What you need before you start
LongVid is in reviewed early access. Submit a request and sign in with ChatGPT to link it to your identity. Signing in alone does not open the studio: the operator must approve access. Requesting access does not start a paid subscription.
You need your own supported text, image and voice provider accounts, API credentials and available usage balance. A provider connection marked configured means settings are saved; run a small sample to check the account and model actually work.
Download your production and finish the edit
Approved creators can download a production package from Guided or Videos. LongVid saves the production, checks the files and offers a ZIP containing the source media, script, scene order, timing notes and draft captions. Large productions use numbered packages; extract all parts into one folder.
Import the media into your editor and use the included guides to arrange the timeline and recreate effects. Finish and render the video there, then upload it to YouTube. Planning a release date in LongVid does not upload or publish it.
A video file, a media package and a timeline are different deliveries
The ZIP download contains source media and editing guides. Separate production JSON and script downloads preserve data and text. Neither format is an automatic project import for another editor.
| Delivery | What it contains | LongVid early-access status |
|---|---|---|
| Finished video (MP4) | The rendered picture and sound. Individual scene layers are not editable. | Requires the configured operator render setup. Invited-account self-service export is not available. |
| Media package | Source images, video, audio, script, edit notes and draft subtitles. Assemble the timeline in your editor. | Available to approved users: checked ZIP download with source media, script, scene order, timing notes and draft captions. Large productions use numbered packages. Effects must be recreated in the editor. |
| Editable timeline | A project interchange file plus linked media, with supported timing and edit decisions. | Planned. No editor or program version has passed a representative LongVid import test. |
| Production JSON / script | LongVid production data or narration text; media files are not bundled. | Available download controls. Not a finished film or a third-party editing project. |
What can I download?
Download your production as a ZIP package with your images, video clips, voiceover, script and a scene-order guide. Large productions come in numbered packages. LongVid checks for missing files first, and you can choose to download a draft while you are still working.
How do I finish and publish my video?
Download the package, unzip it and bring the media into your video editor. Follow the included scene-order and timing guides, make the final adjustments and export your video there. You stay in control of the final cut and upload it to YouTube yourself.
Choose a channel and define the video
Create a channel with its own visual style, writing rules and references. Give the production one question, a target length, a language and a working title. Documentary and sleep formats can use different pacing guidance. For other explainers and visual essays, make the tone and structure explicit in the brief.
Research connections can bring in public video references. Save why a reference is useful: an angle, a visual structure or a title pattern. Popularity is not evidence that a factual claim is correct.
Write and review the script in chapters
Start with an outline for longer videos. Review the opening promise and each chapter’s purpose before generating the full narration. You can also write or paste your own script.
Check the facts, remove unsupported claims and read difficult sentences aloud. Target length is a planning estimate. Spoken delivery and the final timeline determine the actual runtime.
Turn the narration into visual scenes
Scene planning groups passages by their subject, action, location or idea. The suggested shot length guides pacing; it does not force a new image every ten seconds. Review what each image should communicate and where the scene should change.
Generate a sample image before the batch. Use existing images with camera motion, generate selected animation, or import reviewed Pexels footage. Recurring characters can use an approved portrait and saved appearance rules. Check every image for continuity and relevance.
Fit the visuals to the voiceover
Choose an ElevenLabs voice and generate a short passage. Listen for pronunciation, emphasis and gaps. Narration is generated scene by scene; measured audio duration helps align the scene timing. If you rewrite a scene, replace the corresponding narration.
Guided and Editor use the same production. In the timeline, reorder, split and trim clips; change framing and camera moves; add supported transitions, captions, particles and chapter labels. Selective regeneration lets you work on the weak scenes without replacing everything.
Review the whole sequence before release
Check the beginning and end of every scene, narration sync, spelling, readable captions and abrupt transitions. Preview effects can differ from the final render, so inspect the actual exported MP4 where export is available.
Keep a record of provider usage, retries and manual corrections. One short finished test tells you more about your setup than an unreviewed batch of thirty minutes. AI output quality, character consistency and voice performance need your own real sample.
Separate studio access from provider usage
LongVid’s request form takes no payment details and starts no subscription. A paid LongVid plan is not currently offered through the site. This does not make a production free: connected AI providers bill usage under their own terms.
The public calculator estimates scene images and narration. Writing, animation, presenter clips, rendering, storage, provider subscriptions, taxes and your time sit outside that number. Review allowances in the studio and spending settings in your provider accounts.