Home / All Skills / Content & Media Creation / virtual-couple-travel-vlog
English (EN)简体中文 (ZH)日本語 (JA)
Content & Media Creation · vibeshotclub/vsc-skills

virtual-couple-travel-vlog

Create virtual couple travel vlog workflows and assets from a user-provided theme, destination, couple profile, output folder, or reference images. Use when Codex needs to generate nostalgic iPhone travel photo grids, split 4x4 grids into four 2x2 memory sheets, create male/female character reference cards, generate Topview Omni Reference video clips, write manual Chinese music prompts, assemble final vlog videos, and save all assets in a user-selected dedicated project folder.

L4 High-Privilege · Graded auth & full auditV1 Commit-PinnedExecute CommandNetwork AccessWrite FileDual-pass Audit
Author vibeshotclubGitHub vibeshotclub/vsc-skills ↗Stars 832License 仓库未提供 LICENSE 文件（find 全仓无 LICENSE*；GitHub API license 字段为 null）Commit fcee81beb6
Control levels describe skill capabilities and suggested control measures; they do not represent quality or safety certifications, nor do they dictate usability. Usage decisions rest entirely with users and administrators. Different agent hosts apply varying constraints; do not assume hosts provide automatic protection. Third-party skill copyrights belong to their original authors. Use at your own risk.

1How It Works & Mechanism

It is not a single point tool but an end-to-end asset pipeline: starting from a single travel theme, it chains 12 tasks covering 4x4 photo wall -> image splitting -> character cards -> video prompts/generation -> music prompts -> final video assembly -> QA.

virtual-couple-travel-vlog/SKILL.md
Build a repeatable "virtual couple travel vlog" asset pipeline from a short theme such as "a young Japanese couple in Paris" or "a Chinese couple in northern Normandy."
The Task List section in SKILL.md enumerates ownership across all 12 tasks, and references/task_list.md expands them into Task -1 through Task 10 job tickets complete with acceptance criteria.

Key design for identity consistency: lock all 16 photos into a single model context via a single 4x4 grid generation, then slice it locally into four 2x2 sheets, rather than running four separate text-to-image generations.

virtual-couple-travel-vlog/SKILL.md
Prefer generating one 4x4 contact sheet in a single image task, then splitting it into four 2x2 sheets. Do not generate four independent 2x2 text-to-image sheets unless there is already a strong locked character reference image.
README's 'Identity Consistency Strategy from 4x4 to 2x2' repeats the same rationale: independent generations easily suffer from drifting facial features, hairstyles, outfits, and perceived age.

Image cropping is executed by a local Python/Pillow script, established as the sole default cropping path (no design tools, no fallback to independent 2x2 generations).

virtual-couple-travel-vlog/SKILL.md
Use [scripts/split-grid.py](scripts/split-grid.py) to split the approved 4x4 image into four 2x2 memory sheets. This is the only default crop route.
split-grid.py uses Pillow to crop four quadrants and names them memory_sheet_NN.png (appending a _new suffix if the file already exists, avoiding overwrites).

Character cards adhere to strict specifications (1:1 aspect ratio, seven-panel layout, four on top and three on bottom, solid white background, turnaround expressions/postures), generated exclusively via Codex Image Gen / gpt-image-2 by default.

virtual-couple-travel-vlog/SKILL.md
Generate character cards with Codex Image Gen / `gpt-image-2`, not Topview, unless the user explicitly asks for Topview for this step after being warned about possible lower model quality.
references/prompts.md provides complete male and female character card prompts, cleanup prompts, and fallback single-person identity reference prompts.

Video generation is platform-agnostic: it runs a 'provider detection' script first, automating execution if connected, or offering a complete manual handoff package if not.

virtual-couple-travel-vlog/SKILL.md
After the four 2x2 sheets, character cards, and video prompts are ready, run [scripts/check-video-providers.py](scripts/check-video-providers.py).
README lists candidate providers including Seedance API / Krea / Lovart / LibTV / Topview UI along with a 5-step manual platform workflow.

Topview automation is executed via a wrapper script to eliminate cross-platform shell quoting issues: assembling reference images into JSON before passing them to video_gen.py.

virtual-couple-travel-vlog/SKILL.md
Before submitting video tasks, pause for an explicit queue decision:
run-topview-omni.py formats the sheet, male card, and female card into a [{fileId, name: Image1/2/3}] JSON structure and invokes video_gen.py run --type omni via subprocess.

Strict gating before payment: display settings, credit estimates, and queue wait times prior to submission, requiring user selection between regular queue and Quick Generate (extra 15 credits), never silently spending credits.

virtual-couple-travel-vlog/SKILL.md
Do not continue to video submission until the user answers.
tooling.md provides an explicit decision prompt template and specifies routing to the official UI/board link if the local script lacks the quick-generate flag.

All assets reside in a dedicated project directory under the user-selected output root, with strict rules against deleting files or silently installing third-party skills/packages.

virtual-couple-travel-vlog/SKILL.md
Never delete files. If cleanup is needed, ask the user for explicit second confirmation first.
Same section: Do not install third-party skills or packages unless the user approves after a poison/backdoor audit and a system resource check.

Music and video editing follow 'principled fallbacks': music only provides Chinese prompts for manual generation; editing splits into simple/advanced tiers (FFmpeg vs Remotion/HyperFrames), explicitly stating HyperFrames does not generate source footage.

virtual-couple-travel-vlog/SKILL.md
Do not use Gemini CLI or any automatic music generation tool by default. Always write the final Chinese music prompt to `audio/music_prompt_zh.txt`
The HyperFrames specification requires 9:16 aspect ratio, ~60s duration across four 15s clips, running npx hyperframes lint / inspect / render.

The output structure is strictly defined as a contract (brief/prompts/images/videos/audio/qa), facilitating cross-session resumption and manual takeover.

virtual-couple-travel-vlog/SKILL.md
- `brief.md` - theme, couple profile, location, style notes.
The Output Contract section contains 6 clauses covering 4x4 source images, four 2x2 sheets, two character cards in images/, and four clips plus final_vlog.mp4 in videos/.

2Core Capabilities

0112-step task orchestration with sub-agent role mapping (intake/images/assets/character consistency/video/audio/editing/QA)
02Dependency self-check script distinguishing required tools (Python/Pillow/FFmpeg/FFprobe/Node/npm/npx/Remotion) from optional ones (Topview, HyperFrames)
03Local 4x4 to four 2x2 image cropping (Pillow, cropping by row/column blocks with sequential numbering)
04Video platform discovery: checks Topview installation path and environment variable availability for four platform API keys
05Topview Omni Reference wrapper invocation (Image1=2x2 memory sheet, Image2/3=male/female character cards, 9:16, 15s, 720p, sound off)
06FFmpeg stitching (attempting stream copy with -c copy first, falling back to transcoding), with optional user music mixing
07Persisting Chinese music prompts to disk (audio/music_prompt_zh.txt) for manual generation on external music platforms
08QA checklist: identity consistency, no grid artifacts, no text watermarks, all files contained within the project directory

3External Dependencies

KindDependency / Endpoint
cliFFmpeg / FFprobe
packagePython Pillow（PIL）
cliNode / npm / npx（Remotion 与 HyperFrames 路径）
cliTopview skill 的 video_gen.py（第三方技能，付费平台）
cliHyperFrames CLI（npx hyperframes lint / inspect / render）
api第三方视频/图像平台 API 密钥（仅检测存在性，不调用）：Seedance / Krea / Lovart / LibTV
cliCodex Image Gen（gpt-image-2）与 imagegen 技能——角色卡与记忆图的首选图像路径

4Recommended Governance & Evidence

Suggested Controls for Safe Operation
For Agent Runtime
Graded/per-use authorization, sandbox recommended
For Administrators
Approval workflow, rollback support, full audit
Evidence & Triggers (File Paths & Excerpts)
  • scripts/: Local script/command execution: python scripts/check-tools.py --show-install-hints (probe only, no installation)
  • 网络端点: External network access declaration: repository code contains no HTTP/socket calls; network activity occurs solely through invoked external CLIs
  • 本地工作目录: File I/O and artifact generation: creates <Project Title>/ and subdirectories prompts, images, videos, audio under the user-selected output root, writing brief.m
Audit Notes: Rationale: (1) Reads and evaluates third-party platform API keys in environment variables (presence check only, never echoing values) and depends on external paid platforms for core steps; (2) Relies on third-party skills and remote packages (Topview skill, Remotion plugin, npx hyperframes/create-video), representing typical supply chain exposure; (3) Contains execution attack surface: two of the five bundled Python scripts invoke external binaries via subprocess (video_gen.py, ffmpeg); (4) Creates project directories in user-designated paths. Mitigations are concrete: explicit gating before billing (Quick Generate's 15 extra credits require per-call confirmation, preventing silent spending), no file deletions, no silent installations (requires security audit + user approval), aborts after 3 failures with a manual handoff package, and non-committable items like upscaling/2K are explicitly marked as observed behavior rather than firm guarantees.

5Second-Pass Verification

  • [ok] 「仓库内无直接网络调用」是否成立 — 全仓 https?:// 扫描中本技能目录零命中；5 个脚本 import 仅 argparse/json/os/shutil/importlib/subprocess/sys/time/pathlib/PIL，无 urllib/requests/socket/http。
  • [ok] 密钥是否会被回显（凭证泄露风险） — first_env() 只返回命中的环境变量名，print_table() 打印的 evidence 列即该变量名；未出现 os.environ.get(name) 取值的输出路径，结论：不回显密钥值。
  • [ok] 外部依赖条目逐条反查 — FFmpeg/FFprobe/Node/npm/npx/PIL 在 check-tools.py 的 results 列表中逐项存在；Topview 路径解析在 run-topview-omni.py 与 check-video-providers.py 两处互证；npx hyperframes 三连命令在 SKILL.md 与 tooling.md 两处出现。
  • [unlocatable] Topview 平台域名（未能定位） — 本仓库内没有任何 Topview/Seedance/Krea/Lovart/LibTV 的域名或端点字符串，只有平台名与「本机是否安装/是否配置 key」的判断。真实网络端点位于被引用的第三方技能与平台侧，超出本次可读源范围，按纪律标为待补，不臆测域名。
  • [ok] 付费闸门是否真的落在流程里（而非口号） — SKILL.md 有独立的 Topview 段要求提交前展示设置/积分/等待并等用户回答；task_list.md Task 6 给出「Do not submit video tasks until the user replies」与逐次/整批授权规则；tooling.md 提供决策话术模板。「不静默花费」在指令层有三处互证。
  • [ok] 写入范围是否被限定在输出根目录内 — SKILL.md Operating Rules 第一条要求所有资产存于 <Output Root>/<Project Title>/；assemble-vlog.py 的工作目录默认取输出文件父目录；split-grid.py 的 output_dir 由参数传入。未发现写到技能目录或系统目录的路径（唯一例外是 npx 自身的缓存行为，属工具侧）。
  • [ok] HyperFrames 定位是否前后一致（防止第一遍读错） — SKILL.md『HyperFrames editing only』段、README『HyperFrames 不再作为四段源视频的生成 provider 使用』、tooling.md『Do not use HyperFrames as the source provider』三处口径一致，第一遍所述成立。
  • [ok] 任务清单完整性 — task_list.md 实含 Task -1、Task 0 至 Task 10 共 12 段，与 SKILL.md Task List 的 12 条编号一一对应；每条带 Acceptance 或 Fallback，非空壳。

6Verdict & Scenarios

  • Mitigates identity drift through structural mechanisms (single 4x4 -> local crop -> character cards) rather than relying on prompt repetition, backed by observed success paths and fallbacks (falling back to single-person identity reference images after 3 attempts).
  • Strict billing and authorization governance: upfront review of settings/credits/wait times, explicit confirmation for Quick Generate (15 credits), zero silent charges, zero silent package installations, and zero file deletions.
  • Platform-agnostic with an actionable manual handoff package (2x2 sheets + 2 character cards + 4 prompts), functional even without APIs.
  • The 5 bundled scripts are concise, single-purpose, and fully local (discovery, cropping, stitching, wrapping), with clear separation between required and optional dependencies.
  • Non-committable aspects of cost and capability are realistically documented (estimated vs observed, unavailable ETA, 2K feasibility), avoiding treating uncertainties as guarantees.
    • Best for: Ideal for content creators and lightweight production workflows producing virtual couple travel vlogs at scale: turning a single theme into photo walls, character cards, four 15-second clips, and a ~1-minute assembled vlog; suitable for users with Topview or other video APIs, or those willing to use manual platform handoff packages; particularly well-suited for projects requiring multi-session resumption and human handoff (standardized project directory contract + task_list tracking).
    Not for: Not suitable for users only seeking a single couple photograph (vsc-distinction explicitly states single couple photos do not trigger this workflow); not suitable for scenarios where users only want ideas/prompts without selecting an output directory upfront (routing rules require pure idea/prompt handoffs); not suitable for users unwilling to incur third-party platform fees or queue times; not suitable for running fully automated paths in environments lacking Python/Pillow/FFmpeg or where installing third-party packages is prohibited (must fall back to manual handoff package).
    Install Agent-Ready
    ① Verified Mirror 2026-09-15
    Option A · Direct DownloadDownload virtual-couple-travel-vlog.tar.gz
    sha256: c85905174ced23a9…
    Option B · Copy Guide for Agent
    install.json
    Agents automatically download and verify mirrors from this JSON guide.
    ② Upstream GitHub
    GitHub Repo ↗
    Pinned to commit fcee81beb6.