belong to the character they were made for, and only work while it faces Front
Troubleshooting
Section titled “Troubleshooting”- “You have reached your AI generation limit for this month”: you’ve used your AI credits. See AI Studio and AI Credits
- “You have reached your AI action limit”: delete AI actions you no longer use
- “Some new actions could not be created: you reached your custom action limit”: delete unused animations from My animations, then run it again
- The create tiles are missing: the character is facing Side or 3/4. Turn it back to Front