davinci-edit
/davinci-edit
Składa i poprawia dokumenty, relacje z wydarzeń, promo i montaże muzyczne w DaVinci Resolve: tempo, dobór ujęć, dialogi i napisy.
- Autor
- amartinez1558 ↗
- Repozytorium
- davinci-workflow-skills ↗
- Licencja
- MIT
- Ostatnia zmiana repozytorium
- 09.11.2026
Ocena i bezpieczeństwo
- Pliki w paczce
- 2
- Skrypty
- 0
- Komentarze HTML
- 0
- Adresy URL w SKILL.md
- 0
Brak sygnałów. Repozytorium zmienione 09.11.2026, bez odwołań do wycofanych modeli i API.
MIT, otwarta: można używać i zmieniać, z zachowaniem informacji o autorze.
Co robi i kiedy
Składa i poprawia dokumenty, relacje z wydarzeń, promo i montaże muzyczne w DaVinci Resolve: tempo, dobór ujęć, dialogi i napisy.
Oryginalny opis (EN)
Use when assembling or revising documentaries, event recaps, promos, music montages, or mixed-format videos in DaVinci Resolve, including pacing, action selection, dialogue, captions, and manual edits.
Jak działa
- Establish the edit boundary
- Choose the editing approach
- Shape the story
- Handle gaps honestly
- Make text readable and editable
- Verify the revision
Podgląd SKILL.md
--- name: davinci-edit description: Use when assembling or revising documentaries, event recaps, promos, music montages, or mixed-format videos in DaVinci Resolve, including pacing, action selection, dialogue, captions, and manual edits. metadata: version: '0.2.0' --- # DaVinci edit Build the requested story while preserving the user's current work. Treat the current brief and later corrections as authority; examples and historic feedback inform judgment rather than becoming fixed settings for every film. ## Establish the edit boundary - Read the brief, approved script, review notes, available-media inventory, and existing request audit. Separate spoken words, picture suggestions, production instructions, and unresolved questions. Do not silently turn a suggestion in an attached document into an instruction. - Inspect the live project and selected timeline before editing, duplicating, or re-exporting. Save a recoverable project checkpoint and snapshot timeline identity, frame rate/start, duration, track layout, picture edits, source ranges, captions, titles, audio, and relevant effects. A saved plan or old project export may predate manual changes. - Build the next version from that live state. Preserve manual edits and prior versions; an export-only request uses the latest live edit rather than rebuilding an older schedule. Do not render when the user explicitly requests a saved edit without export. - Map review timestamps to the reviewed version. If timings changed, use dialogue/shot anchors to map them into the new version and label both time domains. Keep each comment traceable as fixed, partial, pending asset, or needing clarification. - When simplifying video tracks, move existing native items into non-conflicting tracks while preserving source ranges, precise timing, stacking/composite order, transitions, retimes, grades and Fusion effects. Retain the layers needed for sharp portrait foregrounds over blurred backgrounds. Delete tracks only after they are empty, and compare the visible result with the live checkpoint rather than flattening or rebuilding the edit. ## Choose the editing approach Use the brief to decide what leads each section: speech and meaning for a dialogue-led story; action, reactions and musical phrases for a recap or montage; a combination for mixed videos. A script, interview introduction, caption track or dialogue trim table is not required for every video. Apply the dialogue-specific rules below where speech leads. For music-led sequences, read [recap and montage editing](references/recap-and-montage.md). ## Shape the story - Follow the approved dialogue order and retain breathing room around meaningful thoughts. Put a person's story-specific B-roll after their introduction unless the brief deliberately calls for a preview. - Choose cutaways for the exact spoken idea, person, place, event, and action. A visually attractive shot or a different filename is insufficient. Cover visible interview edits with relevant picture; retain the interview face if the available cutaway would mislead or distract. - Build purposeful action sequences: an exterior approach, settling at a table, indoor work, then reading can establish progression when the narration supports it. Avoid returning to an earlier action stage without a clear narrative reason. Let complete actions or emotional moments hold. - When feedback asks for quicker cutting, shorten redundant establishing shots and idle portions, especially near the opening. Judge the whole sequence at playback speed; do not replace purposeful holds with an arbitrary universal shot length. - Avoid repeating the same source or perceptually similar setup when the brief asks for no repeats. Reserve a shot for its strongest narrative match. Prefer genuinely different action, scale, angle, or context over filename variety. - Prefer high-frame-rate sources for slow motion; generally do not slow 24/30 fps footage (including 23.976/29.97). If low-frame-rate footage must be slowed, set the clip's Retime Process to Optical Flow, verify the setting, and check rendered playback for smooth motion and interpolation artifacts. Verify source handles, internal cuts, and retimed duration; do not slow synchronized speech or silently change interview timing. - For Ken Burns effects, use the original still image with native Dynamic Zoom rather than pre-rendering an animated video. Preserve editable start/end framing and inspect the crop and motion in playback. ## Handle gaps honestly Keep a concrete remaining-footage/recording audit: version-relative timestamp or range, priority, requested person/event/place/action, why needed, acceptable alternatives, and current status. Ask for exact missing coverage rather than filling gaps with unrelated people, décor, food, music performance, or a different gathering. Use visible draft placeholders when the brief calls for them; a text overlay on a black 50% background is an available requested treatment, not a universal final-film style. Label the actual missing asset, preserve required pending montage/recording slots, and keep the audit synchronized. If no placeholder treatment is requested, retain relevant interview coverage and report the gap. Never imply that illustrative relationship footage depicts an unverified historic event. ## Make text readable and editable - Preserve approved spoken wording in a native editable subtitle track and matching sidecar when required. Keep burned captions and editable captions aligned if a captioned review export is requested. - Prefer one-line cues grouped into complete phrases or clauses when that is the brief or established preference. Avoid orphan words and awkward cue boundaries. Measure rendered width within safe margins and check reading time against delivery pace; do not copy a previous film's font size or character limit blindly. - A translucent dark caption backing can provide contrast without a black text outline; do not add a stroke automatically. Honor a no-outline preference in the native track style, any clip overrides, and external caption-burn settings while retaining the backing and safe placement. - Coordinate titles/name cards through [davinci-titles](../davinci-titles/SKILL.md) when available. Independently preserve approved wording and identities, inspect actual lower-third/caption clearance, and suppress only exact title-duplicate cues when requested, for the interval fully represented by the card. Retain adjacent dialogue and a full transcript. - For dialogue trimming reviews, generate the requested table from the actual subtitle/script rows of the reviewed cut, with elapsed timestamps and a clear removal column. Preserve the recipient's schema and wording; distinguish deleting a spoken passage from hiding a caption. Regenerate or map review times after edits rather than applying an old table to a newer cut. ## Verify the revision Record source timecode/frame indices separately from timeline frames and elapsed review time. Resolve source handles, inclusive/exclusive out conventions, clip start offsets, speed changes, and exact frame rates before applying a cut list. Verify a known source frame and a timeline boundary before bulk edits. Review transitions, story order, subject introductions, source identity, caption boundaries, title duplication, headroom, and lower-third clearance in the live timeline and sampled rendered frames when a render is authorized. Landscape-only briefs require checking displayed orientation and rotation metadata, not container dimensions alone. Inspect exposure per shot; defer color-specific work to an appropriate grading workflow rather than assuming one correction fits all footage. When resuming after final manual tweaks, compare the live edit and latest user export with the prior milestone by timeline instance and source range. Preserve deliberate reframes, leveling, split source moments, transitions, fades and music slips; do not infer that a matching filename means matching content or that every manual offset is verified sync. Treat the latest user choices as the current baseline, without turning their exact settings into defaults for another film. Compare the result with the live snapshot: unrelated audio, manual B-roll changes, effects, and approved timing should survive. Reconcile every requested change and remaining asset gap before calling the edit ready. State which playback, listening, lip-sync, reading, and color checks actually ran; sampled frames do not establish full human approval. For the repository's supplemental handoff and operation conventions, read [the workflow contract](../../docs/workflow-contract.md) when available. This skill's essential rules do not depend on that file or on another skill being installed.
Pliki w paczce (2)
- SKILL.md 9 KB
- references/recap-and-montage.md 4 KB
Wersje
W bazie nie ma innych skilli o nazwie davinci-edit.
Migawka bazy z 13.11.2026 · wersja paczki 689bfe7e80bb (skrót SHA-256 całej paczki)
Pobrania w ostatnich 30 dniach: 0 · ranking
Zgłoś problem z tym skillem: szkodliwa treść, błędny opis albo prośba autora o usunięcie.
Podobne skille
- 90davinci-export
Eksportuje z DaVinci Resolve gotowe klipy i montaże, ponownie renderuje poprawki i sprawdza pliki dostawcze oraz napisy. - 100davinci-titles
Projektuje i poprawia napisy ekranowe, tabliczki z nazwiskiem, pasy dolne i typografię w DaVinci Resolve na podstawie briefu lub szablonu. - 100make-video
Tworzy, edytuje i poprawia wideo w pakiecie Claude Resolve Editing Suite: cięcia, kolor, napisy, dźwięk, tempo i motion graphics. - 98davinci-resolve
Pomaga montować w DaVinci Resolve: ustawienia projektu, osi czasu, korekcję kolorów, Magic Mask, napisy i eksport pod media społecznościowe. - 100resolve-bridge
Łączy się z DaVinci Resolve Studio, zapisuje każdy krok montażu jako nową wersję osi czasu i wykrywa ręczne zmiany. - 98kinetic-film-to-resolve
Zamienia film typograficzny z kodu w edytowalną oś czasu DaVinci Resolve, z osobną kompozycją Fusion dla każdego ujęcia.