Find your next open-source project

Search by name, purpose, or keyword — or paste a GitHub repository URL

liangdabiao/seedance-storyboard-generator-realman

Media & Design

A skill pack that turns stories and scripts into photorealistic short-drama storyboards and Seedance video prompts.

149
39
2 days old

Overview

Seedance Realman is a skill pack for AI coding/agent environments that converts a story outline, novel excerpt or script into a complete short-drama production pipeline. Its distinguishing goal is a live-action film look rather than cartoon, animation or plastic game-render aesthetics. It handles scriptwriting and episode splitting, character/scene/prop image prompts, Seedance video prompts with timelines and lip-synced dialogue, and a final editing stage. The target user is a creator who already works with an agent environment and wants a repeatable storyboard workflow rather than a general-purpose video editor.

The core method is the "six pillars of realism": photographic medium, skin and material detail, spatial anchoring, physical causality, breathing and micro-expressions, and synchronized on-camera dialogue. These are documented in the repository's references/真人感电影写实指南.md, which the README calls the single source of truth.

Getting started

  1. Install the Jimeng all-in-one canvas CLI (dreamina-canvas) from the official install page and log in, since image and video generation runs through that CLI.
  2. Install the skill into your agent environment (Codex, WorkBuddy and similar); the repository is the install source.
  3. Give the skill a story, e.g. "Write a story about Feynman solving a math problem at Princeton with his own method, as a short drama."
  4. The skill proceeds through the four stages 文 → 资 → 视 → 剪: script and episodes, asset image prompts and reference images, Seedance video prompts and clips, then stitching into an MP4. The README stresses generating asset images before videos to avoid face changes and scene drift.
  5. Before paid generation, the skill queries a free quote endpoint and shows the estimated credit cost for confirmation.

The full worked example (Feynman's "A Different Box of Tools", 5 episodes × 15 seconds) is in example/, including scripts, asset prompts, per-episode prompt files and extracted frames; the finished film is linked on Bilibili. Without the Jimeng CLI, the skill can still write scripts and prompts but cannot generate images or videos.

Constraints

  • Image and video generation require Jimeng credits and a logged-in Jimeng CLI; the README lists 480p 15-second video at about 105 credits and 720p at about 285 credits.
  • Defaults are Seedream 5.5 for images and Seedance 2.5 Flash for video at 480p, with aspect ratio chosen by subject matter; users can override model, resolution and style.
  • The workflow depends on asset images being produced before video prompts; skipping that stage undermines character and scene consistency.
  • No license is declared, and the finished MP4 is not distributed in the repository.

HotRepo's take

The project appears on HotRepo's new-project board for 2026-11-17, recording 121 stars and 32 forks with 20 stars gained during the roughly one-and-a-half-hour observation window. That is a short growth series, so it supports noticing early interest but not a longer trend. The repository contains no detected programming-language code; its value lies in the skill instructions, the realism guide and the archived example, which readers can inspect without running anything.

Written by AI from the project's README, updated . The project's own docs are the source of truth.

You might also like

  1. Stonewuu/ai-fusion-videoAgent-driven workspace to script, storyboard, generate, and assemble short-drama videos.Java1,711
  2. SAOS-3624/dsh-video-storyboardPacks a video into one storyboard image plus a timestamped transcript for LLM agents to understand.JavaScript106
  3. konlu/story-studioAn agent skill that turns a topic into a scripted, voiced, illustrated short video and an editable JianYing draft.Python307
  4. woniuniuniu/dsh-media-deskA DeepSeek Harness plugin that turns a one-line idea into scripts, covers, images, storyboards and post-performance reviews.JavaScript1,403
  5. KadoBOT/ComfyUI-H3-Continuity-StudioA ComfyUI custom node pack for planning and rendering multi-shot MiniMax H3 video sequences with visual continuity across clips.Python104