Four preparation windows, no act filmed

Four production windows ended on October 14 without a single act being filmed. Each had twenty-one minutes eighteen seconds for preparation. The crew produced choices, the native checks found actual failures, and the deadline closed the work before it reached the recorder.

I had asked for related contributions to be tried together as soon as their inputs existed. Departments completing their separate assignments was not enough if the combined performance still could not run. The replacement needed to bring those choices together early and return precise failures to their owners.

Work reached the shared scene

Version 2 gave owners proposals rather than competing copies of the master. One technical writer retained the inputs and saved immutable scene versions. A changed assumption or shared animation channel could be rejected before one contribution silently overwrote another.

Some connected fixtures worked. A generic 180-second moving scene completed physical simulation, observation, fog refresh, save, reopen and freezing in approximately eight minutes thirty-three seconds. The existing recorder then sampled all seven cameras from that exact frozen scene. Another fixture launched thirteen distinct Opus 5.5 owners before score authoring and resumed its original photography owner into a synthetic post phase.

Those results established equipment and session behavior. The second fixture deliberately invented its recorded state to test the transition. Neither fixture staged the supplied play or delivered an acceptance film.

The actual shows exposed the difference. The first hit scheduling and missing-resource problems. The second caught Yvette’s physical failure early but did not complete the required return to that owner. The third saved an actor revision, then refused projection images that had not been retained. Sound/Projection republished the resources, but the final combination did not finish before expiry.

The latest clock

The fourth run started shortly before eleven in the morning, Chicago time, and closed at 11:21. Eleven CLI owners began before the Director wrote the score; independent native Sound/Projection and photography owners joined. All thirteen roles were present. The Director published the complete action score, and Sound/Projection retained three editable Kintsugi arrangements and their musical mixes.

Actual playback caught a 3.243-millimetre Yvette fastening gap near five seconds, insufficient movement under the bound camera observations, and a bell-contact distance of about 82.5 millimetres against a 50-millimetre limit. The reports supplied real causes and frames. They still arrived through a sequence that could consume several minutes before reporting an early failure.

The writer’s order of work explains that delay:

            report['bake'] = integrate.bake_world(bpy, scene, last_frame)
        if not report['bake']['is_baked']:
            raise ValueError('Native rigid-body bake did not finish.')
        measured = attachment_actions(ready, end_s, combined)
        observations_path = out / 'played-before-save.json'
        announce('observe-before-save')
        before = played_observations.observe(bpy, measured, scene_sha256=None, scene_binding='in-memory-candidate', output=observations_path)

Public source excerpt. Repository snapshot, requiring repository access.

This implemented observation follows the preceding native bake; the public excerpt retains that surrounding order. The writer therefore completed a long simulation before inspecting the action that had failed near the beginning. The code establishes an execution sequence, while the retained run receipts establish its cost.

Three completed failed checks took about 184, 185 and 282 seconds, together ten minutes fifty-one seconds. A final combination attempt spent another two minutes before its bake was interrupted by the deadline. Repeated waiting returns and changed inputs did not converge on a frozen scene during the remaining time.

The recorded owner returns also prevent a simple claim that everyone received and repaired every failure. Some sessions were blocked by command-permission classification, and some other owners repeatedly returned against unchanged dependencies. A crew roster and a launch event cannot establish that the intended recovery actually happened.

What the time establishes

The four preparation budgets total eighty-five minutes twelve seconds, excluding maintenance between commissions. Model work, native jobs and dependency waits overlapped. The records do not provide a reliable exclusive allocation of productive work and idleness, so I cannot turn the unallocated time into a claim that the crew sat idle for that duration.

The result requires no such estimate. There were zero recorded acts and no acceptance film. The closed clocks stayed closed, and the existing recorder remained available. The replacement had improved the accuracy of some observations while still failing to get the crew’s work onto film within the production window.

That is the state the showrunner direction must address. More precise failure receipts help only if the Director and original owners can use them in time to produce the requested performance and leave enough time for its recording and assembly.