ArtArch Newsroom
AI Video TechniquesHow to Write AI Video Character Continuity Prompts
Learn how AI video character continuity prompts track wardrobe, accessories, prop ownership, orientation, and motivated state changes across shots.
Provide one character reference and one empty environment reference to create a multi-shot AI video with stable wardrobe details, persistent props, and believable state changes. This AVT-005 experiment follows a reporter through an old archive building and tests three continuity problems across paired shots: clothing and accessories, object handling, and the gradual effect of rain.
The character has a small set of recognizable anchors: a lean face, short wavy brown hair, a faint scar above the left eyebrow, an open olive cotton jacket, an off-white shirt, dark trousers, and a narrow black watch on his left wrist. The environment adds a long table, wooden archive cabinets, a glass door, wet stone steps, and a dark red hardback file book.
Continuity begins with a state record
Each new shot should inherit a clear state from the previous one. A useful record answers four questions:
- Who is the character?
- What is the character wearing and where are the accessories?
- What object is being held, by which hand, and in what orientation?
- Which changes have already happened in the story world?
Repeating these facts is purposeful. It gives each shot the same visual contract while allowing the action and framing to change.
Test 1: wardrobe and accessory continuity
The first pair moves from a wide entrance shot to a closer view at the archive table. The reporter enters from frame left, walks to the table, opens the file book, and looks toward the glass door.
The continuity prompt repeats the fixed anchors in both shots:
Keep the same lean face, short wavy brown hair, and faint scar above the left eyebrow. The olive cotton jacket remains open over the off-white shirt, the dark trousers remain unchanged, and the round watch with a narrow black strap stays on the left wrist.
The second shot adds direct inheritance language: it begins with the same person already standing at the table. That sentence connects the new composition to the previous ending.
Wardrobe descriptions work best when they specify state as well as appearance. “Olive jacket” describes color. “Olive cotton jacket, still open” also controls material and closure. “Watch on the left wrist” controls accessory location and screen logic.
Test 2: prop ownership and orientation
The second pair follows the dark red file book from the center of the table to the glass door. The reporter picks it up, examines the cover, turns, walks right, stops at the door, and looks back.
The state record includes five properties:
- holder: the reporter;
- hand: right hand;
- position: beside the right side of his body;
- orientation: cover facing outward and spine facing down;
- quantity and appearance: one dark red hardback book.
Keep exploring
More in AI Video Techniques
Motivated Cut AI Video Continuity: Change the Shot for a Reason
Learn how a visible action trigger, clear shot-size change, and locked screen direction create more intentional same-scene AI video edits.
Deep Depth of Field AI Video: Keep Every Story Layer Readable
Learn how deep depth of field AI video prompts keep foreground tools, central action, and distant location cues readable in one layered shot.
Medium Depth of Field AI Video: Balance Subject and Setting
Learn how medium depth of field AI video prompts keep hands and active objects readable while preserving recognizable environmental context.
AI Video Character Reference Framing: Match Portraits to Landscape Shots
Prepare vertical character portraits for landscape AI video by matching aspect ratio, body scale, headroom, floor space, and movement room before generation.