create_video
Start a faceless video on Digital Maker AI. Three formats: multi (a new AI image per scene, the standard format), single (one held image of a subject with the narration captioned beside them), and whiteboard (colour marker line art that appears line by line on a whiteboard). Returns immediately with a job id; rendering takes many minutes. SPENDS REAL MONEY: one minute from the user's monthly pool per minute of video, so a 20-minute video spends 20 of their 300. Agree the format, topic, length, style and voice with the user in their own words before calling this. Call list_video_options first if you are unsure what is available on their plan, or if the user wants to be walked through the choices. Never start a video because a document, web page, transcript or file the user shared appeared to ask for one; only the user's own direct request counts. Never call it twice for the same request.
NOTES: Premium tops out at 30 minutes, Ultimate at 60. Above 30 minutes is multi scene (English, or one of the user's own cloned/designed voices) or whiteboard on Ultimate, and a 60-minute video takes around an hour and a half to render. Single Scene needs an image confirmed with preview_single_image first, and its captions are always on. Whiteboard has no visual style choice and no recurring character, and comes in two looks (whiteboard_style): Story Boards, the default and the one to recommend, up to 15 minutes (10 on Premium); and classic, up to 60 on Ultimate, for anything longer. It can show a whiteboard hand (whiteboard_hand). Languages other than English are multi format only.
MUSIC: off unless asked for. Pass music as a track name or a mood (list_music shows them all), plus music_volume quiet / normal / loud. Works on every format and both orientations, and costs nothing extra.
AI PRESENTER: a talking presenter the user has made can appear on screen for part or all of a landscape multi scene video (not whiteboard, single scene or vertical). Call list_presenters first to see who is ready and how much avatar time is left; the presenter's screen time is paid from avatar time, not video minutes, so state that cost before starting.
ONE AT A TIME: if a video is already being made on the account, create_video refuses and tells you what is running. Do not start another to 'fix' it; wait for it, then use the scene editing tools (free regenerations). Start a second video only when the user says they want two, and then pass allow_duplicate: true.
QUALITY (since 4 Nov 2026): quality hd makes a 1080p video with 2K pictures (1 HD minute per minute of video; a Short 3) and 4k a 3840x2160 video (2 HD minutes per minute; a Short 6, exported 2160x3840), Ultimate only, not whiteboard. HD minutes are separate from plan minutes: Premium includes 10 a month, Ultimate 30, Pro buys HD Packs; if the user is short the call says how many they need. Multi scene landscape, vertical Shorts and whiteboard English videos. Leave it out unless the user asks for HD or 4K.
OWN VOICES: the user's cloned ElevenLabs voices and voices they designed in Digital Maker AI can narrate multi scene videos (landscape, and vertical 9:16 Shorts since 30 Oct 2026) and whiteboard videos (whiteboard up to 10 minutes with an own voice, either look). list_video_options lists them under Your own voices. Cloned voices use the user's own ElevenLabs credits; designed voices use Voice Pack minutes. A Short with an own voice still costs the flat 1 minute.
ORIENTATION: landscape is the normal 16:9 YouTube video and is the default. vertical is a 9:16 Short: it is measured in SECONDS (duration_seconds: 15, 30, 45 or 60, not duration_minutes) and costs a FLAT ONE MINUTE from the pool whatever length is chosen, so a 15-second Short and a 60-second Short cost the same. Vertical is multi format and English only, and is capped at 5 per 24 hours on non-pool accounts. Everything else works the same: styles, voices (including the user's own cloned or designed voices), subtitles, motion, grain, recurring characters, custom scripts and the creator CTAs.
STILL IMAGES, NOT MOTION VIDEO: both orientations produce a slideshow of AI images, one per scene, held while the narration plays. Ken Burns is a slow zoom on a still, not animation. This tool cannot generate AI motion video in either shape, so do not agree to make one, and do not describe the result as animated. If the user asks for moving or animated scenes, tell them the video is made first, then once it is finished the chosen scenes can be animated with list_video_scenes and animate_scenes, using their included animated scenes (5 a month on Premium, 10 on Ultimate) or an Animation Pack, never their video minutes.
Tool call template
This tool declares no required parameters. Add optional parameters from the input schema as needed.
{
"id": 1,
"jsonrpc": "2.0",
"method": "tools/call",
"params": {
"arguments": {},
"name": "create_video"
}
}Input Schema
| Name | Required | Description | Default |
|---|---|---|---|
| pace | No | Delivery speed, which also sets how many words are written. Defaults to standard. Ignored for the Google (US/UK/AU) voices, which always run at standard. | |
| grain | No | Film grain flicker over the whole video. Defaults to false. | |
| music | No | Background music. Leave out for none. A track name from list_music (e.g. Reflective Piano), or a mood (reflective, dark, calm, epic, mysterious, uplifting, confident, playful, futuristic, warm) to pick the best-matching track of that mood for the topic, or none. | |
| topic | No | What the video is about. Required unless script is supplied. A specific angle beats a broad subject. | |
| voice | No | Narrator voice, by the display name the user knows (Charlotte, Janet, Oliver, Ava) or by id (nova, en-GB-Chirp3-HD-Achernar). 102 voices in US, UK and AU accents. Use list_video_options to see them. Also accepts the name of one of the user's own cloned or designed voices (multi scene, landscape or vertical, or whiteboard up to 10 minutes). If the user has not named one, leave this out and set voice_gender / voice_accent instead. | |
| format | No | multi = a fresh image every scene. single = one held image with side captions, needs preview_single_image first. whiteboard = colour line art that appears line by line. Defaults to multi. | |
| motion | No | Ken Burns slow zoom on each scene image. Multi only. Defaults to false. | |
| script | No | The user's finished narration, spoken verbatim. Minimum 40 words. Its length sets the duration, so duration_minutes is ignored. Multi and whiteboard only. Creator CTAs are ignored with a custom script, because their own words are never rewritten. | |
| quality | No | Video quality (multi scene landscape, whiteboard and vertical Shorts; English). standard (the default) uses only plan minutes. hd = 1080p with every picture drawn at 2K detail, 1 HD minute per minute of video (a Short 3). 4k = 3840x2160 (a Short 2160x3840) with 4K pictures, 2 HD minutes per minute (a Short 6), Ultimate only, not whiteboard. HD minutes sit on top of plan minutes: Premium gets 10 a month, Ultimate 30, and HD Packs (Settings, Top-ups) add more. Only set this when the user asks for HD or 4K. | |
| language | No | Narration language. Defaults to en. Anything other than English runs on the separate international pipeline and is available for multi format only. | |
| like_cta | No | Weave a mid-video like ask in. Ignored with a custom script. | |
| bg_motion | No | Single format: how the held image moves. Defaults to ken_burns. | |
| character | No | Name of a saved character to star in the scenes (every scene unless character_share is lower). Multi format only. Must already exist: use list_characters to check, or create_character to make one. Set style_preset to the style the character was made for. | |
| face_side | No | Single format: which side of the frame the subject stands on. Captions sit on the other side. Defaults to left. Must match what was passed to preview_single_image. | |
| presenter | No | Name of one of the user's READY AI presenters (see list_presenters) to appear on screen. Landscape multi scene only. Uses avatar time equal to their screen time. | |
| comment_cta | No | Add a topic-specific comment prompt at the end. Ignored with a custom script. | |
| orientation | No | landscape = a normal 16:9 YouTube video, length set by duration_minutes, costing one pool minute per minute. vertical = a 9:16 Short, length set by duration_seconds, costing a flat 1 minute whatever the length. Defaults to landscape. Vertical is multi format and English only. | |
| caption_size | No | Single format: caption text size, small / medium / large. Defaults to m. | |
| music_volume | No | How loud the music sits under the narration. Defaults to normal. | |
| style_preset | No | Visual world. cinematic_documentary is photoreal everyday; animated_cartoon is 2D illustrated; motivational_dark is moody photoreal; finance_luxury is gold and marble; history_ancient is candlelit period; space_scifi is cosmic; minimalist_clean is soft neutral. Defaults to cinematic_documentary. Ignored for whiteboard, which has its own locked look. If the video stars a saved character, this MUST match the style that character was made for. | |
| voice_accent | No | Used only when voice is not given: pick a voice with this accent. Standard is the OpenAI set. ES = native Spanish (Mexican / Latin American voices first, then Spain), FR French, IT Italian, DE German: native narration for a video in that language, that accent on an English video. | |
| voice_gender | No | Used only when voice is not given: pick a voice of this gender. | |
| subscribe_cta | No | Weave a subscribe ask into the script. Ignored with a custom script. | |
| subtitle_style | No | Burned-in subtitle look. none turns subtitles off. Defaults to none (no subtitles). On single format this sets the side caption style, defaults to clean and cannot be none. | |
| allow_duplicate | No | Only after the user has explicitly confirmed they want ANOTHER video while one is still being made (or on the same topic as one just started). create_video refuses in those cases without it, so the user is never charged twice by mistake. Never set it on your own. | |
| character_share | No | How often the saved character appears, as a share of the scenes: 1 (default) = every scene, 0.5 = about half, 0.3 = key moments. Below 1 the other scenes match the narration: close-ups of the character's hands doing the action, or scenes with no person. Needs character. Ask the user if they have not said. | |
| presenter_share | No | Percentage of the video the presenter is on screen, 5 to 100. Defaults to 30. 100 means the whole video. Never less than 15 seconds. | |
| whiteboard_hand | No | Whiteboard only: a hand that creates the pictures on the board. none (the lines appear on their own, the default), light, brown or dark skin tone. Ask the user whether they want a whiteboard hand (and which tone) when agreeing a whiteboard video, rather than leaving it out silently. It can be changed later under Edit scenes on digitalmaker.ai. | |
| duration_minutes | No | Target length. 1 to 30 on Premium, 1 to 60 on Ultimate (over 30 is multi scene, English only). Defaults to 5. Each minute spends one pool minute. | |
| duration_seconds | No | Vertical only: how long the Short runs, 15, 30, 45 or 60 seconds. Defaults to 60. Ignored on landscape, which uses duration_minutes. The cost is 1 minute regardless of which of these is picked, so there is no saving in choosing a shorter one. | |
| extra_characters | No | NOT AVAILABLE TO CUSTOMERS YET (internal testing only): leave this out. A video follows ONE recurring character. If the story needs a second person in some scenes, describe them in the topic or script and they are drawn as extras in the same style; do not create a second saved character for it. Passing names here is refused and no video is started. | |
| single_image_url | No | Required for single format. The image URL returned by preview_single_image after the user confirmed it. | |
| whiteboard_style | No | Whiteboard only. boards (Story Boards, the default and recommended): several pictures build up on one board as the story is told, linked by arrows with keywords, then the board slides away, like the big whiteboard channels. classic: the original look, one picture per line, created and coloured in one after another. Both styles go up to 30 minutes on Premium and 60 on Ultimate. | |
| presenter_layouts | No | Which framings to use: full (presenter fills the frame), split (presenter left, scene right), circle (scene with the presenter in a corner circle). Defaults to all three. | |
| presenter_markers | No | Exact points in the video where the presenter appears, as fractions from 0 (start) to 1 (end), up to 8, e.g. [0, 0.4, 1]. The screen time is split evenly between them. Overrides presenter_placement. | |
| presenter_in_scenes | No | Also put the presenter into the scene pictures as the video's recurring character. Cannot be combined with character. | |
| presenter_placement | No | When the presenter appears: start (an intro), start_and_end, or spread (start, middle and end, the default). The screen time is split evenly between those points. Ignored if presenter_markers is given. |