prompts.

skills for builders who need Chinese short-video scripts that can be copied into Seedance 2.0 and published fast. Seedance Viral Forge 是一个面向即梦 / Seedance 2.0 的 AI 视频脚本仓库，重点不是“写好看的提示词”，而是产出能直接生成、发布、测数据的中文脚本。 ## What This Project Does This repository helps users generate structured Chinese AI video scripts for three real short-video jobs: - product-selling videos - merchant lead-gen dance videos - trend-driven traffic videos Instead of treating prompt writing like vague inspiration

Text-to-image diffusion models have demonstrated remarkable capability in generating realistic images from arbitrary text prompts. However, they often produce inconsistent results for compositional prompts such as "two dogs" or "a penguin on the right of a bowl". Understanding these inconsistencies is crucial for reliable image generation.

In this paper, we highlight the significant role of initial noise in these inconsistencies, where certain noise patterns are more reliable for compositional prompts than others. Our analyses reveal that different initial random seeds tend to guide the model to place objects in distinct image areas, potentially adhering to specific patterns of camera angles and image composition associated with the seed.

To improve the model's compositional ability, we propose a method for mining these reliable cases, resulting in a curated training set of generated images without requiring any manual annotation. By fine-tuning text-to-image models on these generated images, we significantly enhance their compositional capabilities. For numerical composition, we observe relative increases of 29.3% and 19.5% for Stable Diffusion and PixArt-α, respectively. Spatial composition sees even larger gains, with 60.7% for Stable Diffusion and 21.1% for PixArt-α.

At least 48GB of VRAM for running the CogVLM2 server, or 16 GB for the Int4-quantized version.

4. Generate a dataset with these seeds

AI漫剧赛道很热，但我们观察到一个现象：市面上大量课程和工具集中在"API转发"和"一键生成"的包装上，却很少有人认真面对漫剧创作的核心问题——**剧情**和**分镜**。

这个项目集成了 @山音 的编剧/导演/分镜方法论体系，尝试用 AI Agent 辅助走完从原始素材到AI平台可用分镜的全流程。它不是替代创作者，而是做创作者的工具。

从小说到AI视频，一站式完成：**小说改编 → 剧本创作 → 导演定调 → 分镜拆解 → AI视频生成提示词

核心原则：台词神圣不可改，所有画面必须能被摄影机拍到，每个镜头必须有叙事目的。

├── 巨日禄模式 → 三列Excel（分镜编号/分镜描述/台词）

generate an image

Qi Baishi style ink wash landscape

camera push, a cat jumping off sofa

convert to watercolor style

A Claude Code custom skill that turns Claude into an AI content creator for ByteDance's **Seedance 2.0** (即梦) platform. Generate images and videos directly via dreamina CLI, or get production-ready prompts for the web UI.

Write me a Seedance prompt for a young surfer riding a wave at sunset, tight close-up, with explosive energy

Multi-shot scene: traceur running across Parisian rooftops, then jumping between buildings, then landing and looking at camera

Cinematic Seedance prompt for a woman walking through a neon Tokyo street, with ambient rain audio

**Multi-shot bracket syntax** for generating cinematic multi-shot scenes in one prompt

**Camera movement encyclopedia** (dolly, tracking, orbit, aerial, handheld, locked-off)

The most powerful Seedance 2.0 endpoint. Generate video conditioned on any combination of reference images, videos, and audio - style transfer, lip-sync, motion transfer, character consistency, and scene composition all in one model.

**智能分镜生成** — 自动拆解为分镜镜头，含景别、运镜、动作、对白、氛围等

**分镜帧生成** — 为每个镜头生成首帧图片