AI for Puppet Performance Blocking Notes
As of 2026-10-16 · AZET 운영팀 (the AZET team). We operate azet.io and publish this guide for fun and reference.
"The fox notices the cup" sounds like a complete direction until someone has to perform it. In a fictional tabletop puppet scene, the fox could turn toward the cup, move closer, or pause with its head already facing it. Those choices tell different little stories.
I would keep the character's intention beside the visible action, rather than asking one sentence to serve both purposes. If you use an AI text assistant, ask it to reorganize movement descriptions you provide. The puppeteer supplies observations and judges what can be performed. Every character, arrangement, and rehearsal note in this guide is fictional.
Three layers for one small action
Our hypothetical scene contains a fox puppet, a cup, and a tabletop marked with three named positions: Home, Cup, and Exit. These labels belong only to this exercise. The operator decides what they correspond to in the real setup.
Write the blocking in three layers:
- Story intention: the fox becomes curious about the cup.
- Visible action: the fox turns toward Cup, pauses, then moves from Home to Cup.
- Operator question: can the turn and movement be performed as separate actions with this puppet?
The third layer is not a defect in the notes. It marks a decision that requires the actual puppet and operator. A fluent rewrite should not erase it.
Add an endpoint too. "Moves toward the cup" could mean a small approach or a complete crossing. In this fictional plan, the fox stops beside Cup without touching it. That distinction matters when the next action is a hesitant attempt to investigate.
Draft an action strip
An action strip is simply a short sequence for this exercise. Each entry has a starting position, one visible change, and an ending position. It is not a universal notation system.
Convert my supplied puppet blocking into an action strip. Preserve the named positions. Separate story intention from visible movement. Use one main action per entry. Flag any movement whose feasibility I have not checked. Do not invent grip changes, hidden mechanisms, or additional operators. Notes: [paste observations].
For the fictional fox, a manually composed strip could read:
- P1: At Home, turn the fox toward Cup. End at Home, facing Cup.
- P2: Hold that position for the agreed pause. Duration remains undecided.
- P3: Move from Home to Cup. Stop beside the cup without contact.
- P4: Wait for the operator's next agreed action; investigation is not yet blocked.
Notice that P4 does not pretend the whole scene is finished. The strip covers only the movement the maker has described.
Revise after a person tries it
Suppose a hypothetical rehearsal note says: "The turn is visible, but the crossing begins before the pause reads as a separate moment." An editor could propose combining the turn and crossing. That might simplify the page while removing the intended hesitation.
I would instead preserve the hesitation and ask the operator to test a clearer separation. The next draft could say: "Complete P1. Hold at Home. Begin P3 only after the agreed signal from the rehearsal leader." This is a rehearsal instruction for testing the sequence, not necessarily the final performance cue.
After trying it, the operator supplies a fresh observation. Perhaps the separation now looks too deliberate for this fictional scene. The group can shorten it, remove it, or change the intention. The important part is that a person makes that choice from the performance.
Keep two kinds of notes apart
A story note might say, "The fox is curious but cautious." An operator note might say, "The cup blocks my view of the Home mark." Do not merge them into "make the fox more cautious." That turns a setup problem into an acting instruction.
For the fictional obstruction, the operator could request a different reference mark. Once the group agrees, update every affected position label together. Leave the character intention unchanged unless the group also decides to change it.
On the working strip, keep only the current action sequence and unresolved practical questions. Put alternatives on a separate scratch page. Otherwise, an abandoned move toward Exit may look like the next instruction.
FAQ
How do I describe where a puppet is looking?
Name a visible target supplied by the maker, such as Cup or a second character. If the puppet's orientation is difficult to judge, record that observation and test a clearer action. Avoid substituting an emotion word for a direction that nobody can see.
Should the notes prescribe the operator's hand movements?
Only include movements the operator has supplied and checked for that puppet. A text description cannot establish how an unfamiliar puppet is constructed or handled. Keep unverified handling instructions as questions, and let the operator decide what belongs in the final notes.
What if one operator handles both characters?
List which character acts in each beat and mark any overlap for the operator to review. In a hypothetical fox-and-bird scene, simultaneous gestures might need adjustment. Do not assume that a polished sequence is physically workable; ask the operator to supply the revised order after trying it.
A note from the makers
Our 2026-10-03 introduction describes azet as an AI assistant and an email early-access waitlist. This fictional blocking exercise is not evidence of a current AZET performance-planning feature. For the waitlist described in that dated introduction, visit https://azet.io and consult the website for the latest availability.