short-drama · AI Micro-Drama Production
A complete multi-episode micro-drama production pipeline: from a one-line premise, it delivers the series bible, character references, episode scripts, shot-by-shot video generation, multi-character dubbing, subtitles and BGM — assembled into publish-ready episodes.
Its defining capability is cross-shot, cross-episode consistency: character appearance, costumes, and art style stay locked throughout, avoiding the character drift and style breaks common in multi-episode AI shoots.
Supports horizontal and vertical formats; episode count and per-episode length are unconstrained by the tool and decided by subject matter and platform.
Example invocation: "Produce a 10-episode vertical urban micro-drama."
Full brief
Positioning
short-drama is a full multi-episode micro-drama AI pipeline: a one-line premise or adaptation source in, publish-ready episodes out. It covers every stage from concept to final cut — not a single-purpose video or dubbing utility.
Core capabilities
- Series bible: character profiles, relationships, story arcs, emotional beats, per-episode hooks — built once, locked for the whole series.
- Character consistency: character reference images are established first, then every shot is generated image-to-video — structurally preventing off-model faces and costume changes across shots and episodes.
- Episode scripts & storyboards: four-act causal structure, line-by-line dialogue annotation, per-shot timeline prompts.
- Native audio first: prefers dialogue and ambience generated natively by the video model, verified per shot with ASR; falls back to TTS dubbing only where the native take fails — no blanket robotic voiceover.
- Pipeline quality gates: missing keyframes, missing dubbing, or missing alignment data hard-block progress. Skipped steps are rejected, not silently degraded.
Workflow
- Build the series bible (dramatic promise, cast, arcs, style prefix)
- Write episode scripts and line-by-line dialogue
- Generate character/scene reference images and index them
- Shot-by-shot image-to-video (prepare → generate → audit → align; steps cannot be skipped)
- Multi-character dubbing, subtitles, BGM assembly
- Final episodes plus the full asset trail (resumable, reusable)
Inputs & outputs
| Input | Notes |
|---|---|
| Premise / source | Required — one-line plot or adaptation source |
| Aspect ratio | Required — horizontal or vertical (must be explicitly confirmed, never defaulted) |
| Episode count / length | Optional — decided by subject and platform |
| Visual style | Optional — e.g. urban, period, suspense |
Output: per-episode final.mp4, plus the complete working files (scripts, storyboards, dialogue, reference index).
Boundaries with adjacent skills
- auto-short-video: one-shot short-video pipeline; does not handle serialized multi-episode drama.
- video-script: script output only; no visuals, no final cut.
- ai-video-gen: atomic video-generation capability, invoked internally by short-drama.
Fit
- Solo creators launching a drama channel with no team and no budget
- Novel/IP holders converting text into video to test the market quickly
- Content teams trialling multiple premises at low cost
Before you start
- Image, video, and voice generation run on metered cloud APIs; bring your own keys in the project
.env. Keys stay on your machine — the skill never requests or uploads them. - Run one full episode first to validate style, pacing, and cost before scaling episode count.
short-drama is part of the Aiglade Skill library. Invoke it from the Aiglade chat box in plain language.