From b7007836b8589104cd2566d6b71c82e0e7c11b47 Mon Sep 17 00:00:00 2001 From: Emily2040 <191656017+Emily2040@users.noreply.github.com> Date: Sat, 30 May 2026 13:43:09 +1000 Subject: [PATCH] Release v5.4.4 professional filmmaker workflow --- CHANGELOG.md | 15 ++ README.md | 43 +++++- SKILL.md | 24 ++-- ...ommunity-patterns.seedance-2026-05-30.json | 36 +++++ data/sources.seedance-2026-05-30.json | 108 ++++++++++++++ evals/evals.json | 134 +++++++++++++++++- references/aspect-ratio-delivery.md | 48 +++++++ references/audio-guide.md | 6 + references/audio-post-delivery.md | 54 +++++++ references/cinematography-shot-language.md | 62 ++++++++ references/color-pipeline-aces.md | 58 ++++++++ references/delivery-qc.md | 54 +++++++ references/examples-by-mode.md | 20 +++ references/json-schema.md | 59 +++++++- references/multilingual-community-examples.md | 32 +++++ references/pro-filmmaking-standards.md | 52 +++++++ references/research-2026-05-30.md | 9 +- references/shot-list-continuity.md | 75 ++++++++++ references/source-registry.md | 7 + references/subtitles-localization.md | 54 +++++++ scripts/design_audit.py | 3 + scripts/eval_schema_check.py | 8 ++ scripts/validate_skills.py | 14 +- skills/seedance-antislop/SKILL.md | 2 +- skills/seedance-audio/SKILL.md | 6 +- skills/seedance-camera/SKILL.md | 6 +- skills/seedance-characters/SKILL.md | 4 +- skills/seedance-copyright/SKILL.md | 2 +- skills/seedance-examples-zh/SKILL.md | 2 +- skills/seedance-filter/SKILL.md | 2 +- skills/seedance-interview-short/SKILL.md | 2 +- skills/seedance-interview/SKILL.md | 7 +- skills/seedance-lighting/SKILL.md | 4 +- skills/seedance-motion/SKILL.md | 4 +- skills/seedance-pipeline/SKILL.md | 16 ++- skills/seedance-prompt-short/SKILL.md | 2 +- skills/seedance-prompt/SKILL.md | 7 +- skills/seedance-recipes/SKILL.md | 7 +- skills/seedance-style/SKILL.md | 2 +- skills/seedance-troubleshoot/SKILL.md | 5 +- skills/seedance-vfx/SKILL.md | 2 +- skills/seedance-vocab-es/SKILL.md | 2 +- skills/seedance-vocab-ja/SKILL.md | 2 +- skills/seedance-vocab-ko/SKILL.md | 2 +- skills/seedance-vocab-ru/SKILL.md | 2 +- skills/seedance-vocab-zh/SKILL.md | 2 +- 46 files changed, 1002 insertions(+), 65 deletions(-) create mode 100644 references/aspect-ratio-delivery.md create mode 100644 references/audio-post-delivery.md create mode 100644 references/cinematography-shot-language.md create mode 100644 references/color-pipeline-aces.md create mode 100644 references/delivery-qc.md create mode 100644 references/pro-filmmaking-standards.md create mode 100644 references/shot-list-continuity.md create mode 100644 references/subtitles-localization.md diff --git a/CHANGELOG.md b/CHANGELOG.md index 42a6cc8..7ca62b7 100644 --- a/CHANGELOG.md +++ b/CHANGELOG.md @@ -2,6 +2,21 @@ All notable changes to this project are documented here. +## [5.4.4] — 2026-05-30 + +### Added + +- Added a professional filmmaker reference layer: `pro-filmmaking-standards.md`, `cinematography-shot-language.md`, `shot-list-continuity.md`, `color-pipeline-aces.md`, `aspect-ratio-delivery.md`, `subtitles-localization.md`, `audio-post-delivery.md`, and `delivery-qc.md`. +- Added README professional scope for directors, DPs, producers, editors, colorists, sound teams, localization teams, and delivery/QC teams. +- Added eval coverage for shot contracts, multi-shot continuity, ACES/color handoff, aspect-ratio delivery, subtitles/localization, audio post, QC preflight, and global campaign versioning. +- Added professional workflow source records and community-pattern records for shot contracts, textless localization, and campaign cutdown matrices. + +### Changed + +- Routed the root skill, pipeline, interview, prompt, camera, motion, characters, lighting, audio, recipes, and troubleshooting skills into the new professional production references. +- Expanded JSON schema support for production phase, shot lists, continuity anchors, color pipeline, subtitle plan, audio deliverables, delivery metadata, and QC checks. +- Expanded examples with professional shot-contract, localization-handoff, and delivery-QC examples. + ## [5.4.3] — 2026-05-30 ### Added diff --git a/README.md b/README.md index 3093483..e1b9cdb 100644 --- a/README.md +++ b/README.md @@ -6,17 +6,17 @@ Text-to-video · Image-to-video · Video-to-video · Reference-to-video · Audio-aware prompting · Copyright-safe rewrites · Agent Skills -[![Version](https://img.shields.io/badge/version-5.4.3-111827?labelColor=0f172a)](#changelog) +[![Version](https://img.shields.io/badge/version-5.4.4-111827?labelColor=0f172a)](#changelog) [![Skills](https://img.shields.io/badge/sub--skills-23-0ea5e9?labelColor=0f172a)](#skill-map) -[![References](https://img.shields.io/badge/references-32-8b5cf6?labelColor=0f172a)](#reference-library) -[![Evals](https://img.shields.io/badge/evals-43-22c55e?labelColor=0f172a)](#validation) +[![References](https://img.shields.io/badge/references-40-8b5cf6?labelColor=0f172a)](#reference-library) +[![Evals](https://img.shields.io/badge/evals-51-22c55e?labelColor=0f172a)](#validation) [![License](https://img.shields.io/badge/license-MIT-f59e0b?labelColor=0f172a)](LICENSE) Author: [Iamemily2050 (@iamemily2050)](https://github.com/Emily2040) · [Instagram](https://instagram.com/iamemily2050) · [X](https://x.com/iamemily2050) · [Website](https://iamemily2050.com) Platform context: [ByteDance Seedance 2.0](https://seed.bytedance.com/en/seedance2_0) · Dreamina · Jimeng · Doubao · [Volcengine Ark](https://www.volcengine.com/docs/82379/2291680?lang=zh) · [BytePlus ModelArk](https://docs.byteplus.com/en/docs/ModelArk/2291680) · [Runway Seedance 2](https://docs.dev.runwayml.com/guides/seedance/) -Updated: **2026-05-30** · **v5.4.3 cinematic infographic refresh, multilingual community examples, and safe mixed-language repair patterns** +Updated: **2026-05-30** · **v5.4.4 professional filmmaker workflow expansion** --- @@ -36,11 +36,29 @@ This skill package turns Seedance 2.0 work into a repeatable assistant workflow: - Keeps model and platform claims source-dated so API, pricing, region, quota, and model-ID details are not guessed. - Provides deeper multilingual cinematic vocabulary in Chinese, Japanese, Korean, Spanish, and Russian, including role binding, first/last-frame phrasing, edit/extend wording, safety wording, and audio cues. - Adds original community-informed examples for Chinese-English, Russian-English, Japanese-English, Korean-English, and Spanish-English prompt structures. +- Adds professional filmmaker workflows for treatment-to-shot-list planning, shot contracts, continuity ledgers, ACES/color handoff, audio post, subtitles/localization, aspect-ratio variants, campaign cutdowns, delivery/QC, and client review packets. - Handles safe false-positive repairs by clarifying benign production context, not by hiding unsafe intent. - Rewrites unsafe celebrity, protected IP, private-person, brand, logo, song, or voice requests into safer creative equivalents. - Diagnoses failed outputs with concrete repair levers: camera, lighting, motion, reference role, duration, framing, audio, or safety wording. - Ships validation scripts, eval cases, source data, and design checks so maintainers can review changes before release. +## Professional Filmmaker Scope + +This package is designed for working film and commercial teams, not only for casual prompt writing. It can help an agent produce the artifact the role actually needs: + +| Role | What the skill should produce | +|---|---| +| Director | treatment, scene beat, performance intent, coverage, shot endpoint, review notes | +| Cinematographer / DP | shot contract, shot size, lens feel, camera support, movement, blocking, lighting continuity | +| Producer / agency | client brief, rights map, approval gates, campaign variants, risk log, review packet | +| Editor | selects plan, edit/extend decision, continuity handoff, handles, textless needs, conform notes | +| Colorist | color intent, ACES-aware handoff, show-look notes, HDR/SDR caveats, product-color checks | +| Sound team | dialogue map, ambience/SFX/music layers, sync cues, stems, M&E, dubbing and loudness notes | +| Localization team | subtitles, SDH captions, forced narratives, dubbing guide, market copy, textless plates | +| Delivery/QC | frame rate, aspect ratio, crop, color, loudness, captions, metadata, naming, human QC checklist | + +For these requests, the skill should not stop at a single prompt. It should return the production object first, then the Seedance prompt or prompt batch that fits inside that plan. + ## Start Here | User situation | Load first | Output | @@ -52,6 +70,9 @@ This skill package turns Seedance 2.0 work into a repeatable assistant workflow: | “Use this as first frame and that as final frame.” | [`first-last-frame-guide`](references/first-last-frame-guide.md) | A continuous transition with endpoint locks. | | “It failed or looks bad.” | [`seedance-troubleshoot`](skills/seedance-troubleshoot/SKILL.md) | A root-cause diagnosis and repaired prompt. | | “This uses a character, brand, celebrity, or real person.” | [`seedance-copyright`](skills/seedance-copyright/SKILL.md) | A safer rewrite preserving the creative function. | +| “I need this for a film, client, campaign, or delivery.” | [`pro-filmmaking-standards`](references/pro-filmmaking-standards.md) | A professional workflow plan, role-specific artifact, and prompt path. | +| “Turn this treatment into shots.” | [`shot-list-continuity`](references/shot-list-continuity.md) | Shot list, continuity ledger, and prompt batch structure. | +| “This needs subtitles, dubbing, color, sound, or QC.” | [`delivery-qc`](references/delivery-qc.md) | Post, localization, audio, color, and delivery checks. | | “I need API, Runway, pricing, model ID, or production workflow guidance.” | [`api-workflow`](references/api-workflow.md) | A source-gated operational checklist. | | “Is this Seedance Pro/Fast/V2?” | [`model-name-map`](references/model-name-map.md) | Source-dated naming and surface caveats. | | “I want Chinese/Russian/Japanese/Korean/Spanish or mixed-language prompt examples.” | [`multilingual-community-examples`](references/multilingual-community-examples.md) | Safe community-informed structures and false-positive repair patterns. | @@ -76,14 +97,16 @@ The v5.4 release line adds a dated research layer for safer data mining and plat - [`model-name-map.md`](references/model-name-map.md) prevents `Seedance 2.0`, `Seedance 2.0 Fast`, `Seedance V2`, and ambiguous Pro labels from being mixed together. - [`community-source-methodology.md`](references/community-source-methodology.md) explains how to mine public prompt corpora without copying unsafe examples. - [`multilingual-community-examples.md`](references/multilingual-community-examples.md) captures safe mixed-language and localized prompt structures from community pattern mining. +- [`pro-filmmaking-standards.md`](references/pro-filmmaking-standards.md) adds industry workflow boundaries for shot lists, continuity, color, audio, localization, and delivery. ## Operating System At A Glance ![Seedance 2.0 Skill OS infographic: source registry, prompt router, multimodal references, safety gates, and eval loop](assets/skill-os-infographic.png) -The visual map is intentionally text-light so it remains clean in GitHub and reusable in docs. It represents the five lanes this package keeps separate: +The visual map is intentionally text-light so it remains clean in GitHub and reusable in docs. It represents the six lanes this package keeps separate: - Research sources: dated official, academic, platform, and community evidence. +- Production spine: brief, shot list, continuity, post handoff, localization, and delivery/QC. - Prompt router: interview, prompt writing, compression, recipes, and troubleshooting. - Multimodal references: image, video, audio, first-frame, last-frame, and role-bound assets. - Safety gates: IP, likeness, voice, brand, real-person, filter, and platform-policy checks. @@ -140,6 +163,14 @@ The visual map is intentionally text-light so it remains clean in GitHub and reu | [`research-2026-05-30.md`](references/research-2026-05-30.md) | Dated source and field-observation snapshot. | | [`agent-compatibility.md`](references/agent-compatibility.md) | Agent Skills structure, Codex compatibility, and packaging notes. | | [`api-workflow.md`](references/api-workflow.md) | Volcengine, BytePlus, Runway, async task, reference-file, pricing, and production workflow checklist. | +| [`pro-filmmaking-standards.md`](references/pro-filmmaking-standards.md) | Professional production spine and source boundaries for film, commercial, post, localization, and delivery work. | +| [`cinematography-shot-language.md`](references/cinematography-shot-language.md) | Shot contracts, shot size, lens feel, camera support, movement, blocking, and coverage language. | +| [`shot-list-continuity.md`](references/shot-list-continuity.md) | Treatment-to-shot-list workflow, continuity ledger, and professional handoff fields. | +| [`color-pipeline-aces.md`](references/color-pipeline-aces.md) | ACES-aware color intent, show-look notes, HDR/SDR handoff, and color QC boundaries. | +| [`aspect-ratio-delivery.md`](references/aspect-ratio-delivery.md) | Creative framing, delivery containers, social cutdowns, safe areas, and textless/version planning. | +| [`subtitles-localization.md`](references/subtitles-localization.md) | Subtitle, SDH, forced narrative, dubbing, textless, and cultural localization planning. | +| [`audio-post-delivery.md`](references/audio-post-delivery.md) | Dialogue, SFX, music, stems, M&E, loudness, dubbing, and sync handoff guidance. | +| [`delivery-qc.md`](references/delivery-qc.md) | Professional preflight for picture, color, audio, captions, rights, metadata, versioning, and human QC. | | [`examples-by-mode.md`](references/examples-by-mode.md) | Mode-specific prompt examples for T2V, I2V, V2V, R2V, FLF2V, edit, extend, and troubleshooting. | | [`multilingual-community-examples.md`](references/multilingual-community-examples.md) | Original Chinese, Russian, Japanese, Korean, Spanish, and mixed-language prompt structures from safe community pattern mining. | | [`platform-surface-matrix.md`](references/platform-surface-matrix.md) | Model-vs-surface claim boundaries. | @@ -214,7 +245,7 @@ The v5.4 front page uses a generated cinematic bitmap hero, a generated operatin ## Changelog -See [`CHANGELOG.md`](CHANGELOG.md). Current release: **v5.4.3**. +See [`CHANGELOG.md`](CHANGELOG.md). Current release: **v5.4.4**. ## License diff --git a/SKILL.md b/SKILL.md index 574e3f7..89154e8 100644 --- a/SKILL.md +++ b/SKILL.md @@ -5,7 +5,7 @@ license: MIT user-invocable: true tags: [seedance] metadata: - version: "5.4.3" + version: "5.4.4" --- # seedance-20 @@ -14,15 +14,16 @@ Seedance 2.0 operating loop for agent-directed video work. Use this root skill t ## Operating Loop -1. Intake: identify the user's goal, target surface, mode, duration, aspect ratio, references, audio needs, and safety/IP risks. +1. Intake: identify the user's goal, production phase, target surface, mode, duration, aspect ratio, references, audio needs, deliverables, and safety/IP risks. 2. Source gate: before platform claims, load `[ref:api-status]` and `[ref:source-registry]`. For Runway or Volcengine specifics, also load `[ref:platform-surface-matrix]`. -3. Mode gate: choose T2V, I2V, V2V, R2V, FLF2V, edit, extend, or troubleshoot before writing prose. -4. Reference map: assign every asset one primary role: identity, first frame, last frame, product, environment, motion, camera, timing, audio, or style. State what must not transfer. -5. Multilingual gate: if the prompt uses Chinese, Russian, Japanese, Korean, Spanish, or code-mixed wording, load `[ref:multilingual-community-examples]` and preserve reference tags exactly. -6. Safety gate: route IP, likeness, voice, brand, real-person, graphic, or evasion-like wording through `[skill:seedance-copyright]` or `[skill:seedance-filter]`. -7. Prompt build: route to `[skill:seedance-interview]`, `[skill:seedance-prompt]`, `[skill:seedance-prompt-short]`, or a domain skill for camera, motion, audio, characters, VFX, style, recipes, or pipeline. -8. Quality pass: run anti-slop, check one visible beat, one primary camera move, physical light, sound intent, constraints, and source-date caveats. -9. Repair loop: if output fails, diagnose root cause before adding adjectives; use `[skill:seedance-troubleshoot]`. +3. Professional gate: if the user asks for film, ad, campaign, client, delivery, localization, color, sound, subtitle, post, QC, or multi-shot work, load `[ref:pro-filmmaking-standards]` before drafting. +4. Mode gate: choose T2V, I2V, V2V, R2V, FLF2V, edit, extend, or troubleshoot before writing prose. +5. Reference map: assign every asset one primary role: identity, first frame, last frame, product, environment, motion, camera, timing, audio, or style. State what must not transfer. +6. Multilingual gate: if the prompt uses Chinese, Russian, Japanese, Korean, Spanish, or code-mixed wording, load `[ref:multilingual-community-examples]` and preserve reference tags exactly. +7. Safety gate: route IP, likeness, voice, brand, real-person, graphic, or evasion-like wording through `[skill:seedance-copyright]` or `[skill:seedance-filter]`. +8. Prompt build: route to `[skill:seedance-interview]`, `[skill:seedance-prompt]`, `[skill:seedance-prompt-short]`, or a domain skill for camera, motion, audio, characters, VFX, style, recipes, or pipeline. +9. Quality pass: run anti-slop, check one visible beat, one primary camera move, physical light, sound intent, continuity anchors, constraints, delivery caveats, and source-date caveats. +10. Repair loop: if output fails, diagnose root cause before adding adjectives; use `[skill:seedance-troubleshoot]`. ## Load Map @@ -30,12 +31,15 @@ Seedance 2.0 operating loop for agent-directed video work. Use this root skill t |---|---| | Vague idea or missing brief | `[skill:seedance-interview]` or `[skill:seedance-interview-short]` | | Production prompt | `[skill:seedance-prompt]`, `[ref:quick-ref]`, `[ref:prompt-examples]` | +| Professional film, commercial, campaign, or delivery workflow | `[ref:pro-filmmaking-standards]`, `[ref:shot-list-continuity]`, `[ref:delivery-qc]` | | Compact prompt or Chinese compression | `[skill:seedance-prompt-short]`, language vocab reference | +| Camera, lens, blocking, shot contract | `[skill:seedance-camera]`, `[ref:cinematography-shot-language]` | | Image reference / first frame | `[ref:i2v-guide]`, `[ref:reference-workflow]` | | First and last frame | `[ref:first-last-frame-guide]` | | API, Runway, Volcengine, workflow, pricing, model IDs | `[skill:seedance-pipeline]`, `[ref:api-workflow]`, `[ref:model-name-map]` | +| Color, ACES, HDR/SDR, aspect ratio, subtitles, audio post, or QC | `[ref:color-pipeline-aces]`, `[ref:aspect-ratio-delivery]`, `[ref:subtitles-localization]`, `[ref:audio-post-delivery]`, `[ref:delivery-qc]` | | Genre template or examples | `[skill:seedance-recipes]`, `[ref:examples-by-mode]`, `[ref:genre-guides]` | | Chinese/Russian/Japanese/Korean/Spanish or mixed-language examples | `[ref:multilingual-community-examples]`, language vocab reference | | Bad result | `[skill:seedance-troubleshoot]` | -Preserve reference tags exactly, keep prompts short, and never convert field-observed community tricks into official platform guarantees. +Preserve reference tags exactly, keep prompts short, and never convert field-observed community tricks into official platform guarantees. For professional filmmaker requests, deliver the workflow object the role needs: shot list, shot contract, continuity ledger, prompt, post handoff, localization plan, or QC checklist. diff --git a/data/community-patterns.seedance-2026-05-30.json b/data/community-patterns.seedance-2026-05-30.json index 11df766..22c905c 100644 --- a/data/community-patterns.seedance-2026-05-30.json +++ b/data/community-patterns.seedance-2026-05-30.json @@ -204,6 +204,42 @@ "risk_flags": ["ip-risk", "brand-risk", "real-person-risk"], "features": ["large raw prompt gallery", "mixed languages", "dialogue and audio blocks", "many unsafe identity or brand surfaces"], "safe_use": "Classify and rewrite into original authorized examples; extract structure only." + }, + { + "id": "professional-shot-contract", + "label": "safe-structure-only", + "source_url": "https://theasc.com/article/shot-craft-camera-movement/", + "source_language": "en", + "surface": "professional filmmaking", + "claim_status": "field-observed", + "confidence": "medium", + "risk_flags": [], + "features": ["shot size", "angle", "lens feel", "camera move", "start frame", "endpoint", "fragile anchors"], + "safe_use": "Use to convert vague cinematic language into one professional shot contract per Seedance clip." + }, + { + "id": "global-localization-textless", + "label": "safe-structure-only", + "source_url": "https://partnerhelp.netflixstudios.com/hc/en-us/articles/215758617-Timed-Text-Style-Guide-General-Requirements", + "source_language": "multi", + "surface": "professional localization", + "claim_status": "field-observed", + "confidence": "medium", + "risk_flags": [], + "features": ["textless plate", "subtitles in post", "dubbing guide", "M&E", "caption-safe framing"], + "safe_use": "Use for global filmmaker deliverables instead of asking Seedance to burn in small multilingual text." + }, + { + "id": "campaign-cutdown-matrix", + "label": "safe-structure-only", + "source_url": "https://www.thedpp.com/specs/", + "source_language": "en", + "surface": "professional delivery", + "claim_status": "field-observed", + "confidence": "medium", + "risk_flags": [], + "features": ["30s hero", "15s cutdown", "6s bumper", "9:16 vertical", "1:1 feed", "textless localized master"], + "safe_use": "Use to plan commercial variants and delivery notes while keeping platform specs recheck-required." } ] } diff --git a/data/sources.seedance-2026-05-30.json b/data/sources.seedance-2026-05-30.json index 15a1353..2d9fe12 100644 --- a/data/sources.seedance-2026-05-30.json +++ b/data/sources.seedance-2026-05-30.json @@ -494,6 +494,114 @@ "volatile_fields": ["language_support", "example_quality", "surface_behavior"], "repo_actions": ["add safe mixed-language repair boundary", "add multilingual examples"], "do_not_use_for": ["official language support list", "filter behavior guarantees"] + }, + { + "id": "asc-studiobinder-shot-language-2026-05", + "title": "Professional Shot Language And Shot List Fields", + "url": "https://theasc.com/article/shot-craft-camera-movement/", + "language": "en", + "source_type": "professional_workflow", + "retrieved_at": "2026-05-30", + "published_or_updated": null, + "confidence": "medium", + "surfaces": ["filmmaking", "cinematography"], + "claims": [ + "professional camera language benefits from shot size, angle, camera movement, and subject relationship", + "shot lists should preserve production intent and handoff fields" + ], + "volatile_fields": ["article_availability"], + "repo_actions": ["add cinematography shot contract reference", "add professional shot eval"], + "do_not_use_for": ["Seedance platform capability"] + }, + { + "id": "screenskills-script-supervisor-continuity-2026-05", + "title": "Script Supervisor Continuity Practice", + "url": "https://www.screenskills.com/job-profiles/browse/film-and-tv-drama/technical/script-supervisor-film-and-tv-drama/", + "language": "en", + "source_type": "professional_workflow", + "retrieved_at": "2026-05-30", + "published_or_updated": null, + "confidence": "medium", + "surfaces": ["filmmaking", "continuity"], + "claims": [ + "continuity work tracks action, dialogue, camera, props, wardrobe, and scene consistency", + "AI video shot plans should keep a continuity ledger across prompts" + ], + "volatile_fields": ["article_availability"], + "repo_actions": ["add shot-list continuity reference", "add multi-shot continuity eval"], + "do_not_use_for": ["Seedance model behavior guarantee"] + }, + { + "id": "aces-color-management-2026-05", + "title": "ACES Overview And AMF Specification", + "url": "https://docs.acescentral.com/background/overview/", + "language": "en", + "source_type": "professional_standard", + "retrieved_at": "2026-05-30", + "published_or_updated": null, + "confidence": "high", + "surfaces": ["color", "post"], + "claims": [ + "ACES provides color-management vocabulary and metadata handoff concepts", + "AMF can carry ACES metadata through professional workflows" + ], + "volatile_fields": ["spec_version"], + "repo_actions": ["add color pipeline reference", "add ACES no-hallucination eval"], + "do_not_use_for": ["claiming Seedance output is ACES-certified"] + }, + { + "id": "subtitles-localization-standards-2026-05", + "title": "Timed Text, WebVTT, And Caption Accessibility Sources", + "url": "https://partnerhelp.netflixstudios.com/hc/en-us/articles/215758617-Timed-Text-Style-Guide-General-Requirements", + "language": "en", + "source_type": "professional_standard", + "retrieved_at": "2026-05-30", + "published_or_updated": null, + "confidence": "medium", + "surfaces": ["localization", "accessibility"], + "claims": [ + "subtitles, SDH, forced narrative, and captions should be planned as post deliverables", + "generated burned-in text is not a substitute for professional timed text" + ], + "volatile_fields": ["buyer_requirements", "language_rules"], + "repo_actions": ["add subtitles localization reference", "add localization eval"], + "do_not_use_for": ["universal delivery requirement"] + }, + { + "id": "audio-loudness-post-delivery-2026-05", + "title": "Professional Audio Loudness And Mix Delivery Sources", + "url": "https://www.itu.int/rec/R-REC-BS.1770", + "language": "en", + "source_type": "professional_standard", + "retrieved_at": "2026-05-30", + "published_or_updated": null, + "confidence": "medium", + "surfaces": ["audio", "post", "delivery"], + "claims": [ + "professional audio delivery requires loudness, sync, mix, and stem checks", + "M&E and stems should be planned when localization is required" + ], + "volatile_fields": ["buyer_targets", "spec_version"], + "repo_actions": ["add audio post delivery reference", "add loudness and stems eval"], + "do_not_use_for": ["claiming Seedance produces final compliant mixes"] + }, + { + "id": "delivery-qc-imf-dpp-movielabs-2026-05", + "title": "Professional Delivery And QC Sources", + "url": "https://www.smpte.org/standards/st2067", + "language": "en", + "source_type": "professional_standard", + "retrieved_at": "2026-05-30", + "published_or_updated": null, + "confidence": "medium", + "surfaces": ["delivery", "qc", "metadata"], + "claims": [ + "professional delivery requires picture, audio, subtitles, metadata, naming, rights, and human QC checks", + "IMF/DCP/social delivery assumptions must be separated from Seedance prompt design" + ], + "volatile_fields": ["buyer_specs", "spec_version", "platform_requirements"], + "repo_actions": ["add delivery QC reference", "add delivery preflight eval"], + "do_not_use_for": ["universal buyer spec"] } ] } diff --git a/evals/evals.json b/evals/evals.json index cc20582..f9294ee 100644 --- a/evals/evals.json +++ b/evals/evals.json @@ -1,5 +1,5 @@ { - "version": "5.4.3", + "version": "5.4.4", "updated": "2026-05-30", "cases": [ { @@ -668,6 +668,138 @@ "skills_expected_to_activate": [ "seedance-pipeline" ] + }, + { + "id": "professional_shot_contract", + "prompt": "I am a DP. Turn this coffee commercial idea into one precise Seedance shot: a hand sets the cup down and steam reveals the logo.", + "expected_output": "Professional shot contract with shot size, angle, lens feel, camera move, subject relation, start/end frame, action endpoint, and fragile anchors.", + "assertions": [ + "includes shot size angle lens feel camera move and endpoint", + "protects logo hand and product geometry as fragile anchors", + "does not answer with generic cinematic adjectives", + "routes to cinematography shot-language reference" + ], + "failure_mode": "generic camera prompt instead of a DP shot contract", + "skills_expected_to_activate": [ + "seedance-camera", + "seedance-prompt" + ] + }, + { + "id": "multi_shot_continuity_handoff", + "prompt": "Turn this short film beat into four Seedance shots and keep wardrobe, screen direction, eyeline, prop, and lighting continuous.", + "expected_output": "Shot list and continuity ledger with one endpoint per shot and handoff notes.", + "assertions": [ + "creates shot IDs and purpose for each shot", + "tracks wardrobe props eyeline screen direction and lighting state", + "uses final-frame or start-frame handoff language", + "does not cram the whole scene into one overloaded prompt" + ], + "failure_mode": "multi-shot continuity drift", + "skills_expected_to_activate": [ + "seedance-pipeline", + "seedance-recipes", + "seedance-characters", + "seedance-camera" + ] + }, + { + "id": "aces_color_pipeline_no_hallucination", + "prompt": "Make this Seedance sequence ACES compliant and ready for HDR and SDR delivery.", + "expected_output": "Two-layer answer: prompt-level color intent plus ACES-aware post handoff, without claiming Seedance output itself is ACES certified.", + "assertions": [ + "separates prompt color look from post color management", + "mentions ACES or AMF as handoff metadata not a prompt guarantee", + "tracks HDR SDR output transform or trim assumptions", + "does not hallucinate certification from the generated clip" + ], + "failure_mode": "pretends prompt wording guarantees ACES compliance", + "skills_expected_to_activate": [ + "seedance-lighting", + "seedance-pipeline" + ] + }, + { + "id": "aspect_ratio_delivery_surface", + "prompt": "We need 16:9, 9:16, 1:1, and 2.39 versions of the same product reveal.", + "expected_output": "Aspect-ratio and delivery-surface plan separating creative framing, crop strategy, safe areas, and version naming.", + "assertions": [ + "separates creative aspect ratio from delivery container", + "protects product and logo from edge crops", + "recommends textless or post-added typography for variants", + "does not assume every Seedance surface supports every ratio" + ], + "failure_mode": "single prompt ignores variant framing and delivery constraints", + "skills_expected_to_activate": [ + "seedance-pipeline", + "seedance-recipes", + "seedance-camera" + ] + }, + { + "id": "subtitle_localization_accessibility", + "prompt": "Prepare this dialogue scene for Spanish, Japanese, Korean, Russian, and Chinese subtitles plus SDH captions.", + "expected_output": "Localization plan with subtitle/SDH/forced-narrative boundaries, caption-safe framing, textless plate, and no burned-in generated text.", + "assertions": [ + "distinguishes subtitles SDH forced narrative and dubbing", + "keeps lower third or safe area clear for captions", + "requires textless plate or post-added text", + "does not ask Seedance to generate final tiny multilingual subtitles" + ], + "failure_mode": "treats generated on-screen text as professional subtitles", + "skills_expected_to_activate": [ + "seedance-audio", + "seedance-pipeline", + "seedance-prompt" + ] + }, + { + "id": "audio_post_loudness_stems", + "prompt": "This ad needs dialogue, SFX, music, M&E, stems, and a loudness check for delivery.", + "expected_output": "Audio post plan separating prompt-level sound cues from final mix, stems, M&E, dubbing, sync, and loudness verification.", + "assertions": [ + "includes dialogue ambience SFX music or silence layers", + "plans stems or M&E when localization is needed", + "states loudness target must follow buyer or platform spec", + "does not promise Seedance creates a certified final mix" + ], + "failure_mode": "confuses prompt audio with final delivery mix", + "skills_expected_to_activate": [ + "seedance-audio", + "seedance-pipeline" + ] + }, + { + "id": "delivery_qc_preflight", + "prompt": "Before I send the Seedance spot to a client, give me a final QC checklist.", + "expected_output": "Delivery preflight covering picture, color, audio, captions, continuity, rights, metadata, versioning, naming, and human review.", + "assertions": [ + "checks picture color audio text continuity rights metadata and versioning", + "includes human watch-through at normal speed", + "routes failures to edit color audio localization or regeneration appropriately", + "does not mark prompt output done without QC" + ], + "failure_mode": "client delivery without preflight", + "skills_expected_to_activate": [ + "seedance-pipeline", + "seedance-troubleshoot" + ] + }, + { + "id": "imf_localization_versioning", + "prompt": "Plan masters for a global campaign: hero, 15s cutdown, vertical, textless, localized subtitles, and archive metadata.", + "expected_output": "Versioning and delivery plan for hero/cutdown/social/textless/localized/archive assets with spec caveats.", + "assertions": [ + "creates a campaign version matrix", + "includes textless and localized subtitle or dubbing deliverables", + "records model surface prompt version rights and approval metadata", + "states contracted platform or buyer specs must be rechecked" + ], + "failure_mode": "no localization or version-control plan for global campaign", + "skills_expected_to_activate": [ + "seedance-pipeline", + "seedance-recipes" + ] } ] } diff --git a/references/aspect-ratio-delivery.md b/references/aspect-ratio-delivery.md new file mode 100644 index 0000000..e771529 --- /dev/null +++ b/references/aspect-ratio-delivery.md @@ -0,0 +1,48 @@ +# Aspect Ratio And Delivery Surfaces + +Use this reference when planning professional framing, social cutdowns, broadcast/web deliverables, DCP/theatrical framing, or campaign versions. + +## Separate Creative Frame From Delivery Container + +| Layer | Decision | +|---|---| +| Creative aspect ratio | the composition the shot is designed for: 16:9, 9:16, 1:1, 4:5, 1.85, 2.39 | +| Platform or container | web, broadcast, DCP, IMF, social, presentation, pitch deck | +| Safe area | where faces, products, captions, and logos must remain readable | +| Reframe plan | crop, letterbox, pillarbox, alternate render, or edit-specific version | +| Text policy | on-screen text should usually be added in post, not generated into motion | + +## Common Ratios + +| Ratio | Use | Prompt advice | +|---|---|---| +| 16:9 | web, broadcast, presentation, horizontal campaigns | strongest default for cinematic scene work | +| 9:16 | vertical social, mobile ads | keep hero subject centered; avoid edge-critical action | +| 1:1 | social grid, product loops | symmetrical product framing works well | +| 4:5 | social feed ads | allow headroom and caption room | +| 1.85 | theatrical/narrative modern frame | compose for horizontal storytelling, not tall product text | +| 2.39 | widescreen drama, premium ads | protect faces/products from being too small | + +## Campaign Version Matrix + +| Version | Purpose | Planning notes | +|---|---|---| +| 30s hero | full story | more shots, full arc, strongest brand/product proof | +| 15s cutdown | paid social/broadcast | keep hook, product proof, end state | +| 10s bumper | recall | one action beat and final product or title card | +| 6s bumper | fast placement | one visual gag/reveal/product beat | +| 9:16 vertical | mobile | center action, avoid wide geography, captions lower third | +| 1:1 or 4:5 | feed | product and face readable without edge details | +| textless | localization/VFX | no burned-in copy, no captions, no end card text | + +## Prompt Framing Rules + +- Say the target ratio and composition goal early. +- Keep faces, products, and logos away from frame edges when multiple crops are needed. +- For vertical versions, simplify camera moves and avoid fast lateral action that exits frame. +- For 2.39 or very wide frames, keep the hero subject large enough to survive generation and review. +- For title cards, packshots, captions, and legal text, prefer clean visual plates plus post-added typography. + +## Delivery Caveat + +Always verify the active platform, buyer, broadcaster, festival, or ad network spec before final export. The skill may plan ratio and crop strategy, but the contract spec controls the master. diff --git a/references/audio-guide.md b/references/audio-guide.md index fe56daf..b589ef7 100644 --- a/references/audio-guide.md +++ b/references/audio-guide.md @@ -2,6 +2,8 @@ Use this reference for detailed audio, dialogue, beat-sync, ambience, and lip-sync workflows. Keep audio roles explicit and avoid promising exact platform behavior unless the active surface documents it. +For professional audio post, stems, M&E, dubbing, loudness, or delivery checks, also load `audio-post-delivery.md`. + ## Dialogue - Keep lines short, preferably one sentence per speaker turn. @@ -45,3 +47,7 @@ Use separate speaker turns when reliability matters. For two-person exchanges, g - Overbusy mix: choose ambience plus one key SFX; remove music if dialogue matters. - Lip-sync drift: use a locked medium close-up, no head turn, short quoted line, and simple expression. - Audio-reference conflict: mute the video reference, remove competing SFX/music, and describe one visible event per beat. + +## Post Handoff Boundary + +Prompt audio can shape performance and visible timing, but final mixes need post-production review. For paid or delivery work, record spoken language, subtitle/dubbing needs, M&E/stem needs, sync cues, and buyer loudness target separately from the prompt. diff --git a/references/audio-post-delivery.md b/references/audio-post-delivery.md new file mode 100644 index 0000000..e743edd --- /dev/null +++ b/references/audio-post-delivery.md @@ -0,0 +1,54 @@ +# Audio Post And Delivery + +Use this reference when Seedance work involves dialogue, sound design, music, stems, M&E, dubbing, sync, or final delivery checks. + +## Prompt-Level Audio + +Use prompt audio to direct story timing and mood: + +- dialogue: short, quoted, assigned to one speaker; +- ambience: room tone, traffic, rain, crowd bed, machine hum; +- SFX: one or two story-relevant sounds tied to visible action; +- music: broad tempo, energy, instrumentation, or silence; +- sync cue: door slam at endpoint, product click on final frame, light pulse on downbeat. + +Avoid exact reproduction claims for protected songs, real voices, or licensed performances unless rights and platform support are clear. + +## Post-Level Audio Deliverables + +| Deliverable | Purpose | +|---|---| +| Full mix | final stereo/5.1/Atmos or platform-required mix | +| Dialogue stem | spoken lines for edits, dubbing, and cleanup | +| Music stem | music-only layer | +| Effects stem | SFX and designed sound | +| M&E | music and effects without original dialogue for localization | +| Printmaster | approved final master mix | +| Dubbing guide | speaker timing, tone, pronunciation, pauses | +| Loudness report | delivery-target compliance evidence | + +## Audio Planning Template + +`Dialogue: [speaker/line/language]. Ambience: [bed]. SFX: [visible sync cue]. Music: [tempo/mood or none]. Reference: [Audio1 role]. Post: [stems/M&E/loudness/sync notes].` + +## Sync And Loudness Checks + +For professional delivery, check: + +- lip-sync against final picture; +- music/SFX sync after frame-rate conversion; +- no dialogue masked by music or effects; +- no unintended source-video audio in R2V workflows; +- loudness target and true peak per buyer/platform spec; +- M&E or stems exist if localization is required. + +## Common Repairs + +| Failure | Repair | +|---|---| +| dialogue desync | shorter line, locked framing, less head motion, one speaker | +| wrong speaker | tag the speaker and split turns | +| music overwhelms line | remove music during dialogue, keep room tone | +| audio reference ignored | map `[Audio1]` to tempo or mood and bind a visible event to it | +| video and audio refs conflict | mute video reference or assign video to camera only | +| localization impossible | plan M&E/stems and textless picture before final edit | diff --git a/references/cinematography-shot-language.md b/references/cinematography-shot-language.md new file mode 100644 index 0000000..b346d6c --- /dev/null +++ b/references/cinematography-shot-language.md @@ -0,0 +1,62 @@ +# Cinematography Shot Language + +Use this reference when a user needs director/DP-grade shot design instead of generic cinematic adjectives. + +## Shot Contract + +Every professional shot should declare: + +| Field | Prompt decision | +|---|---| +| Shot size | extreme wide, wide, medium wide, medium, medium close-up, close-up, extreme close-up, macro | +| Angle | eye-level, low angle, high angle, overhead, profile, over-shoulder, insert, product three-quarter | +| Lens feel | wide spatial energy, natural 35mm perspective, portrait compression, macro material detail | +| Camera support | locked-off, handheld, slider, dolly, crane, drone, gimbal, virtual product table | +| Movement | push-in, pull-back, lateral track, orbit, pan, tilt, pedestal, crane up, reveal, rack focus | +| Subject relation | camera follows, leads, discovers, holds, observes, or blocks with the subject | +| Start frame | first readable composition | +| End frame | changed state, pose, reveal, product hero, or continuity handoff | +| Fragile anchors | face, hands, logo, text, product shape, wardrobe, prop position | + +Prompt pattern: + +`Shot: [size/angle/lens feel]. Camera starts [composition], [one movement] at [speed] while [subject action], ending on [clear endpoint]. Preserve [fragile anchors].` + +## Shot Size Use + +| Shot size | Use for | Seedance caution | +|---|---|---| +| Extreme wide | scale, arrival, environment | small faces/logos will drift; do not demand facial acting | +| Wide | blocking, dance, movement, product-in-environment | keep action simple and readable | +| Medium | dialogue, product use, handoff | good default for character commercials | +| Close-up | emotion, texture, lip-sync, product detail | keep camera stable and action small | +| Macro | material, food, jewelry, mechanics | avoid large motion and text redraw | + +## Camera Movement Grammar + +- **Locked-off:** use for lip-sync, product identity, text, logos, precise VFX, and continuity anchors. +- **Dolly/push-in:** use for discovery, realization, intimacy, product reveal. +- **Lateral track:** use for travel, procession, choreography, passing foreground layers. +- **Orbit:** use for product hero and statuesque subjects; avoid if identity must remain stable from one angle only. +- **Crane/drone:** use for scale, arrival, geography, reveal; avoid for dialogue or tiny product text. +- **Handheld:** use for realism or tension; keep movement subtle when identity matters. +- **Rack focus:** use for attention shift between two anchored objects; avoid stacking with complex camera moves. + +## Blocking and Coverage + +For multi-shot scenes, keep camera grammar motivated: + +1. Establish geography and screen direction. +2. Move to character/product action. +3. Use close detail only when the detail changes the story. +4. End each clip with a frame that can become the next first-frame reference. + +Coverage pattern: + +`Shot 1 wide establishes [space/screen direction]. Shot 2 medium follows [action]. Shot 3 close-up reveals [consequence]. Continuity: same wardrobe, prop, light direction, and eyeline.` + +## Professional Avoid List + +Avoid: `cinematic camera`, `dynamic shot`, `epic zoom`, `film look`, `Hollywood style`, `beautiful composition` without physical shot choices. + +Replace with: shot size, support, movement, subject relation, motivated light, endpoint, and fragile anchors. diff --git a/references/color-pipeline-aces.md b/references/color-pipeline-aces.md new file mode 100644 index 0000000..a33bd12 --- /dev/null +++ b/references/color-pipeline-aces.md @@ -0,0 +1,58 @@ +# Color Pipeline And ACES Notes + +Use this reference when Seedance output must enter a professional edit, grade, HDR/SDR pass, agency review, or delivery workflow. + +## Honest Boundary + +Seedance prompts can describe color intent, lighting motivation, contrast, palette, material response, and mood. They cannot replace measured color management, calibrated monitoring, conform, grade, legal-range checks, or delivery transforms. Keep prompt language creative; keep pipeline language as metadata for post. + +## Prompt-Level Color Intent + +Use: + +- source light: tungsten practical, overcast daylight, sodium streetlight, neon sign, cool moon rim; +- contrast: soft low contrast, hard noir contrast, clean product contrast, high-key beauty; +- palette: restrained warm/cool split, muted winter palette, saturated music-video palette; +- material response: brushed metal highlight, skin rolloff, glossy acrylic reflection, wet asphalt speculars; +- transition: practical lamp warms the face, lightning briefly hardens the silhouette. + +Avoid: + +- unsupported claims such as exact ACES compliance from a prompt alone; +- impossible stacks like HDR Dolby Vision, 16mm, neon, bleach bypass, and pastel commercial all in one short shot; +- using LUT names as magic style words without describing the visible result. + +## Post Metadata To Track + +For professional handoff, record: + +| Field | Meaning | +|---|---| +| capture/source | generated source, reference clips, stills, source frame | +| working color space | project working assumption, often ACEScct/ACEScg or editor-managed alternative | +| IDT/source transform | how source media is interpreted, if applicable | +| show look | creative look description, LUT/CDL/LMT notes | +| output transform | SDR Rec.709, HDR PQ, theatrical/DCP, social platform conversion | +| trim pass | separate SDR/HDR/social review notes | +| QC notes | clipping, illegal levels, banding, skin tone, product color, logo color | + +## ACES-Friendly Handoff + +When a user asks for ACES, respond with a two-layer answer: + +1. Prompt: visible color and lighting instructions that Seedance can understand. +2. Handoff: ACES/AMF/color notes for the editor or colorist to verify outside Seedance. + +Example: + +`Prompt look: cool overcast daylight with a warm practical lamp reflected in the bottle, soft contrast, clean highlight rolloff, no crushed blacks. Post note: conform generated clip into the project color pipeline, verify source interpretation, preserve product color, create SDR Rec.709 and HDR trim review if required.` + +## Color Failure Repairs + +| Symptom | Repair | +|---|---| +| Flat image | add motivated key source, rim/separation, and one material highlight | +| Overprocessed color | reduce style names; specify natural contrast and neutral skin/product color | +| Inconsistent color across shots | repeat light direction, time of day, palette, and show-look note in every shot | +| Product color wrong | use I2V product reference, locked camera, and product-color preservation constraint | +| HDR/social mismatch | keep prompt neutral; plan separate grade/export versions in post | diff --git a/references/delivery-qc.md b/references/delivery-qc.md new file mode 100644 index 0000000..e39b6e0 --- /dev/null +++ b/references/delivery-qc.md @@ -0,0 +1,54 @@ +# Delivery And QC + +Use this reference when output must be reviewed, handed to a client, prepared for platform upload, localized, archived, or delivered as part of a professional campaign. + +## Delivery Preflight + +| Area | Check | +|---|---| +| Picture | frame rate, resolution, aspect ratio, crop, safe area, stabilization, flicker, banding | +| Color | working space, look notes, HDR/SDR target, product/brand color, legal range if required | +| Audio | sync, loudness, true peak, dialogue clarity, stems, M&E, silence where intended | +| Text | captions, subtitles, forced narrative, on-screen copy, title-safe placement | +| Continuity | wardrobe, props, screen direction, product orientation, light direction, last-frame handoffs | +| Rights | references, music, voice, likeness, product/brand authorization, stock licenses | +| Metadata | job ID, model/surface, prompt version, seed/settings if available, source URLs, approval owner | +| Versioning | hero, cutdown, vertical, square, textless, localized, archival | +| Human QC | watch all outputs at normal speed and pause on fragile frames | + +## Naming Pattern + +Use consistent names: + +`PROJECT_CAMPAIGN_VERSION_RATIO_LANG_DATE_STATUS` + +Example: + +`LUMA_BOTTLE_HERO_15S_9x16_TEXTLESS_2026-05-30_REVIEW01` + +## Client Review Packet + +Include: + +- concept/treatment; +- shot list and accepted prompts; +- reference role map and rights notes; +- exported review links/files; +- known issues and recommended fixes; +- approval questions; +- delivery spec assumptions. + +## QC Failure Routing + +| Failure | Route | +|---|---| +| face/product/text drift | I2V lock, edit pass, composite in post, or regenerate from stable frame | +| continuity mismatch | update continuity ledger and regenerate only affected shot | +| color mismatch | grade/conform first; only regenerate if lighting intent is wrong | +| caption/text issue | remove generated text and add typography in post | +| loudness/sync issue | fix mix or edit timing; do not rely on prompt repair alone | +| unsafe or rights issue | rewrite to original/authorized material and document rights | + +## Done Definition + +A professional Seedance asset is done only when the creative owner approves the shot, the rights map is clean, continuity is tracked, the post handoff is explicit, and the delivery target has passed human QC. diff --git a/references/examples-by-mode.md b/references/examples-by-mode.md index 29c08ba..23ead08 100644 --- a/references/examples-by-mode.md +++ b/references/examples-by-mode.md @@ -34,6 +34,26 @@ Use these as safe structures, not fixed scripts. They are original examples and `Use the returned last frame from the previous clip as the new first frame. Preserve character, wardrobe, light direction, and camera angle. Continue only one action: smoke thins around the doorway and the character lowers the lantern. No new scene, no new camera move, no extra VFX.` +## Professional Shot Contract + +**Brief:** 15s product film, 16:9 hero plus 9:16 cutdown, textless plate required. + +| Shot | Purpose | Seedance prompt spine | +|---|---|---| +| S01 | establish use context | `Wide 16:9 kitchen counter at dawn, original athlete places the bottle beside running shoes. Camera: locked medium-wide, slight push-in, warm window light, quiet room tone. End frame holds product left of center for vertical crop.` | +| S02 | material proof | `[Image1] is the product reference; preserve logo, label, cap, color, and geometry exactly. Locked macro shot, condensation beads merge and slide down glass, narrow warm strip light sweeps once. SFX: single glass tick at endpoint.` | +| S03 | packshot | `[Image1] product preserved exactly, black acrylic reflection, slow slider from silhouette to front three-quarter hero angle. No generated text; leave clean right-side negative space for localized copy in post.` | + +Continuity anchors: same bottle orientation, label color, dawn palette, clean counter surface, no burned-in text. Delivery notes: make textless 16:9 and 9:16 versions; add claims/captions in post. + +## Localization Handoff Example + +`Character A stands still in a locked medium close-up and says "We start now." Spoken language: English. Post: create subtitles in Spanish, Japanese, Korean, Russian, and Chinese; keep lower third clear; no burned-in captions; preserve textless plate and room tone for dubbing/M&E.` + +## Delivery QC Example + +`Before delivery: verify frame rate, aspect ratio, crop-safe product placement, product color, no logo drift, dialogue sync, loudness target per buyer spec, subtitle placement, textless export, rights notes for Image1/Audio1, and human watch-through at normal speed.` + ## Multilingual Community-Informed Examples These examples are original structures distilled from public multilingual practice. They do not copy community prompt text. diff --git a/references/json-schema.md b/references/json-schema.md index 789c9a7..ab022df 100644 --- a/references/json-schema.md +++ b/references/json-schema.md @@ -13,15 +13,72 @@ Use this schema when the user wants structured output or when an automation pipe {"tag": "Audio1", "role": "voice | rhythm | ambience | music | tempo | reference_audio"} ], "characters": [], + "production": { + "phase": "brief | preproduction | generation | review | post | localization | delivery", + "role": "director | dp | producer | editor | colorist | sound | localization | qc", + "delivery_surface": "web | broadcast | social | theatrical | client_review | archive", + "approval_owner": "" + }, + "shot_list": [ + { + "shot_id": "S01_SH01", + "purpose": "establish | reveal | demonstrate | emotional_turn | end_card", + "shot_contract": "shot size, angle, lens feel, camera move, endpoint", + "start_frame": "", + "end_frame": "", + "risks": [] + } + ], + "continuity_anchors": { + "character": [], + "product": [], + "wardrobe": [], + "props": [], + "location": "", + "screen_direction": "", + "eyeline": "", + "lighting_state": "", + "audio_state": "" + }, "scene": "", "camera": "", "motion": "", "lighting": "", "style": "", "audio": "", + "color_pipeline": { + "look_intent": "", + "working_assumption": "", + "output_transform": "SDR Rec.709 | HDR PQ | theatrical | social", + "show_lut_or_cdl_notes": "", + "qc_notes": [] + }, + "subtitle_plan": { + "subtitles": false, + "sdh": false, + "forced_narrative": false, + "dubbing": false, + "textless_required": false, + "languages": [] + }, + "audio_deliverables": { + "full_mix": true, + "stems": [], + "m_and_e": false, + "loudness_target": "", + "sync_cues": [] + }, + "delivery": { + "frame_rate": "", + "resolution": "", + "aspect_ratio": "", + "safe_area": "", + "version_name": "", + "qc_checks": [] + }, "safety_notes": [], "final_prompt": "" } ``` -The JSON wrapper is for planning. The final prompt still needs to read naturally. +The JSON wrapper is for planning. The final prompt still needs to read naturally. For professional work, keep the production, shot-list, continuity, localization, audio, color, and delivery fields as handoff metadata; do not cram all of them into the prompt. diff --git a/references/multilingual-community-examples.md b/references/multilingual-community-examples.md index a2aaeef..6fa532f 100644 --- a/references/multilingual-community-examples.md +++ b/references/multilingual-community-examples.md @@ -113,3 +113,35 @@ When a multilingual prompt fails, repair in this order: 4. Keep dialogue in the speaker language, but keep camera and constraints in the clearest language for the operator. 5. Reduce the prompt to one visible beat and one camera move. 6. If it still blocks, change the creative surface, not just the language. + +## Global Production Handoff Patterns + +Use these when the collaborator is not only prompting, but preparing a shot for a director, editor, localization team, or client. + +| Language context | Production-safe structure | +|---|---| +| Chinese-English shot list | `Shot ID + @图/@视频 role map + action endpoint + Camera in English + 后期备注: textless/localized copy in post` | +| Japanese review notes | `ショット目的 + 保持する要素 + 修正する動き + postで追加する字幕/コピー` | +| Korean dialogue handoff | `대사 + locked framing + speaker tag + 자막/더빙은 후반 작업에서 처리` | +| Spanish client versioning | `versión 15s/9:16 + producto protegido + texto en post + subtítulos separados` | +| Russian delivery note | `роль референса + что не менять + textless plate + отдельные субтитры/озвучка` | + +### Chinese-English client shot + +`S01_SH02: @图1锁定产品logo、标签、瓶身比例；@视频1仅参考 slow slider rhythm，不复制环境。产品在黑色亚克力台面上，暖色条形光扫过瓶身，最后停在正三分之二角度。Camera: locked macro-to-medium push-in. Post: no generated text; add Chinese/English campaign copy in edit.` + +### Japanese localization handoff + +`[Image1]の人物と衣装を保持。Character A says "I am ready" in English, locked medium close-up, no head turn. Post note: 日本語字幕と吹替は後処理で作成、画面下部を空ける、焼き込み文字なし。` + +### Korean social cutdown + +`9:16 모바일 컷다운, original product centered, no edge-critical action. Camera: slow push-in only. Sound: one clean product click. Post: 한국어 자막과 법적 문구는 편집에서 추가, textless plate required.` + +### Spanish delivery note + +`Versión 15s horizontal y 6s vertical. [Image1] conserva producto, etiqueta y color. No texto generado dentro de la imagen; entregar placa limpia para copy localizado. Subtítulos, claims y CTA se agregan en postproducción.` + +### Russian QC handoff + +`[Image1] сохраняет оригинального персонажа и костюм; [Audio1] задает только темп. Камера фиксированная, средний план, один короткий жест. Post/QC: отдельные русские субтитры, textless version, проверить синхрон губ и отсутствие изменения лица.` diff --git a/references/pro-filmmaking-standards.md b/references/pro-filmmaking-standards.md new file mode 100644 index 0000000..4ffd656 --- /dev/null +++ b/references/pro-filmmaking-standards.md @@ -0,0 +1,52 @@ +# Professional Filmmaking Standards + +last_verified: 2026-05-30 + +Use this reference when Seedance work must support directors, cinematographers, editors, producers, localization teams, commercial agencies, or delivery/QC teams. Treat these standards as workflow guidance, not as a promise that a video-generation surface can render or export every technical deliverable directly. + +## Source Boundaries + +| Area | Preferred source type | Use in this skill | Boundary | +|---|---|---|---| +| Shot design | ASC, cinematography education, production shot-list practice | Translate creative intent into shot contracts. | Do not over-specify frame-perfect camera physics for short AI clips. | +| Continuity | Script supervision and production continuity practice | Track wardrobe, props, screen direction, eyelines, lighting state, and final-frame handoff. | AI continuity is probabilistic; use references and smaller shots for fragile anchors. | +| Color | ACES docs, AMF, delivery partner guidance | Track color intent, show look, LUT/CDL/LMT notes, HDR/SDR expectations. | A prompt can request a look; finishing must still be verified in color tools. | +| Aspect ratio | Delivery partner specs, DCI/ISDCF, platform specs | Separate creative framing from delivery containers and safe areas. | Do not assume every surface supports every ratio or crop. | +| Audio | ITU BS.1770, EBU R128, ATSC A/85, delivery partner mix specs | Plan dialogue, ambience, SFX, music, M&E/stems, loudness, and sync checks. | Seedance prompt audio is not a certified final mix. | +| Subtitles | Netflix timed text, WebVTT, accessibility and caption rules | Plan subtitles, SDH, forced narratives, reading time, placement, and localization notes. | Do not rely on generated burned-in text as final deliverable subtitles. | +| Delivery/QC | SMPTE IMF, DPP, Netflix delivery specs, DCI | Create preflight checks for frame rate, resolution, color, audio, captions, textless, metadata, and human QC. | Always follow the buyer/platform spec actually contracted for the job. | + +## Professional Operating Spine + +1. **Brief:** define client/creative goal, audience, territory, duration, aspect ratio, deliverables, references, rights, approval owner, and hard constraints. +2. **Pre-production:** create treatment, reference map, shot list, continuity ledger, color/audio/localization intent, and risk log. +3. **Generation plan:** split into stable shots. Assign one visible beat, one camera idea, and one endpoint per Seedance clip. +4. **Review loop:** evaluate identity, product, action, camera, continuity, audio sync, text, safety, and rights before extending or editing. +5. **Post plan:** edit, conform, stabilize, sound, color, captions, versioning, textless, and archival metadata. +6. **Delivery/QC:** check spec, naming, frame rate, resolution, color pipeline, loudness, captions, safe areas, rights notes, and human review. + +## When To Load Detailed References + +| User need | Load | +|---|---| +| Treatment, production plan, client brief, campaign | `shot-list-continuity.md`, `delivery-qc.md` | +| Camera, lens, shot size, blocking | `cinematography-shot-language.md` | +| Multi-shot continuity, scene handoff | `shot-list-continuity.md` | +| ACES, HDR/SDR, LUT/CDL/look language | `color-pipeline-aces.md` | +| 16:9, 9:16, 1.85, 2.39, social cutdowns | `aspect-ratio-delivery.md` | +| Dialogue, mix, stems, M&E, loudness | `audio-post-delivery.md` | +| Subtitles, dubbing, captions, forced narrative | `subtitles-localization.md` | +| IMF/DCP/social export, QC checklist | `delivery-qc.md` | + +## Professional Answer Contract + +For professional filmmaker requests, return: + +- production phase and role: director, DP, editor, producer, sound, localization, or delivery; +- assumptions and source-date caveats for volatile platform claims; +- shot contract or workflow checklist; +- reference and rights map; +- continuity anchors; +- post/delivery notes when the output will leave the prompt stage. + +Avoid giving only a final prompt when the user is asking for a film, commercial, campaign, localization package, or delivery-ready workflow. diff --git a/references/research-2026-05-30.md b/references/research-2026-05-30.md index a647216..da95984 100644 --- a/references/research-2026-05-30.md +++ b/references/research-2026-05-30.md @@ -2,7 +2,7 @@ last_verified: 2026-05-30 -This snapshot records the evidence used for the v5.4.3 section-depth, multilingual, visual-asset, and surface-refresh pass. It is a dated working layer, not a permanent guarantee of platform access, pricing, limits, or regional availability. +This snapshot records the evidence used for the v5.4.4 section-depth, multilingual, visual-asset, professional-filmmaker, and surface-refresh pass. It is a dated working layer, not a permanent guarantee of platform access, pricing, limits, or regional availability. ## Source Priority @@ -10,8 +10,9 @@ This snapshot records the evidence used for the v5.4.3 section-depth, multilingu 2. Official Volcengine or BytePlus ModelArk documentation. 3. Official or partner workflow documentation such as ComfyUI partner nodes. 4. Academic/model-card material published by the provider. -5. Agent-skill packaging docs for repository structure. -6. Community prompt guides, forums, prompt corpora, and wrapper documentation. +5. Professional film, post-production, color, subtitle, audio, and delivery standards. +6. Agent-skill packaging docs for repository structure. +7. Community prompt guides, forums, prompt corpora, and wrapper documentation. ## Confirmed Official Signals @@ -29,6 +30,7 @@ This snapshot records the evidence used for the v5.4.3 section-depth, multilingu - ByteDance's own launch material lists remaining weaknesses: detail stability, hyper-realism, dynamic vitality, multi-subject consistency, text rendering, complex editing, and occasional audio distortion. - OpenAI's current Codex Agent Skills docs describe a skill as a directory with a required `SKILL.md` plus optional `scripts/`, `references/`, `assets/`, and `agents/` folders. This repo follows that shape. - Recent audio-video benchmark work such as AVBench and VABench strengthens eval vocabulary around cross-modal consistency, synchronization, lip-speech consistency, and audio-video QA, but it should not be treated as a Seedance access or performance guarantee. +- Professional workflow references reinforce that AI-video prompting should separate creative shot intent from delivery obligations: shot contracts, continuity ledgers, ACES/color handoff, subtitle/localization planning, audio stems/M&E/loudness, aspect-ratio variants, and QC are workflow metadata unless an active platform explicitly implements them. ## Field-Observed Signals @@ -45,6 +47,7 @@ Use these as practitioner guidance, not platform guarantees: - Editing or extending a failed segment can be cheaper and more controllable than regenerating the entire clip. - Product logos, faces, readable text, hands, fast gestures, and multi-character blocking benefit from locked framing and stronger constraints. - Agent-skill repos stay more usable when README explains the workflow at human level while scripts enforce file, source, and metadata integrity. +- Professional users need role-specific artifacts: directors need shot purpose and performance beats, DPs need shot contracts, producers need rights/approval gates, editors need continuity handoffs, colorists and sound teams need post metadata, localization teams need textless/caption/dubbing plans, and QC teams need delivery preflight. ## Excluded From Active Guidance diff --git a/references/shot-list-continuity.md b/references/shot-list-continuity.md new file mode 100644 index 0000000..f290352 --- /dev/null +++ b/references/shot-list-continuity.md @@ -0,0 +1,75 @@ +# Shot List And Continuity + +Use this reference when a user asks for a film scene, ad sequence, multi-shot plan, storyboard, treatment-to-shot-list conversion, or continuity repair. + +## Pre-Production Brief + +Collect only what changes production decisions: + +- objective: film scene, product ad, music video, trailer, social cutdown, internal pitch, test render; +- audience and territory; +- duration and target platforms; +- hero subject, product, character, or brand; +- references and what each one controls; +- rights/authorization status; +- dialogue/audio needs; +- aspect ratios and delivery surfaces; +- deadline, approval owner, and risk constraints. + +## Shot List Fields + +| Field | Example | +|---|---| +| Shot ID | `S01_SH03` | +| Duration | `5s` | +| Mode | T2V, I2V, R2V, FLF2V, edit, extend | +| Purpose | establish, reveal, demonstrate, emotional turn, end card | +| Subject/action | product rotates once; character lowers letter | +| Shot contract | medium close-up, locked-off, 50mm portrait compression | +| References | `[Image1] product identity`, `[Video1] camera rhythm`, `[Audio1] beat` | +| Continuity anchors | wardrobe, prop, screen direction, light state, background layout | +| Start frame | where the shot begins | +| End frame | next-shot handoff or final target | +| Audio | dialogue, ambience, SFX, music cue, silence | +| Risks | face drift, logo text, hand complexity, unsafe likeness, prompt block | +| Review notes | accept, retry, edit, extend, replace | + +## Continuity Ledger + +Track these anchors across shots: + +| Anchor | What to record | +|---|---| +| Character | tag, wardrobe, hair, silhouette, prop, emotional state | +| Product | label, logo, geometry, material, color, packshot angle | +| Location | layout, left/right geography, doorway/window positions | +| Screen direction | subject moves left-to-right or right-to-left | +| Eyeline | where each character looks and whether it matches reverse shots | +| Lighting | key direction, practical sources, time of day, atmosphere | +| Camera | lens feel, height, stability, movement family | +| Action state | what happened at previous endpoint | +| Sound state | music, ambience, dialogue, sync cue | + +## Treatment To Shot List + +1. Extract the dramatic or commercial beats. +2. Assign one Seedance clip per beat unless the beat is tiny and stable. +3. Choose the fragile anchor for each clip: face, product, text, choreography, or endpoint. +4. Select the safest mode: I2V for identity/product preservation, FLF2V for exact endpoints, R2V for role-bound motion or camera, edit for one-layer changes, extend for continuation. +5. Create a shot contract and continuity anchors for every shot. +6. Write prompts only after the shot list is stable. + +## Three-Shot Commercial Pattern + +| Shot | Purpose | Prompt spine | +|---|---|---| +| 1 | Problem or world | wide or medium environment, simple action, establish product context | +| 2 | Product/material proof | close or macro, locked or slow push, material detail and one SFX | +| 3 | Hero packshot/end state | product three-quarter, clean light, logo preserved, tagline handled in post | + +## Handoff Rules + +- Use the accepted last frame as the next first-frame reference when continuity matters. +- Keep wardrobe, product orientation, screen direction, and light direction stable unless the story explicitly changes them. +- If a prompt creates a good performance but bad product/logo, edit or composite the product layer rather than regenerating the whole shot. +- If a shot fails three times, simplify the shot contract or split the action into two clips. diff --git a/references/source-registry.md b/references/source-registry.md index ff4de25..88463e8 100644 --- a/references/source-registry.md +++ b/references/source-registry.md @@ -33,6 +33,13 @@ Use this registry before making factual claims about Seedance 2.0 platform behav | Agent Skills structure | OpenAI Codex Agent Skills docs: https://developers.openai.com/codex/skills, OpenAI Academy plugins/skills explainer: https://openai.com/academy/codex-plugins-and-skills/, OpenAI Codex Plugins docs, and Agent Skills open standard: https://agentskills.io/ | confirmed | Recheck before changing install guidance or root skill layout. | Packaging guidance, not Seedance platform capability. | | Runway MCP agent surface | Runway MCP announcement: https://runwayml.com/news/mcp | confirmed | Use for agent-surface availability only. | Does not change Seedance model capability; plan and connector access are Runway-specific. | | Audio-video eval vocabulary | AVBench and VABench papers: https://arxiv.org/abs/2605.24652 and https://openaccess.thecvf.com/content/CVPR2026/papers/Hua_VABench_A_Comprehensive_Benchmark_for_Audio-Video_Generation_CVPR_2026_paper.pdf | field-observed | Use for eval dimensions such as audio-video sync and cross-modal consistency. | Benchmark framing, not product access or official Seedance performance proof. | +| Professional shot language | ASC camera-movement education and production shot-list practice: https://theasc.com/article/shot-craft-camera-movement/ and https://www.studiobinder.com/blog/shot-list-template-free-download/ | field-observed | Use for shot contracts, camera movement, and shot-list fields. | Industry workflow guidance, not Seedance capability proof. | +| Continuity practice | ScreenSkills script-supervisor role: https://www.screenskills.com/job-profiles/browse/film-and-tv-drama/technical/script-supervisor-film-and-tv-drama/ | field-observed | Use for continuity anchors such as wardrobe, props, eyeline, screen direction, and notes. | Role guidance, not platform behavior. | +| Color management and ACES | ACES docs and AMF specification: https://docs.acescentral.com/background/overview/ and https://docs.acescentral.com/amf/specification/ | confirmed | Use for color pipeline vocabulary and handoff metadata. | A prompt can describe a look; it cannot certify ACES compliance. | +| Aspect ratio and delivery containers | DCI, ISDCF, and buyer/platform specs: https://www.dcimovies.com/dci-specification/ and https://registry-page.isdcf.com/ | field-observed | Use for aspect-ratio/container separation and naming caution. | Always follow the contracted delivery spec. | +| Subtitles and captions | Netflix timed text, WebVTT, and accessibility rules: https://partnerhelp.netflixstudios.com/hc/en-us/articles/215758617-Timed-Text-Style-Guide-General-Requirements, https://w3c.github.io/webvtt/, and https://www.law.cornell.edu/cfr/text/47/79.1 | volatile | Use for subtitle/SDH/forced-narrative planning and caption-safe framing. Recheck buyer, language, region, and platform requirements. | Delivery requirements vary by buyer, language, region, and platform. | +| Audio loudness and post | ITU BS.1770, EBU R128, ATSC A/85, and buyer sound specs: https://www.itu.int/rec/R-REC-BS.1770, https://tech.ebu.ch/fr/publications/r128, https://www.atsc.org/atsc-documents/a85-techniques-for-establishing-and-maintaining-audio-loudness-for-digital-television/, and https://partnerhelp.netflixstudios.com/hc/en-us/articles/360001794307-Netflix-Sound-Mix-Specifications-Best-Practices-v1-6 | volatile | Use for stems, M&E, sync, loudness, and mix handoff language. Recheck the target buyer or platform spec. | Prompt audio is not a certified final mix. | +| Delivery and QC | SMPTE IMF, DPP specs, Netflix delivery specs, and MovieLabs OMC: https://www.smpte.org/standards/st2067, https://www.thedpp.com/specs/, https://partnerhelp.netflixstudios.com/hc/en-us/sections/10066414335891-Delivery-Specifications, and https://movielabs.com/ontology-for-media-creation/ | volatile | Use for delivery preflight and metadata/versioning guidance. Recheck contracted platform specs before final delivery. | Contracted platform specs override generic guidance. | | Community practice | Douyin, Bilibili, CSDN, Reddit, Habr, creator notes, workflow screenshots | field-observed | Use only as practitioner guidance. | Mark as non-official; do not state as model guarantee. | | Localized and mixed-language prompting | Japanese, Korean, Spanish, Russian, and multilingual community guides plus forum observations | field-observed | Recheck public pages before release; use for vocabulary and example structure only. | Code-mixing can clarify safe prompts, but never treat it as official filter behavior or safety evasion. | | Community prompt corpora | YouMind/OpenLab, public prompt galleries, forum collections | field-observed | Mine structure, timing, vocabulary, and failure patterns only after safety classification. | Do not copy unsafe, IP-sensitive, or real-person prompts into active examples. | diff --git a/references/subtitles-localization.md b/references/subtitles-localization.md new file mode 100644 index 0000000..b9f7040 --- /dev/null +++ b/references/subtitles-localization.md @@ -0,0 +1,54 @@ +# Subtitles And Localization + +Use this reference when a Seedance project needs global release, subtitles, captions, forced narratives, dubbing, multilingual prompts, or market-specific copy. + +## Localization Plan + +| Deliverable | Purpose | Prompt/post boundary | +|---|---|---| +| Subtitles | translate spoken dialogue | create in post from approved script, not generated as moving text | +| SDH captions | dialogue plus important sound cues | plan sound cues and speaker IDs; author captions in post | +| Forced narrative | translate signs, texts, or off-language lines | avoid burned-in AI text; keep clean plates | +| Dubbing | localized voice performance | use short speaker turns, stable framing, and post-sync review | +| M&E | music and effects without dialogue | plan audio layers separately | +| Textless | picture without titles/lower thirds | generate clean background/action plates | +| Market copy | local tagline or legal claim | add in design/edit tools after legal approval | + +## Prompting For Subtitle-Friendly Footage + +- Keep dialogue short and assigned to a speaker. +- Use stable medium or medium close-up for important spoken lines. +- Leave negative space for captions when needed. +- Avoid generated subtitles or small moving text unless the task is explicitly a rough concept preview. +- Preserve clean plates for markets where copy changes. +- For multilingual dialogue, specify which language is spoken and which language is captioned in post. + +## Reading And Placement Checks + +For professional output, check: + +- captions do not cover faces, product claims, logos, legal disclaimers, or key action; +- subtitles have enough reading time for the target language; +- speaker changes are clear; +- SDH sound cues describe story-relevant sounds only; +- forced narratives are used only when needed; +- line breaks preserve meaning; +- formal/informal address matches region and character relationship. + +## Global Prompt Handoff + +| Language need | Safe wording | +|---|---| +| Chinese prompt with English camera terms | Chinese for role binding, English for camera/lens terms if clearer; never for evasion | +| Japanese market copy | keep generated shot textless; add Japanese copy in post | +| Korean dialogue | short quoted line, stable face framing, no head turn | +| Spanish captions | plan caption-safe lower third and avoid burned-in source text | +| Russian localization | deliver textless plate plus separate Russian subtitle/copy file | + +## Cultural Localization + +Ask what must localize: dialogue, product claim, holiday/season, gesture, sign, food, wardrobe, legal text, or music cue. Do not assume a literal translation is a market-ready localization. + +## Safe False-Positive Repair + +Mixed-language wording may clarify benign production context and reference roles, but it must not hide unsafe content. If a blocked prompt includes violence, real-person likeness, protected IP, sexual content, or evasion-like phrasing, repair the underlying issue in every language. diff --git a/scripts/design_audit.py b/scripts/design_audit.py index e0f15a8..75c58d6 100644 --- a/scripts/design_audit.py +++ b/scripts/design_audit.py @@ -31,10 +31,13 @@ def main() -> int: "assets/skill-os-infographic.png", "assets/skill-map-cinematic.png", "## What This Skill Does", + "## Professional Filmmaker Scope", "## Operating System At A Glance", "## Start Here", "## Skill Map", "api-workflow.md", + "pro-filmmaking-standards.md", + "delivery-qc.md", "examples-by-mode.md", "multilingual-community-examples.md", "## Validation", diff --git a/scripts/eval_schema_check.py b/scripts/eval_schema_check.py index 4847f88..8f34c1c 100644 --- a/scripts/eval_schema_check.py +++ b/scripts/eval_schema_check.py @@ -42,6 +42,14 @@ "extension_quality_degradation", "multilingual_false_positive_repair", "cinematic_infographic_front_page", + "professional_shot_contract", + "multi_shot_continuity_handoff", + "aces_color_pipeline_no_hallucination", + "aspect_ratio_delivery_surface", + "subtitle_localization_accessibility", + "audio_post_loudness_stems", + "delivery_qc_preflight", + "imf_localization_versioning", } diff --git a/scripts/validate_skills.py b/scripts/validate_skills.py index 4fcb23d..df0b8e3 100644 --- a/scripts/validate_skills.py +++ b/scripts/validate_skills.py @@ -22,6 +22,14 @@ "references/research-2026-05-30.md", "references/agent-compatibility.md", "references/api-workflow.md", + "references/pro-filmmaking-standards.md", + "references/cinematography-shot-language.md", + "references/shot-list-continuity.md", + "references/color-pipeline-aces.md", + "references/aspect-ratio-delivery.md", + "references/subtitles-localization.md", + "references/audio-post-delivery.md", + "references/delivery-qc.md", "references/examples-by-mode.md", "references/multilingual-community-examples.md", "references/platform-surface-matrix.md", @@ -152,8 +160,8 @@ def validate_skill(path: Path, root: Path, errors: list[str], warnings: list[str if metadata_value(frontmatter, "parent") != "seedance-20": errors.append(f"{rel}: missing metadata.parent: seedance-20") - if metadata_value(frontmatter, "version") != "5.4.3": - errors.append(f"{rel}: metadata.version must be 5.4.3") + if metadata_value(frontmatter, "version") != "5.4.4": + errors.append(f"{rel}: metadata.version must be 5.4.4") description = value_for(frontmatter, "description") or "" if not description.startswith("This skill should be used when"): @@ -237,7 +245,7 @@ def main() -> int: print(f"- {error}") return 1 - print(f"Validated root plus {len(EXPECTED_SKILLS)} sub-skills and required v5.4.3 files.") + print(f"Validated root plus {len(EXPECTED_SKILLS)} sub-skills and required v5.4.4 files.") return 0 diff --git a/skills/seedance-antislop/SKILL.md b/skills/seedance-antislop/SKILL.md index ae99071..6f3dd35 100644 --- a/skills/seedance-antislop/SKILL.md +++ b/skills/seedance-antislop/SKILL.md @@ -8,7 +8,7 @@ tags: - anti-slop - seedance-20 metadata: - version: "5.4.3" + version: "5.4.4" updated: "2026-05-30" parent: "seedance-20" author: "Iamemily2050 (@iamemily2050)" diff --git a/skills/seedance-audio/SKILL.md b/skills/seedance-audio/SKILL.md index a5b694d..fc76065 100644 --- a/skills/seedance-audio/SKILL.md +++ b/skills/seedance-audio/SKILL.md @@ -9,7 +9,7 @@ tags: - dialogue - seedance-20 metadata: - version: "5.4.3" + version: "5.4.4" updated: "2026-05-30" parent: "seedance-20" author: "Iamemily2050 (@iamemily2050)" @@ -23,7 +23,7 @@ metadata: Use this for dialogue, lip-sync, sound layers, music, ambience, beat-sync, audio-reference mapping, desync troubleshooting, or sound-driven visual timing. Audio should support the visible beat instead of becoming a second competing prompt. -Load `[ref:audio-guide]` for detailed constraints, beat-sync, desync repair, audio-reference conflicts, and multi-character workarounds. +Load `[ref:audio-guide]` for detailed constraints, beat-sync, desync repair, audio-reference conflicts, and multi-character workarounds. Load `[ref:audio-post-delivery]` when the user needs stems, M&E, dubbing, loudness, sync, mix, or delivery guidance. ## Core Rules @@ -53,4 +53,4 @@ If audio and video references fight each other, mute the reference video before ## Output Contract -Return speaker map, quoted dialogue, sound layers, audio reference role, lip-sync constraints, and a compact prompt-ready audio block. +Return speaker map, quoted dialogue, sound layers, audio reference role, lip-sync constraints, post/delivery notes if needed, and a compact prompt-ready audio block. diff --git a/skills/seedance-camera/SKILL.md b/skills/seedance-camera/SKILL.md index a1c766f..626f4ee 100644 --- a/skills/seedance-camera/SKILL.md +++ b/skills/seedance-camera/SKILL.md @@ -8,7 +8,7 @@ tags: - cinematography - seedance-20 metadata: - version: "5.4.3" + version: "5.4.4" updated: "2026-05-30" parent: "seedance-20" author: "Iamemily2050 (@iamemily2050)" @@ -22,7 +22,7 @@ metadata: Use one clear camera idea per short clip unless the user asks for a multi-shot sequence. The best camera direction has a start frame, movement, speed, subject relationship, and endpoint. Avoid stacking moves that fight each other, such as drone rise, dolly-in, handheld shake, and orbit in the same five-second shot. -Load `[ref:quick-ref]` for prompt assembly and `[ref:vocab/zh]` or `[ref:vocab/ru]` when camera wording must be multilingual. +Load `[ref:quick-ref]` for prompt assembly, `[ref:cinematography-shot-language]` for professional shot contracts, and `[ref:vocab/zh]` or `[ref:vocab/ru]` when camera wording must be multilingual. ## Camera Contract @@ -56,4 +56,4 @@ If the user gives several incompatible moves, choose one primary camera move and ## Output Contract -Return the selected camera phrase, why it fits the shot, conflicts removed, and a prompt-ready integrated sentence. +Return the selected camera phrase, why it fits the shot, conflicts removed, fragile anchors, endpoint, and a prompt-ready integrated sentence. diff --git a/skills/seedance-characters/SKILL.md b/skills/seedance-characters/SKILL.md index baef17c..d86ef17 100644 --- a/skills/seedance-characters/SKILL.md +++ b/skills/seedance-characters/SKILL.md @@ -9,7 +9,7 @@ tags: - consistency - seedance-20 metadata: - version: "5.4.3" + version: "5.4.4" updated: "2026-05-30" parent: "seedance-20" author: "Iamemily2050 (@iamemily2050)" @@ -23,6 +23,8 @@ metadata: Use this for identity, consistency, multi-character blocking, wardrobe continuity, hand safety, expression control, and likeness-sensitive character guidance. Character prompting must remove ambiguity before adding style. +Load `[ref:shot-list-continuity]` when character identity, wardrobe, props, eyeline, screen direction, or emotional state must survive across multiple shots. + ## Character Contract Assign each character a stable tag: `Character A`, `Character B`, `[Image1] subject`, or a user-provided original name. After more than one character appears, do not use ambiguous pronouns. Keep tag, role, appearance, wardrobe, position, action, and emotional beat consistent. diff --git a/skills/seedance-copyright/SKILL.md b/skills/seedance-copyright/SKILL.md index 5d93e91..6731800 100644 --- a/skills/seedance-copyright/SKILL.md +++ b/skills/seedance-copyright/SKILL.md @@ -10,7 +10,7 @@ tags: - safety - seedance-20 metadata: - version: "5.4.3" + version: "5.4.4" updated: "2026-05-30" parent: "seedance-20" author: "Iamemily2050 (@iamemily2050)" diff --git a/skills/seedance-examples-zh/SKILL.md b/skills/seedance-examples-zh/SKILL.md index ce1fb99..acd6abb 100644 --- a/skills/seedance-examples-zh/SKILL.md +++ b/skills/seedance-examples-zh/SKILL.md @@ -8,7 +8,7 @@ tags: - examples - seedance-20 metadata: - version: "5.4.3" + version: "5.4.4" updated: "2026-05-30" parent: "seedance-20" author: "Iamemily2050 (@iamemily2050)" diff --git a/skills/seedance-filter/SKILL.md b/skills/seedance-filter/SKILL.md index 4b3e8e1..9c44efb 100644 --- a/skills/seedance-filter/SKILL.md +++ b/skills/seedance-filter/SKILL.md @@ -8,7 +8,7 @@ tags: - safe-rewrite - seedance-20 metadata: - version: "5.4.3" + version: "5.4.4" updated: "2026-05-30" parent: "seedance-20" author: "Iamemily2050 (@iamemily2050)" diff --git a/skills/seedance-interview-short/SKILL.md b/skills/seedance-interview-short/SKILL.md index ba29c39..9048257 100644 --- a/skills/seedance-interview-short/SKILL.md +++ b/skills/seedance-interview-short/SKILL.md @@ -9,7 +9,7 @@ tags: - compression - seedance-20 metadata: - version: "5.4.3" + version: "5.4.4" updated: "2026-05-30" parent: "seedance-20" author: "Iamemily2050 (@iamemily2050)" diff --git a/skills/seedance-interview/SKILL.md b/skills/seedance-interview/SKILL.md index 1d2f652..df2ff69 100644 --- a/skills/seedance-interview/SKILL.md +++ b/skills/seedance-interview/SKILL.md @@ -9,7 +9,7 @@ tags: - brief - seedance-20 metadata: - version: "5.4.3" + version: "5.4.4" updated: "2026-05-30" parent: "seedance-20" author: "Iamemily2050 (@iamemily2050)" @@ -28,10 +28,11 @@ Use this as the full director interview when the user has a rough idea rather th 1. Build a safe draft premise immediately from the user input. 2. Ask only the missing questions that materially affect the prompt: subject, action, reference assets, camera feel, emotional direction, duration, and risk constraints. 3. Identify the genre path: product, lifestyle, drama, music video, landscape, commercial, animation, UGC, or experimental. -4. End with a concise creative brief and route to `[skill:seedance-prompt]` or `[skill:seedance-prompt-short]`. +4. If the user is a filmmaker, agency, producer, editor, localization team, or client-review owner, load `[ref:pro-filmmaking-standards]` and collect deliverables, territory, aspect ratio, approval owner, rights, and post/delivery needs. +5. End with a concise creative brief and route to `[skill:seedance-prompt]`, `[skill:seedance-prompt-short]`, or `[skill:seedance-pipeline]`. ## Output Contract -Return: concept summary, reference asset request, core scene, mood, camera intent, sound intent, safety notes, and next prompt path. +Return: concept summary, production phase, reference asset request, core scene, mood, camera intent, sound intent, safety/rights notes, deliverables if known, and next prompt path. Do not ask a long questionnaire when the user already supplied enough information to write the prompt. diff --git a/skills/seedance-lighting/SKILL.md b/skills/seedance-lighting/SKILL.md index c618ce1..6d84759 100644 --- a/skills/seedance-lighting/SKILL.md +++ b/skills/seedance-lighting/SKILL.md @@ -8,7 +8,7 @@ tags: - atmosphere - seedance-20 metadata: - version: "5.4.3" + version: "5.4.4" updated: "2026-05-30" parent: "seedance-20" author: "Iamemily2050 (@iamemily2050)" @@ -22,6 +22,8 @@ metadata: Lighting should describe physical sources and transitions, not abstract beauty. A useful lighting prompt tells the model where the light comes from, its color temperature, how shadows behave, what atmosphere catches the light, and whether the light changes during the clip. +Load `[ref:color-pipeline-aces]` when the user asks for ACES, HDR/SDR, show look, grade, LUT, CDL, product color, or professional color handoff. + ## Lighting Contract State: key source, direction, color temperature, atmosphere, shadow behavior, reflective behavior, and any transition. diff --git a/skills/seedance-motion/SKILL.md b/skills/seedance-motion/SKILL.md index 467600e..bc21b7e 100644 --- a/skills/seedance-motion/SKILL.md +++ b/skills/seedance-motion/SKILL.md @@ -9,7 +9,7 @@ tags: - physics - seedance-20 metadata: - version: "5.4.3" + version: "5.4.4" updated: "2026-05-30" parent: "seedance-20" author: "Iamemily2050 (@iamemily2050)" @@ -23,7 +23,7 @@ metadata: Use physical verbs and consequences. Motion should be observable on screen, timed within the clip, and assigned to a subject or object. Prefer one strong action with a visible endpoint over several vague actions competing for attention. -Load `[ref:reference-workflow]` for video-motion references and `[ref:examples-by-mode]` for safe edit, extend, and R2V patterns. +Load `[ref:reference-workflow]` for video-motion references, `[ref:shot-list-continuity]` for action handoffs across shots, and `[ref:examples-by-mode]` for safe edit, extend, and R2V patterns. ## Motion Contract diff --git a/skills/seedance-pipeline/SKILL.md b/skills/seedance-pipeline/SKILL.md index 28f0448..852e40f 100644 --- a/skills/seedance-pipeline/SKILL.md +++ b/skills/seedance-pipeline/SKILL.md @@ -9,7 +9,7 @@ tags: - integration - seedance-20 metadata: - version: "5.4.3" + version: "5.4.4" updated: "2026-05-30" parent: "seedance-20" author: "Iamemily2050 (@iamemily2050)" @@ -27,17 +27,19 @@ Use this for operational workflows, APIs, web surfaces, post-production, and int Always load `[ref:api-status]` for current API and platform claims. Load `[ref:model-name-map]` when a user says Pro, Fast, V2, or a wrapper model ID. Do not rely on old release-status memory. Load `[ref:api-workflow]` for implementation planning, task lifecycle, Runway/Volcengine field differences, pricing caveats, upload handling, and production readiness. +Load `[ref:pro-filmmaking-standards]` for professional film, commercial, agency, localization, post, and delivery workflows. Load `[ref:delivery-qc]` before saying an asset is delivery-ready. ## Workflow Split 1. Web workflow: Dreamina/Jimeng surface, references, prompt, output review. 2. API workflow: Volcengine, BytePlus, or Runway docs, model ID, auth, file handling, task creation, polling/querying, cancellation/deletion, task ledger, and retrieval. -3. Post workflow: stitching, audio cleanup, captions, color, upscale, delivery. -4. First/last-frame workflow: map first frame, last frame, transition action, identity locks, and ending target. -5. Runway workflow: model `seedance2`, `runway://` uploads, audio-reference combination rules, plan/region caveats, and SDK type lag are Runway-specific. -6. Community workflow: ComfyUI or unofficial nodes must be labeled community/unverified unless sourced. -7. Corpus-mining workflow: classify sources before reuse; extract structure and vocabulary, not unsafe raw prompts. +3. Professional production workflow: treatment, shot list, continuity ledger, reference rights map, review loop, post handoff, and delivery/QC. +4. Post workflow: edit, conform, stitching, stabilization, audio cleanup, captions/subtitles, color, localization, versioning, textless, and delivery. +5. First/last-frame workflow: map first frame, last frame, transition action, identity locks, and ending target. +6. Runway workflow: model `seedance2`, `runway://` uploads, audio-reference combination rules, plan/region caveats, and SDK type lag are Runway-specific. +7. Community workflow: ComfyUI or unofficial nodes must be labeled community/unverified unless sourced. +8. Corpus-mining workflow: classify sources before reuse; extract structure and vocabulary, not unsafe raw prompts. ## Output Contract -Return the workflow path, source status, required inputs, validation steps, and risks. +Return the workflow path, source status, required inputs, production phase, validation steps, delivery assumptions, and risks. For professional jobs, include the next artifact to create: brief, shot list, continuity ledger, prompt batch, review packet, localization matrix, or QC preflight. diff --git a/skills/seedance-prompt-short/SKILL.md b/skills/seedance-prompt-short/SKILL.md index cc28c32..019915d 100644 --- a/skills/seedance-prompt-short/SKILL.md +++ b/skills/seedance-prompt-short/SKILL.md @@ -8,7 +8,7 @@ tags: - chinese-prompt - seedance-20 metadata: - version: "5.4.3" + version: "5.4.4" updated: "2026-05-30" parent: "seedance-20" author: "Iamemily2050 (@iamemily2050)" diff --git a/skills/seedance-prompt/SKILL.md b/skills/seedance-prompt/SKILL.md index d1865e4..e77c3e7 100644 --- a/skills/seedance-prompt/SKILL.md +++ b/skills/seedance-prompt/SKILL.md @@ -8,7 +8,7 @@ tags: - video-generation - seedance-20 metadata: - version: "5.4.3" + version: "5.4.4" updated: "2026-05-30" parent: "seedance-20" author: "Iamemily2050 (@iamemily2050)" @@ -22,7 +22,7 @@ metadata: Build production-ready Seedance prompts from clear concepts or supplied reference assets. Treat the prompt as a short shooting brief: it must say what changes on screen, what the camera does, what the light and sound contribute, and what must stay stable. Keep final prompts under the platform prompt budget and remove filler before delivery. -Load `[ref:quick-ref]` for the checklist, `[ref:reference-workflow]` for multimodal references, `[ref:i2v-guide]` for image-to-video, `[ref:first-last-frame-guide]` for first/last-frame work, `[ref:examples-by-mode]` when examples are useful, and `[ref:multilingual-community-examples]` for Chinese/Russian/Japanese/Korean/Spanish or mixed-language prompts. +Load `[ref:quick-ref]` for the checklist, `[ref:reference-workflow]` for multimodal references, `[ref:i2v-guide]` for image-to-video, `[ref:first-last-frame-guide]` for first/last-frame work, `[ref:examples-by-mode]` when examples are useful, `[ref:shot-list-continuity]` for multi-shot professional plans, and `[ref:multilingual-community-examples]` for Chinese/Russian/Japanese/Korean/Spanish or mixed-language prompts. ## Director Formula @@ -68,6 +68,7 @@ Return: 2. Reference role map, if any. 3. Final prompt under 2000 characters. 4. Optional Chinese compressed version when useful. -5. Safety or copyright note when relevant. +5. Shot-list or delivery note when the prompt belongs to a professional sequence. +6. Safety or copyright note when relevant. Before finalizing, run an anti-slop pass and remove vague quality boosters. diff --git a/skills/seedance-recipes/SKILL.md b/skills/seedance-recipes/SKILL.md index b864eaa..81c0e26 100644 --- a/skills/seedance-recipes/SKILL.md +++ b/skills/seedance-recipes/SKILL.md @@ -9,7 +9,7 @@ tags: - recipes - seedance-20 metadata: - version: "5.4.3" + version: "5.4.4" updated: "2026-05-30" parent: "seedance-20" author: "Iamemily2050 (@iamemily2050)" @@ -23,7 +23,7 @@ metadata: Use recipes as starting patterns, not rigid prompt templates. Pick the recipe that matches the user's outcome, then customize subject, action, camera, lighting, audio, and constraints. Recipes should preserve the one-beat discipline of a short clip. -Load `[ref:genre-guides]` for genre patterns, `[ref:examples-by-mode]` when the user needs copy-ready examples, and `[ref:multilingual-community-examples]` when the recipe should reflect Chinese/Russian/Japanese/Korean/Spanish community-style structures. +Load `[ref:genre-guides]` for genre patterns, `[ref:examples-by-mode]` when the user needs copy-ready examples, `[ref:shot-list-continuity]` for professional multi-shot sequences or commercials, and `[ref:multilingual-community-examples]` when the recipe should reflect Chinese/Russian/Japanese/Korean/Spanish community-style structures. ## Recipe Families @@ -38,6 +38,7 @@ Load `[ref:genre-guides]` for genre patterns, `[ref:examples-by-mode]` when the | Animation | Original characters and stylized motion. | `medium + shape language + palette + elastic or weighted motion` | | VFX | Transformations, particles, weather, energy. | `source + material behavior + interaction + dissipation endpoint` | | First/last frame | In-between transitions, product state changes, character pose targets. | `first frame + last frame + continuous transition + identity locks` | +| Commercial campaign | 6/10/15/30s variants, vertical/social cutdowns, textless/localized masters. | `hook + product proof + end state + cutdown matrix + delivery notes` | ## Prompt Skeletons @@ -57,4 +58,4 @@ If a user gives many goals, choose the recipe that protects the most fragile req ## Output Contract -Return one selected recipe, why it fits, the customized prompt skeleton, and a compact final prompt. +Return one selected recipe, why it fits, the customized prompt skeleton, compact final prompt, and campaign/delivery notes when relevant. diff --git a/skills/seedance-style/SKILL.md b/skills/seedance-style/SKILL.md index 4495afd..476a0a6 100644 --- a/skills/seedance-style/SKILL.md +++ b/skills/seedance-style/SKILL.md @@ -9,7 +9,7 @@ tags: - ip-safe - seedance-20 metadata: - version: "5.4.3" + version: "5.4.4" updated: "2026-05-30" parent: "seedance-20" author: "Iamemily2050 (@iamemily2050)" diff --git a/skills/seedance-troubleshoot/SKILL.md b/skills/seedance-troubleshoot/SKILL.md index 246f938..a30d8d4 100644 --- a/skills/seedance-troubleshoot/SKILL.md +++ b/skills/seedance-troubleshoot/SKILL.md @@ -8,7 +8,7 @@ tags: - troubleshooting - seedance-20 metadata: - version: "5.4.3" + version: "5.4.4" updated: "2026-05-30" parent: "seedance-20" author: "Iamemily2050 (@iamemily2050)" @@ -22,7 +22,7 @@ metadata: Diagnose failure before rewriting. Do not simply add more adjectives. Identify whether the failure came from mode mismatch, overload, ambiguity, fragile identity, unsafe wording, unsupported platform behavior, or missing preservation constraints. -Load `[ref:field-observed-tips]`, `[ref:reference-workflow]`, and `[ref:api-workflow]` when the failure involves continuation, edit/extend, source clips, audio references, or platform-specific errors. +Load `[ref:field-observed-tips]`, `[ref:reference-workflow]`, and `[ref:api-workflow]` when the failure involves continuation, edit/extend, source clips, audio references, or platform-specific errors. Load `[ref:shot-list-continuity]` for multi-shot drift and `[ref:delivery-qc]` for final-client or delivery failures. ## Diagnostic Tree @@ -38,6 +38,7 @@ Load `[ref:field-observed-tips]`, `[ref:reference-workflow]`, and `[ref:api-work | Extension quality degrades | No last-frame anchor or too many new variables across continuations. | Use returned last frame as first frame and change one variable. | | Audio reference ignored | Competing video sound, no visual beat mapping, or unsupported combo. | Mute competing video and map one visible event to the beat. | | Text/logos break | Small text asked to move or be redrawn. | Keep text static, centered, and protected; animate light around it. | +| Client QC fails | Prompt output treated as final delivery without post/QC. | Route to delivery preflight, post fix, or regenerate only the failing shot. | ## Repair Process diff --git a/skills/seedance-vfx/SKILL.md b/skills/seedance-vfx/SKILL.md index d137a94..b1b1fd6 100644 --- a/skills/seedance-vfx/SKILL.md +++ b/skills/seedance-vfx/SKILL.md @@ -9,7 +9,7 @@ tags: - effects - seedance-20 metadata: - version: "5.4.3" + version: "5.4.4" updated: "2026-05-30" parent: "seedance-20" author: "Iamemily2050 (@iamemily2050)" diff --git a/skills/seedance-vocab-es/SKILL.md b/skills/seedance-vocab-es/SKILL.md index d28aacc..f5e04e3 100644 --- a/skills/seedance-vocab-es/SKILL.md +++ b/skills/seedance-vocab-es/SKILL.md @@ -8,7 +8,7 @@ tags: - vocabulary - seedance-20 metadata: - version: "5.4.3" + version: "5.4.4" updated: "2026-05-30" parent: "seedance-20" author: "Iamemily2050 (@iamemily2050)" diff --git a/skills/seedance-vocab-ja/SKILL.md b/skills/seedance-vocab-ja/SKILL.md index c2b8402..30ba9ed 100644 --- a/skills/seedance-vocab-ja/SKILL.md +++ b/skills/seedance-vocab-ja/SKILL.md @@ -8,7 +8,7 @@ tags: - vocabulary - seedance-20 metadata: - version: "5.4.3" + version: "5.4.4" updated: "2026-05-30" parent: "seedance-20" author: "Iamemily2050 (@iamemily2050)" diff --git a/skills/seedance-vocab-ko/SKILL.md b/skills/seedance-vocab-ko/SKILL.md index baf1143..24a3a0f 100644 --- a/skills/seedance-vocab-ko/SKILL.md +++ b/skills/seedance-vocab-ko/SKILL.md @@ -8,7 +8,7 @@ tags: - vocabulary - seedance-20 metadata: - version: "5.4.3" + version: "5.4.4" updated: "2026-05-30" parent: "seedance-20" author: "Iamemily2050 (@iamemily2050)" diff --git a/skills/seedance-vocab-ru/SKILL.md b/skills/seedance-vocab-ru/SKILL.md index 4899560..a50d0ed 100644 --- a/skills/seedance-vocab-ru/SKILL.md +++ b/skills/seedance-vocab-ru/SKILL.md @@ -8,7 +8,7 @@ tags: - vocabulary - seedance-20 metadata: - version: "5.4.3" + version: "5.4.4" updated: "2026-05-30" parent: "seedance-20" author: "Iamemily2050 (@iamemily2050)" diff --git a/skills/seedance-vocab-zh/SKILL.md b/skills/seedance-vocab-zh/SKILL.md index 918b925..a6a873c 100644 --- a/skills/seedance-vocab-zh/SKILL.md +++ b/skills/seedance-vocab-zh/SKILL.md @@ -8,7 +8,7 @@ tags: - vocabulary - seedance-20 metadata: - version: "5.4.3" + version: "5.4.4" updated: "2026-05-30" parent: "seedance-20" author: "Iamemily2050 (@iamemily2050)"