news-rough-cut

News Rough Cut (Intelligent News Rough Cut)

Rough-cut news footage into a short news video with complete content, clear logic and tight pacing. Stay faithful to the original footage, add no external audio, and maintain an objective, formal, tight and clear informational news style.

This is an OpenChatCut-native workflow. Use the current project's assets, transcript, word-level editing, timeline, and editing tools. Do not depend on external download or transcode pipelines.

Workflow Overview

1. Fully analyze the footage(mandatory before editing): Identify the news events, core topics, key figures, important conclusions and effective visuals in the footage, then determine the editing mainline and final video duration. Use read_project、transcribe_track、view_timeline_frames to check the footage content segment by segment.
2. Topic analysis and duration determination: Determine how many topics the footage contains and distinguish core topics from secondary content.
3. Content organization: Open directly with the most important news result/core conclusion/latest development/key on-site visuals, without buildup.
4. Editing execution: Filter according to the keep/delete rules, and cut speech at semantically complete points.
5. Keep only the original audio of the target news footage: Before editing begins, include only the news footage and its original on-site sound explicitly specified/selected by the user as allowed sources; when not explicitly specified, use only the news visuals and on-site sound already present on the active timeline. BGM, sound effects, voiceover, narration and other unselected footage in the media pool must not enter the allowed sources even if they already exist.
6. Final check: Play back segment by segment to verify factual fidelity, semantic completeness of speech, and natural transitions at cut points.

Topic Analysis and Duration Determination

• In principle, a single final video revolves around onlyone core news mainline.
• If the footage contains multiple independent topics, prioritize the topic withthe highest news value, the most complete information and the most sufficient visualsfor editing;do not forcibly splice unrelated topicsinto the same video.
• The final video duration has no fixed limit and is determined automatically based on the following factors:
  • the information volume of the core news;
  • the length of effective speech by figures;
  • the stage of event development and latest progress;
  • the number of key on-site visuals;
  • the duration required to ensure semantic completeness of the news.
• When there is less information,shorten the final videoand avoid adding irrelevant content merely to extend the duration; when there is more information, it may be extended appropriately, butdo not cut off figures' speech, omit key facts or break the news logic in order to compress the duration。

Content Organization Logic

The final video opens directly with the most important news result, core conclusion, latest progress or key on-site visuals, without lengthy buildup. Organize the whole according to the following logic:

1. What happened;
2. What the latest developments are;
3. The final result, subsequent impact or related responses.

If the news event has not yet concluded, end withthe latest developments confirmed so far, and do not speculate on the outcome on your own.

Content retention rules

Prioritize retaining the following content:

• the core facts of the news event;
• the time, place, people, and outcome of the event;
• the latest developments and authoritative responses;
• statements by important figures that carry substantive information;
• news scenes, interviews, press conferences, surveillance footage, and related effective material;
• key footage that directly explains how the event unfolded, its outcome, or its impact.

All retained content must serve the core news storyline.

Content removal rules

Remove the following content:

• advertisements and commercial promotions;
• program promotions, channel packaging, and opening and closing credits;
• host pleasantries and transitional segments with no informational value;
• repetitive statements and repeated footage;
• dead pauses, filler words, and obvious gaps;
• lengthy background unrelated to the core event;
• secondary content that does not affect understanding of the news;
• segments that cannot be verified, are ambiguous in meaning, or are likely to cause misunderstanding.

Rules for editing statements by people

• Statements by people must remainsemantically complete。
• Prefer to cut at the following points:
  • after a complete sentence ends;
  • at natural pauses by the speaker;
  • at points where the content of the statement clearly shifts;
  • at natural shot transitions.
• Do notcut forcibly in the middle of a sentence,Do notretain only part of a statement in a way that changes the original meaning,Do notincorrectly splice together statements made at different times or in different contexts.
• If a statement is long, repetitive, vague, or irrelevant sentences may be removed, but the retained content must be able toexpress a complete meaning independently。
• Use word-level editing tools (transcripts) to delete and edit by word, ensuring cut points fall on complete sentence boundaries.

Factual and logical requirements

Editing must be faithful to the original news material and must not:

• change the original meaning of statements by people;
• exaggerate or understate facts;
• incorrectly associate different events;
• create false causal relationships through shot splicing;
• present speculative content as established fact;
• use footage that does not correspond to the news event to mislead viewers;
• delete necessary causes and consequences in pursuit of pacing.

Audio requirements

• Do not addany background music, voiceover, narration, sound effects, transition sound effects, or other external sounds.
• Retain only the human voices and necessary on-site sounds in the original material that are directly related to the news content.
• Remove advertising music, program packaging music, and sounds unrelated to the core news.
• When handling cut points, ensure the original human voices connect naturally, avoiding sudden cutoffs, overlaps, clipping, or obvious volume jumps (simply fine-tune the cut points with edit_item fadeInSeconds/fadeOutSeconds; do not add music).

Overall style

• An objective, formal, compact, and clear informational news style.
• The editing rhythm is determined by the news content; do not deliberately pursue a fixed duration or high-frequency shot changes.
• Ensure every retained segment has clear informational value, and increase information density while keeping the content complete.
• No captions/filters/transition effects; a basic cross-dissolve may be used to avoid hard cuts, but restraint in the style of news is the principle.

Corresponding OpenChatCut tools

• read_project / read_timeline: first read the project and timeline state, and record the allowed sources within the above scope sourceAssetId and src; it is forbidden to automatically include all assets in the media pool.
• transcribe_track + transcript word-level editing: cut speech at complete semantic boundaries, remove filler words/repetitions.
• view_timeline_frames: verify the picture content and key on-site footage.
• edit_item(trim / ripple delete / fade) and split_item(split): edit according to the keep/delete rules and handle the continuity at cut points.
• edit_track: when multi-track organization is needed, create, adjust, or tighten the voice track and the on-site sound track.
• Before output, once again read_project, confirm that all final video / audio items come from allowed sources, and that there is no BGM, sound effects, dubbing, or narration; then use preview/export preflight to verify duration and content completeness.

Reference files

• When executing the final cut, follow the fixed deterministic workflow and the mandatory acceptance checklist, see references/deterministic-execution-and-acceptance.md: environment check → media parsing → generate a single edit plan → one-time execution → item-by-item final inspection and acceptance (at least 1 video track clip, no newly added external audio after the final cut, complete speech semantics, natural continuity at cut points).