AI Short-Drama Glossary: 20 terms explained
AI short drama, vertical micro-drama, novel-to-video, storyboard script, segments vs shots, character consistency, reference images, native audio, AI labeling and more — each with a one-line definition, how it works, common mistakes and related tools.
AI short drama
A serialized show produced with AI generation rather than live shooting, typically vertical, 60–120 s per episode, dozens of episodes per title. Pipeline: parse the novel/script → character and scene reference art → storyboard → segment video → voice → compositing. Differs from 'AI video' by cross-episode consistency and series structure.
Vertical micro-drama
A phone-first 9:16 drama format: 1–2-minute episodes, a hook in the first 3 seconds, a cliffhanger per episode; monetized by per-episode unlocks or ads; internationally led by apps such as ReelShort and DramaBox.
Novel-to-video
The workflow that adapts a novel into a video series automatically. The real barrier is the parsing layer: episode splitting, appearance text an image model can reproduce, debris-free scene descriptions, a conflict and hook per episode. Tools without parsing are just video generators.
Storyboard script
Per-shot text a video model executes: framing and camera, who is in frame, action written as cause not result, dialogue with speaker, lighting as source direction and colour temperature, seconds. In AI drama the storyboard text is the prompt.
Segment vs shot
A segment is one video-model call, usually 10–15 s (up to 30 s on Wan 3.0 and Seedance 2.5); a shot is a camera/framing change inside it, 3–6 per segment. Rule: long segments, short shots — fragmenting segments doubles cost, long shots drag.
Character consistency
Keeping one character's face, build and costume identical across shots and episodes. Achieved in four layers: reproducible appearance text, neutral-light white-background sheets, references injected per segment, and storyboards that list only who is in frame.
Reference image
An image passed with the prompt to lock a character's look, a scene's space or a prop's material. Limits vary (MiniMax H3 up to 9, Wan 3.0 up to 10, Veo 3.1 up to 3, Qwen-Image 3.0 1–3, Seedream 5.0 Pro up to 10 and 5.0 Lite up to 14); MiniMax H3 and Wan 3.0 cannot combine a first frame with references in one request.
Character sheet
A character's reference image: pure white background, frontal full body, key + fill + rim light, neutral expression, 9:16. The goal is recognisability, not beauty — any baked-in mood or lighting is inherited by every shot.
Scene plate
A people-free, debris-free scene reference: three layers of depth, materials, light source position. Test: 'if another scene were shot here, what shouldn't be there?' — broken mirrors, blood, crowds are plot debris and would appear in every episode before the event.
First / last frame
Images that pin a clip's opening or closing frame. First frames give continuity between segments, but on some models (e.g. H3) they exclude character references — continuity costs you the cast lock, the mechanical cause of blocking drift.
Native audio (audio-video generation)
The video model generates speech and ambience together with the picture, lip-synced to whoever speaks on screen. Supported by MiniMax H3, Seedance 2.x, Wan 3.0, Veo 3.1, Kling 3.0 and Vidu Q3, among others (Sora 2 also, until its API shuts down on 2026-09-24); models without it need post-production voice and lip-sync.
Reference audio
A 3–8-second voice sample bound to a character and sent with each segment so the voice stays consistent. H3 caps total reference audio at 15 s and 3 clips; matching from a preset library is two orders of magnitude cheaper than generating voices per character.
Dialogue timing
Converting a line into shot seconds by speech rate: ≈ 5 Chinese characters or 2.5 English words per second, +0.75 s per sentence-final mark. Too long and the model repeats or stretches; too short and the tail lands on the next shot.
Hook / cliffhanger
The per-episode ending suspense (identity reveal, unresolved emotion, reversal, rotated) and the attention grab inside the first 3 seconds. They decide completion and paid conversion, so the parser should produce them per episode.
Credit-based billing
How most AI video platforms bill: buy credits, deduct per call. Compare on cost per finished second or per series, and check expiry (Runway's Standard and Pro credits reset monthly while Max rolls over up to one month; invideo credits don't roll over; Vidu subscription credits last 30 days).
720p / 1080p / 2K tiers
Video output tiers. How much the lower tier saves depends on the vendor: on official lists checked 2026-09-12, 720p costs 50% of 1080p on Wan 3.0, 75% on the Kling 3.0 API and the same price on Veo 3.1 Standard. Draft at the low tier, finalize at the high tier; vertical delivery is 1080×1920.
Prompt rewriting (prompt extend)
Automatic LLM expansion of your prompt (on by default on Wan, Qwen-Image and others). Turn it off for drama storyboards: the rewriter overrides cut points, durations and facing rules.
Content moderation
Compliance screening of prompts and outputs. Chinese models reject gore, corpses and nudity terms (e.g. Wan's DataInspectionFailed). Whether a blocked request is charged varies: Google does not charge for blocked Veo videos, while Runway's API charges moderated generations the same as successful ones. Keep violence at cause and consequence in storyboards.
AI-generated content label
Visible or machine-readable labeling of AI-generated content — required by rules such as Article 50 of the EU AI Act and by platform policies on YouTube, TikTok and Meta, and carried in files through standards such as C2PA content credentials.
IAA / IAP drama
Two monetization models: IAP (pay to unlock episodes or subscribe) and IAA (free viewing paid for by ads). IAA earns per episode watched, so it rewards output volume and completion, which is where AI production cuts cost; many apps now mix both.