Home/ Materials/ Session 3 Facilitation

Pilot script

Session 3: Video

◯ 60–90 min ▤ Facilitation guide ✦ Pilot ready

Use this script for a 60–90 minute pilot focused on temporal coherence, drift, anchoring, and why video generation is harder than generating a single image.

Core question. What changes when generation has to work across time?

Materials

Run of show

TimeSegmentFacilitator moveParticipant action
0–5Welcome & bridgeConnect Session 2 to Session 3: a good frame is not enough; video must stay coherent across time.Choose a participation pathway.
5–35Temporal Telephone round 1Open Temporal Telephone in previous-frame-only mode. Use “a person at work” as the vague prompt, then check motion defaults after playback: job, body, setting, tool, pace, gesture.Draw/save frames or observe where defaults and small changes accumulate into drift.
35–50Temporal Telephone round 2Switch to anchor mode and repeat with the same basic subject or motion.Compare whether identity, layout, and motion stay more stable.
50–70Failure analysisUse curated clips, still frames, or optional generated video examples. Name failure modes precisely.Mark identity drift, physics breaks, camera inconsistency, style drift, or temporal jumps.
70–90Video Test ReportAsk participants to complete one A/B/C or failure-analysis section.Share one coherence claim and one responsible-use boundary.

Optional pilot evidence move: use the A/B/C Comparison Board to document a hypothesis, one changed variable, a baseline motion prompt, one continuity anchor, and a stronger anchor or stress test. Keep real-person likeness and consent concerns visible before exporting.

Facilitator prompts

  • “Where did the sequence first begin to drift?”
  • “For the vague prompt ‘a person at work,’ what job, body, setting, tool, pace, or gesture appeared by default?”
  • “What stayed consistent when only the previous frame was visible?”
  • “What did the anchor preserve?”
  • “What does video require that a single image does not?”
  • “Which failure type is easiest to spot frame by frame?”
  • “What would you need to specify to keep identity, motion, and setting stable?”

Investigation prompt

Compare a baseline motion prompt with one revision that adds an anchor: a consistent subject, fixed camera, repeated setting, or explicit continuity constraint. What improves, and what still breaks?

Low-AI / No-AI pathway

Participants can complete the session using Temporal Telephone and curated clips only. No one needs to generate video to investigate temporal coherence.

Fallback plan

  • If drawing with the tool is awkward, have one facilitator draw while participants direct changes verbally.
  • If playback is too fast or hard to see, step through thumbnails manually.
  • If no curated clips are available, use Temporal Telephone outputs as the failure gallery.
  • If participants do not want to analyze AI video, focus on continuity, consent, and classroom-safe critique protocols.

Pilot QA notes

During the pilot, note:

  • Whether participants understood previous-frame-only versus anchor mode.
  • Whether the Save frame and Play controls were discoverable.
  • Whether thumbnails were large enough for discussion.
  • Which failure categories participants used naturally.
  • Any ethical concerns around generating motion, simulating people, or using real likenesses.