Skip to content
Vidsly

Documentary channels

The documentary, shot from your script.

Vidsly turns a documentary script — a history, a true story or an explainer — into a narrated 16:9 video: a generated scene for every beat (a photoreal film still by default), the same people held from the first scene to the last, a narrator from 85 voices and an optional music bed.

0 credits / clip16:99:1670 scenes85 voices
Start a documentaryHow characters stay consistent
History
Documentary

Two 9:16 scenes made in Vidsly. Long-form runs the same pipeline in 16:9.

7:00
longest 16:9 film in practice
70 scenes × 6s clips
85
narrator voices
16 languages
1
reference per character, used in every scene
kept for the next episode
0
credits per clip on the unlimited lane
20 clips a day on Creator
01Who it’s for

Three kinds of documentary. One pipeline.

What each one gets from it — and the one thing each should know before the first render.

History

Periods, events and the people in them — a 1920s press room mid-run, a lighthouse in a storm — rendered scene by scene and animated with a planned camera move.

Watch for
There is no archive behind it. Every scene is a generated reconstruction, so a real photograph, a document or a dated map has to come from somewhere else.

True story

True crime, disasters, survival stories, scandals. Reporting on violence and crime is allowed — the moderation pass refuses explicit depiction, not difficult subjects.

Watch for
Scene images stay non-graphic whatever the script describes: no visible injuries, no gore, no weapons pointed at anyone. The narration carries the weight.

Explainer

Science, business, how things work — a narrator walking through an idea, with a scene built for every step instead of a stock clip that roughly matches.

Watch for
Scene images contain no words, numbers or labels, so charts, diagrams and on-screen figures are added in your editor.

02One pass

Research in. Film out.

You write the script and pick the voice. Casting, shooting, scoring and the final mix happen in one render.

  1. 01

    Script

    Paste the narration you have researched, or have the writer draft from a topic in any of 15 languages. Each scene lasts as long as its line takes to say.

  2. 02

    Cast

    Before a frame is drawn, every person and recurring place gets one locked description. Saved characters and uploaded references are drawn from the same sheet in every scene.

  3. 03

    Narrator

    85 voices across 16 languages. On every Azure voice pitch and pace each run from −50% to +50%, for a slower, graver read.

  4. 04

    Scenes

    One generated image per beat — a photoreal film still by default, one of 6 styles — then a camera move planned across the whole film, so the cuts read as one documentary rather than a slideshow.

  5. 05MP4

    Deliver

    An optional music bed under the voice, then one 1920×1080 H.264 MP4 — or 1080×1920 for a 9:16 cut-down. We email you when it is ready.

The same pipeline, script-first: how script-to-video works.

03Recurring people

The same faces, scene after scene.

A documentary depends on the audience recognising the people in it. Draw each scene from text alone and the lead’s face changes every cut. Here every scene image is drawn from one reference per person — never from the previous frame — so drift has no chance to compound across 70 scenes.

Cast several people in one film and keep them for the next episode. Two limits worth knowing: recognizable copyrighted characters are refused before you start, and photographs of real people are often refused by the image model’s safety check — describe a historical figure in words instead.

How consistent characters work

01
Harbour, dawn
02
Records room, noon
03
Cliff, storm

Three scenes from one Vidsly render: one saved character, three places, three kinds of light.

04Long-form

Long-form, with the arithmetic shown.

The 16:9 format allows up to 10 minutes across 70 scenes. Full videos render on the unlimited lane, where each clip is at most 6 seconds, so the real ceiling for one film is about 7 minutes. Each scene is one clip against your daily allowance.

A video needing more clips than the day has left starts now with what fits and finishes automatically after the midnight-UTC reset — over up to 5 days for a very long one. You never re-submit it, and it never costs credits. HD clips spend more of the same allowance: a 720p clip on Unlimited Premiere counts 2× and a 1080p clip on Unlimited Ultra counts 4×.

16:9
A 16:9 scene from the paid Director's Cut engine. Full videos on the unlimited lane use the same Director at a 480p source (720p on Unlimited Premiere, Pro and Studio; 1080p on Unlimited Ultra, Studio), so expect a softer picture.

Creator

Price
$29 / mo
Clips a day
20
New footage a day
up to 2:00
A 70-scene film
about 4 days

Pro

Price
$49 / mo
Clips a day
40
New footage a day
up to 4:00
A 70-scene film
about 2 days

Studio

Price
$99 / mo
Clips a day
60
New footage a day
up to 6:00
A 70-scene film
about 2 days
Start a long-form videoSee the plans
05Questions

Before you commit a series to it.

Can AI make a documentary-style YouTube video?
Yes — that is what this pipeline is for. You paste a researched script; Vidsly cuts it into scenes, generates an image for each beat (a photoreal film still by default, or one of 5 other styles), animates it with a planned camera move, narrates it and mixes an optional music bed under it, then returns a 1920×1080 MP4. What it is not is an archive: every scene is a generated reconstruction, so real photographs, footage and maps have to come from somewhere else.
How long can an AI documentary be?
The 16:9 format allows up to 10 minutes and 70 scenes, but full videos render on the unlimited lane, where each clip is at most 6 seconds — so about 7 minutes is the real ceiling for one film. On Creator a full 70-scene film needs about 4 days of allowance and finishes on its own after each reset; on Pro or Studio about 2 days. Longer stories work best as a series of parts.
Will the same people look the same in every scene?
Yes. Pick saved characters, describe them, or upload a reference image, and every scene image is drawn from one reference per person rather than from the previous frame — so the face in the last scene is the face in the first, and drift never compounds across a long film. You can cast several people in one video and keep them for the next episode.
Can I use photos of real historical figures?
Often not. Photographs of real people are frequently refused by the image model's safety check when the reference is turned into a character sheet — before you are charged — and recognizable copyrighted characters are refused up front. Describing the person in words (age, build, clothing, era) works, and that description is then held across every scene.
Is true crime or war history allowed?
Yes. The moderation pass lets violence, crime and conflict through as subjects, because reporting on them is the point. It refuses sexual content, hate speech, threats against real people, instructions for violent wrongdoing and graphic gore. Scene images are kept non-graphic whatever the script says — no visible injuries, no gore, no weapons pointed at anyone — so a battle or a crime scene is shown without being dwelt on. Refusals happen before anything is charged.
Which narrator voices are there?
51 voices in the main picker plus 34 more in the full library (85 in all), across 16 languages. 11 emotional styles besides the default are available on 18 expressive voices (up to 10 on any one voice), and on every Azure voice pitch and pace each run from −50% to +50% — slow it down and lower the pitch for a graver documentary read.
Does it add captions, dates or maps?
No. There is no subtitle step on the AI pipeline and no .srt file, and scene images are generated without any text — no dates, labels, signs or map legends. Captions, chapter titles and annotated maps go on in your editor. Burned-in captions exist only in the photo-video tool, where you bring your own images and audio.
Is there documentary music?
There are 5 music beds, and the default is none. The one made for narration-led films is “Atmosphere” (slow, gentle pad — voice-led and documentary). There is no tense or dramatic score in the library yet, so a true-crime or war film gets a soft pad or silence; if you need tension, add your own score in the edit.
Do I have to disclose that the video is AI-generated?
On YouTube, probably. YouTube asks creators to disclose AI-generated content that looks realistic — including realistic depictions of real people and of events — and labels the video when they do. Vidsly's default scene style is a photoreal film still, so a history or true-story video made in it is likely to fall under that rule; clearly non-realistic animation generally does not. An AI-written script on its own does not need disclosure.
What does a documentary channel cost to run?
Full videos run on the unlimited lane at 0 credits per clip — scene images included — on Creator, Pro and Studio: Creator is $29 a month, Pro $49, Studio $99. The bound is a daily allowance of 20 clips on Creator, 40 on Pro and 60 on Studio, resetting at midnight UTC, with one clip per scene. HD clips spend more of the same allowance: a 720p clip on Unlimited Premiere counts 2× and a 1080p clip on Unlimited Ultra counts 4×. A video needing more clips than the day has left starts now with what fits and finishes automatically after the midnight-UTC reset — over up to 5 days for a very long one. You never re-submit it, and it never costs credits. The 1920×1080 file is scaled up from a 480p source (720p on Unlimited Premiere, Pro and Studio; 1080p on Unlimited Ultra, Studio), so expect a softer picture than a native render.
Related

Keep exploring

Your research. Every scene shot for it.

0 credits per clip on the unlimited lane — 20 clips a day on Creator from $29/mo, up to about 7 minutes a film.

Start a documentarySee the plans