Observation happens alongside performance
The 3P Manifesto expands its name into Performing, Peering and Processing. Its Chinese passages discuss real time and unpredictability, auditory segmentation, and deliberate cutting and arrangement. English and Japanese versions are preserved alongside them.[S1]
The words place performance, observation and processing together. What happens in a room and how it is later selected, arranged and heard can become different parts of one creative practice. Yet a manifesto records a language of making; it does not establish that an event followed its principles. This reading traces how the words entered editing projects and how those projects organize positions from which to watch.
The manifesto enters actual tracks
Two projects labelled Titles1.0_FC and Breakup1.0_FC place Chinese, English and Japanese manifesto text on an actual video track. The first three items occupy approximately 0–63.864, 63.864–127.628 and 127.628–191.742 seconds. The extracted text matches the three TXT passages after whitespace is removed; the corresponding decoded text payloads also have identical fingerprints across the two projects.[S1][S2][S3]
This establishes reuse of text within an editing structure, beyond the presence of similarly named files in one folder. It does not establish what an opening sequence finally displayed. Tracks may overlap, text may be obscured, and missing media may affect output. The projects have not been relinked or rendered for this reading, so text in a track is not treated as a verified final image.
A title and five working positions in 1.0
Breakup1.0_FC contains the Chinese title text 分手, meaning “breakup,” at approximately 28.362–30.898 seconds. That is a located title-text record within the 1.0 project. Breakup, 1.0, 2.0, FC and RC remain project labels here, rather than verified release titles or grades of completion.[S3]
The closing text assigns Zhang Shaohua to left-side video photography, Huang Lucang to middle-video recording, and He Fa to right-video recording. All three are labelled 3P group members. Each name can be traced to its own text item and timeline position.[S9][S10][S11]
The same closing section gives Shanshan a directing label and Guangzi a whole-event video-recording label. Only working roles are summarized here; playful character descriptions are omitted. These entries establish how this project describes a division of work. Their final-screen presentation has not been verified, and the names are not assigned to the three source files in 2.0 or to people seen in candidate images.[S12][S13]
Other projects’ public articles also carry 3P credits. Those lists and roles remain specific to their projects; they do not establish a permanent membership roster.[R2]
In 2.0, segmented images accompany long audio items
Sequence 02 in Breakup2.0_RC divides three nested groups labelled right, middle and left into 18, 7 and 8 video segments: 33 in total. Each of three audio tracks retains one long item. Every video segment’s source in-point equals its timeline start; the checked fields do not indicate that these passages were rearranged in time.[S4]
This structure separates the choice of what to see at a given moment from the continuation of sound. Thirty-three segments do not mean thirty-three camera switches: one cut retains the same selected index on both sides. Some segments select a transparent black-video layer, and one opening selection points to a track without an item. Such states require playback verification. They do not establish that the later image is entirely black, that content was deleted, or that the event ended.[S4]
Another arrangement: a base image and three later entries
Sequence 04 in the same main project keeps a base image lasting about 18 minutes 45.783 seconds. A different set of right, middle and left sequences enters at approximately 39.483, 39.783 and 40.217 seconds. This is a different arrangement from Sequence 02. The nested sequences have different identities despite sharing names.[S4]
Left/right positions and differing start points make layout and alignment inspectable work. They are compatible with an attempt to arrange multiple images, but do not establish successful audiovisual synchronization or explain why the makers chose them. Track activation, transparency, overlap and media availability can all affect visibility. The Purple Action trailer project offers a separate comparison; the comparison does not establish influence between the projects.[R1]
Auto-saves preserve different structural states
Another saved 2.0 project uses the same three source paths directly. One starts at zero and two at approximately 2.667 seconds, with both video and audio items for each. A fourth file appears in the media library but is not placed in the sequence. Importing a file and using it in a timeline are different relationships.[S5]
A further auto-save contains five sequences. Sequence 07 has no items in its parsed tracks, while other sequences still use media. Its rightMulticam sequence retains a long left item and places right only at approximately 59.683–71.533 seconds. An empty sequence does not make the entire project empty. Save-time filenames do not, by themselves, establish a revision chronology or explain an author’s decisions.[S6]
A copy of the 2.0 main project in the archival snapshot is byte-identical to the main project examined here. This provides a reliable version connection, but does not establish that 2.0 inherited the opening, ending or credits from 1.0.[S4]
Candidate images make the space concrete, without identifying people
One readable candidate matches the base material’s filename, dimensions and duration of approximately 18 minutes 45.79 seconds. A fingerprint comparison with the file at the original referenced location is unavailable. The other three originals have not been located; this candidate cannot stand in for every viewpoint.[S7]
Two existing contact sheets cover 33 distinct time points. Chairs and seated people appear against a pale background; people in the foreground stand, bend or hold small devices, sometimes obscuring the rear. Later samples show people approaching chairs, standing up or moving towards the camera. These images make viewing positions and obstruction concrete. They do not supply the intervening dialogue, establish how the activity ended, or identify anyone’s name or formal responsibility.[S7]
Text, project and finished film remain separate checks
A subtitle file in the same archive contains 964 entries across four layers, ending at 22 minutes 50.61 seconds, with empty speaker-name fields. That differs from the two main 2.0 sequences, ending at 17 minutes 49.567 seconds and 18 minutes 45.783 seconds. Without checked audio and precise anchors, these subtitles cannot be applied to 2.0 or used to identify speakers.[S8]
The manifesto document’s creation date of 30 September 2020 and modification date of 1 October are document properties, not filming or production dates. Its Latin-script metadata name is not automatically merged with a person credited in the project.[S1]
The evidence currently connects manifesto text, working credits within 1.0, and differing viewing structures preserved in 2.0. A readable final export, three missing originals and reliable audio correspondence are still needed. This publication provides edited documentary and version notes. Original projects, subtitles, private drafts and candidate footage are not newly attached, and no unverified video link is created.