# recurring-show + character cloning support — spoolcast

engine-level rules added to spoolcast that make recurring video series + locked recurring-character casts work end-to-end

Type: project entry

Published: 2026-04-29

[Original page](<https://artlu.ai/project/recurring-show-character-cloning-support-spoolcast>)

[Browse this group](<https://artlu.ai/ai/groups/spoolcast.md>)

Parent project: spoolcast

Series: features

Recorded status: launched

Updated: 2026-09-16T16:22:25.455Z

Technologies: spoolcast, ffmpeg, Google Cloud TTS, gpt-image-2, kie.ai, Kling

## Details

**What is this?**
A capability addition to spoolcast: support for recurring video series with locked recurring characters that hold identity across episodes.

**What it does**
Documents the engine-level rules that make recurring shows work. Covers how to clone a real public figure into an anime caricature consistently (gpt-image-2 two-input redraw recipe), how to feed locked refs to kling 3.0 video without tripping content filters, how to handle the kling_elements lead-frame flash, how to handle ffmpeg concat timestamp drift on variable-duration clips, how to render captions correctly across resolutions, how to position captions to avoid covering on-screen text, how to use ssml for comedic pacing in chirp3-hd, why variable beat durations beat fixed-grid clips.

Validated against the daily ai news satire pilot — first recurring show built on the spoolcast engine. Show repo lives at `spoolcast-content/shows/news-anime-bot/` (new `shows/` subdir convention for series, sibling to existing `sessions/` for one-offs).

**How it works**
Nine engine-rule sections consolidated into VIDEO_OUTPUT_RULES.md in the spoolcast repo. Lives with the code, gets versioned alongside the pipeline. Future video projects inherit by reading it alongside existing spoolcast rules.

**Built with**
Spoolcast, kie.ai (kling 3.0 video, gpt-image-2 image-to-image), google chirp3-hd tts, ffmpeg.


Status and dates describe the published record, not a recent test of an external service. Missing details are not inferred. Public links do not imply an open-source licence or permission to reuse material.

[Directory](<https://artlu.ai/llms.txt>)