VidSeedsVidSeeds.ai

使い方ガイド

How to Use VidSeeds Across the Current Product

Code-checked walkthroughs for the workflows that exist today: projects, thumbnails, translation, dubbing, auto-clips, analytics, and account setup.

Create & Optimize

Start projects, diagnose uploads, tighten edits, and build thumbnails

How to Run the Full Project Workflow

Create a project from a local video, let the entry flow prepare the analysis, then continue through optimize, schedule, and results.

  1. 1.
    Choose publishing accounts first — Open /projects/new. If you already have connected accounts, the workflow starts with account selection before upload.
  2. 2.
    Upload a local video file — Drag in a local file on /projects/new. The page probes metadata, tracks extraction progress, and asks you to confirm the project name.
  3. 3.
    Finish the pre-create analysis — The New Project screen completes probing, extraction, transcription, metadata generation, and frame prep before the project is created.
  4. 4.
    Edit platform configs and publish — When you confirm creation, the app opens /projects/[id]/optimize. From there continue into schedule and results to publish or review status.

How to Re-Optimize a Published YouTube Video

Pick a published video, attach the source file, and push refreshed metadata back to YouTube.

  1. 1.
    Open the YouTube video picker — Go to /reoptimize and choose from the connected channel's cached video list.
  2. 2.
    Open the video-specific optimizer — Selecting a card takes you to /reoptimize/[videoId], where the workflow is tied to that published video.
  3. 3.
    Attach the original local video — The page asks for a local file so it can fetch, transcribe, extract frames, and generate updated metadata options.
  4. 4.
    Choose fields to update on YouTube — Review the generated title, description, tags, and optional thumbnail, then push only the fields you want.

How to Run a Pre-Upload Diagnosis

Upload a video before publishing to check audio, visuals, hook quality, retention risk, and platform compatibility.

  1. 1.
    Upload a local video — Open /diagnose and drag in a file. The page accepts local uploads and charges seeds only when a run completes.
  2. 2.
    Wait for the diagnosis pipeline — The flow extracts audio, transcribes, samples frames, runs visual checks, and synthesizes rule-based findings.
  3. 3.
    Review compatibility and prioritized fixes — Results include an overall score, per-platform publishability, content signals, rule checks, strengths, and prioritized fixes.
  4. 4.
    Reset and test another cut — Use the results page as a pre-upload QA pass, then run the check again after revisions.

How to Use Precision Trim

Upload a local file, let AI find hook, cut, and reorder suggestions, then export an improved edit.

  1. 1.
    Upload a video or audio file — Open /precision-trim and add a local media file from the drop zone.
  2. 2.
    Let AI build the critique plan — The analysis pipeline extracts audio, transcribes speech, and generates hook, cut, and reorder suggestions.
  3. 3.
    Review the plan in the editor — The flow continues to /precision-trim/editor where you approve the specific suggestions you want to keep.
  4. 4.
    Export the edited file — Browser export is available from the editor for supported local files, with a client-side size cap for final export.

How to Use Thumbnail Studio

Generate standalone thumbnails from a video, a structured brief, or a full prompt, then download or publish the winner.

  1. 1.
    Pick a source mode — Open /thumbnails and start from a video, a short brief, or a fully written prompt.
  2. 2.
    Add channel style or trend context — You can load connected account style, detect niche trends, choose a preset, and adjust overlay text generation.
  3. 3.
    Generate variants — The studio creates one or more thumbnails, lets you compare variants, and can open the selected result in the editor.
  4. 4.
    Download, export, or publish — From the results area you can download the image, export a first-frame video swap, or publish the thumbnail to YouTube when a channel is selected.

How to Refine a Thumbnail in the Editor

Use prompt-based edits on the dedicated thumbnail editor route after generating an initial result.

  1. 1.
    Generate a thumbnail first — Start in /thumbnails and create at least one result. The editor works on a saved thumbnail ID.
  2. 2.
    Open the edit route — Use the Edit Selected action to navigate to /thumbnails/[id]/edit?from=studio.
  3. 3.
    Describe the change — Enter an edit prompt like background, lighting, text, or composition changes and submit a new iteration.
  4. 4.
    Keep the best revision — Each successful edit replaces the current editor state with a new thumbnail you can download or keep refining.

Repurpose & Localize

Translate, dub, cut shorts, and package deliverables for upload

How to Translate a YouTube Video

Choose a YouTube video, generate translations, review them on the results screen, and publish selected localizations to YouTube.

  1. 1.
    Pick a YouTube video — Open /translate and select a video from the connected channel grid.
  2. 2.
    Choose target languages — The language dialog builds the seed cost from the number of selected languages before you start.
  3. 3.
    Review results on the session page — Completed jobs land on /translate/results?sessionId=..., where you can filter, edit, retry failed languages, and download everything.
  4. 4.
    Publish selected translations to YouTube — The publish flow applies selected title and description localizations through the YouTube localizations API.

How to Dub a Local Video or Audio File

Upload media, choose language and tier, then download dubbed audio or a remuxed video.

  1. 1.
    Upload a local media file — Open /dubbing and add a local video or audio file. The page extracts audio before enabling the dubbing controls.
  2. 2.
    Choose target language, accent, and tier — Once the audio is ready, set the target language, available accent, and automatic or professional tier.
  3. 3.
    Start the dubbing job — The page tracks pending and processing jobs, estimates seeds from duration, and keeps completed jobs visible for later download.
  4. 4.
    Download audio or rebuild the video — Completed jobs let you download the dubbed audio or remux the new audio back into the original video in the browser.

How to Turn Long Videos into Shorts

Upload a long-form video, find standout moments, then reframe the selected clips into vertical shorts.

  1. 1.
    Upload a local long-form video — Open /auto-clips and add a video file from the upload screen.
  2. 2.
    Let analysis find candidate moments — The pipeline reads duration, extracts audio, transcribes when possible, and falls back to visual analysis when transcript data is missing.
  3. 3.
    Choose moments on the selection page — The flow moves to /auto-clips/select, where detected moments can be reviewed and selected before clip generation.
  4. 4.
    Generate and reframe the clips — Selected moments are cut first, then sent through the reframing flow so completed shorts can be delivered as vertical exports.

How to Download the Shorts Package

After auto-clips finish, download individual shorts or a ZIP package with captions and upload notes.

  1. 1.
    Finish clip generation first — The route is populated after /auto-clips/select finishes creating at least one completed clip.
  2. 2.
    Open the delivery page — Successful runs redirect to /shorts-download?mode=shorts.
  3. 3.
    Review captions and metadata — The page builds title, description, hashtags, TikTok text, Instagram copy, and alt text from the primary generated moment.
  4. 4.
    Download the assets — You can grab single clips or a ZIP bundle with metadata files, platform captions, and upload instructions.

Analyze & Strategize

Channel intelligence, idea generation, monetization checks, and video autopsies

How to Use Intelligence

Run a full channel analysis and load 18 intelligence tabs in batches.

  1. 1.
    Start a channel run — Open /intelligence with a connected YouTube channel and trigger the analysis request.
  2. 2.
    Hydrate tab data in batches — The page loads audience, meaning, visual, integrity, pre-publish, metadata, competitive, retention, and other tabs through the sequential hydrator.
  3. 3.
    Inspect per-tab metrics and AI summaries — Each loaded tab renders typed analytics, while AI insight cards fetch deeper explanations for selected areas.
  4. 4.
    Turn it into an action plan — The pre-publish and action-plan views are meant to translate channel findings into next steps.

How to Generate Video Ideas

Generate ten ideas from your last 100 cached uploads plus trend context assembled by the service.

  1. 1.
    Load seed balance and channel context — Open /video-ideas with a connected YouTube channel. The page checks seed status before generation.
  2. 2.
    Generate ten ideas — The request sends the active channel, current language, and a fixed idea count of 10 to the generator.
  3. 3.
    Review strategic explanations — Results include idea cards, channel analysis, and the cached trend inputs used to shape the recommendations.
  4. 4.
    Use them for planning — The page is built as an ideation screen, so regenerate when you want a fresh batch rather than editing in place.

How to Generate a Channel Description

Pull recent channel context, generate a new description, then publish it back to YouTube.

  1. 1.
    Load a connected YouTube channel — Open /channel-description-generator and select the channel context you want to analyze.
  2. 2.
    Review current context — The page loads the current channel description plus recent video context from YouTube or saved project analysis.
  3. 3.
    Generate a new description — The POST flow sends channel name, current description, videos, language, and connection IDs to the generator.
  4. 4.
    Publish to YouTube — When you're happy with the suggestion, use the publish action to update the connected channel description.

How to Place Mid-Roll Ads with Captions

Analyze caption timing to place mid-roll ads at natural breaks, then jump into YouTube Studio.

  1. 1.
    Pick a YouTube video — Open /midroll-ads and select a video from the connected channel.
  2. 2.
    Require usable timecoded captions — The flow fetches captions first and stops with an error state when there is not enough timecoded transcript data.
  3. 3.
    Review suggested timestamps — Successful runs return suggested breaks with timestamps, confidence, reasoning, and surrounding context.
  4. 4.
    Open YouTube Studio — The results screen includes a direct studio link for the selected video plus a rendered guide explaining how to place the breaks.

How to Run a Video Autopsy

Run an autopsy on a published YouTube video and inspect the report before making fixes.

  1. 1.
    Choose a published video — Open /why and select a video from the connected YouTube grid.
  2. 2.
    Launch the autopsy report — The page switches from the grid into the report mode for the selected video and active channel.
  3. 3.
    Review the full diagnosis — The report combines performance context, best-practice analysis, key findings, and prioritized action recommendations.
  4. 4.
    Return and compare more videos — Use the back action to run the same autopsy flow against another video from the grid.

Manage & Operate

Return through the dashboard, reopen projects, manage accounts, and review voice identity

How to Use the Dashboard

Use the dashboard as the main return point for quick actions and recent work.

  1. 1.
    Open the hub — Go to /dashboard to see the welcome header, connect-channel banner, quick actions, and recent projects list.
  2. 2.
    Jump into the right tool — Quick actions link to the main flows like projects, thumbnails, dubbing, translate, diagnose, mid-rolls, and intelligence.
  3. 3.
    Continue recent work — Recent projects reopen directly on each project's current step instead of sending you back to the start.
  4. 4.
    Use it as the default return page — The dashboard is optimized for resuming work, not for deep editing, so treat it as your operating home base.

How to Reopen Existing Projects

Use the dedicated projects list to jump back into the exact current step for each project.

  1. 1.
    Open the projects page — Go to /projects to see the same recent-project list pattern as the dashboard, but focused entirely on project history.
  2. 2.
    Scan status and platform badges — Each row shows project status, connected platform icons or counts, and the created date.
  3. 3.
    Jump straight into the current step — Project links resolve to /projects/{id}/{currentStep}, so unfinished work opens where you left it.
  4. 4.
    Clean up stale projects — The list includes delete actions when you want to remove old draft or completed work.

How to Manage Social Accounts

Manage connected social accounts and group them for publishing flows. Available platform sections depend on the current domain.

  1. 1.
    Open social accounts — Go to /settings/accounts to see connected totals, account cards, and only the platform sections enabled on the current domain.
  2. 2.
    Connect or reconnect platforms — Each platform section exposes connect, add-another, reconnect, or disconnect actions based on connection state.
  3. 3.
    Organize groups for publishing — The page also supports account groups so you can bundle connected accounts for later selection.
  4. 4.
    Use those connections in publishing tools — Project account selection, thumbnail publishing, and YouTube-based workflows all read from this connection data.

How to Review Voice Identity

Inspect the per-channel voice fingerprint that VidSeeds derives from connected channels and re-run it when needed.

  1. 1.
    Open Voice Identity — Go to /settings/voice. The page loads each connected channel and any saved voice fingerprint records.
  2. 2.
    Review the fingerprint details — Each card shows personality, trait bars, signature words, and voice strength for the connected channel.
  3. 3.
    Re-analyze when the channel evolves — Use Re-analyze to refresh the fingerprint through /api/author-voice/analyze and then reload the saved profile.
  4. 4.
    Use it as your voice reference — This same voice data is reused by voice-aware generation flows such as project optimization and thumbnail prompt context.

よくある質問

Where does the project flow go after /projects/new?

Creating a project is only the entry step. The New Project flow prepares analysis before creation, then lands on /projects/[id]/optimize. The same project route family also includes /projects/[id]/analyze, /projects/[id]/schedule, and /projects/[id]/results.

Which pages need a connected YouTube channel?

YouTube-driven pages like Re-Optimize, Translate, Mid-Roll Ads, Why, Intelligence, Video Ideas, and Channel Description Generator depend on a connected channel. Upload-first pages like Projects, Thumbnails, Dubbing, Auto-Clips, Diagnose, and Precision Trim start from local files.

Does Translate publish to every platform?

No. The current translate flow creates a review session, lets you edit and retry languages, and publishes selected title and description localizations to YouTube.

Does Dubbing work from my YouTube library?

No. The /dubbing page is upload-first. You provide a local video or audio file, then the page extracts audio, runs dubbing, and lets you download the result.

What upload and export limits matter on these pages?

Server uploads are capped at 1 TB. Some browser-side export flows have stricter limits, including the 500 MB client-side transform cap used for local video rebuilds such as thumbnail first-frame export, dubbed video remux, and Precision Trim export.

Ready to start a real workflow?

Open the current product flows, start with a local file or connected YouTube channel, and work through the same routes the app uses today.