【Yingce】An open-source infinite canvas workspace for AI film and television creation, integrating multimodal generation, storyboard planning, asset management, and Agent workflows.
影策 is an open-source workspace for AI film and short-video production, using a freeform canvas and structured workflows to organize scripts, characters, scenes, storyboards, media assets, and generated text, images, video, and audio. It is for creators who manage audiovisual projects through local or trusted deployments, with features for task tracking, asset management, timelines, and AI-agent workflows.
This repository also configures its own agents. See what open-ai-canvas tells them →
Skill Claude CodeCodex
A method guide for directing AI-generated character performances through facial expressions, eye focus, body movement, and close-up shots. It also addresses keeping a character’s appearance consistent across scenes.
Skill Claude CodeCodex
A prompt-writing skill for 10-second vertical ASMR unboxing videos of collectible toys. ASMR means a close, calming audio-visual style built around sounds such as peeling, rustling, tapping, and objects clicking together.
Skill Claude CodeCodex
A prompt-planning guide for matching a visual idea to a style template and assembling the final image-generation prompt from six sections. It also lists common problems such as unclear text, mixed styles, and overcrowded layouts.
Skill Claude CodeCodex
A prompt-writing toolkit for MiniMax H3 video generation. It uses a six-part prompt structure, explicitly links reference assets to the video, and helps choose a visual style and keep shots consistent.
Skill Claude CodeCodex
A short-video script editor for improving how an original character, or OC, is introduced in the first 30 seconds of a micro-drama. It checks whether the character quickly shows kindness, strength, weakness, hardship, or relatable flaws.
Skill Claude CodeCodex
A set of planning methods for making multi-shot videos with Seedance 2.5, a video-generation model. It records which characters, places, and objects must stay consistent and checks the chosen mode and timing.
Skill Claude CodeCodex
A short-form product-selling video script skill for 15–30-second TikTok-style videos using a presenter, a product, and an optional mood. It describes product benefits through actions and scenes rather than reading a product description.
Skill Claude CodeCodex
A video-planning skill for turning a character reference image, a chosen theme, and a visual group such as zodiac signs into 3D Chinese-animation transformation video scripts and prompts.
Skill Claude CodeCodex
A set of methods for writing and performing comedy. It covers setups, contrasts, reversals, punchlines, escalating patterns, character jokes, timing, pauses, and arranging a full set.
Skill Claude CodeCodex
A creative planning assistant for developing an original character or fictional brand identity for short videos. It produces character settings, visual traits, recurring actions, props, world details, image prompts, and video topic ideas.
Skill Claude CodeCodex
A script and prompt workflow for making a 10-second, vertical TikTok-style user-generated video promoting an app. It uses an app screenshot, collects product details, creates a short storyboard, and can submit the confirmed prompt for video generation.
Skill Claude CodeCodex
A story-planning entry point that checks whether a scene is structurally worth producing, identifies the kind of work requested, and routes it to the right film-making module.
Skill Claude CodeCodex
A film-style planning module that sets the overall visual and color baseline for a project, including style references, palettes, color moods, and film-look simulations.
Skill Claude CodeCodex
A film-rendering planning module that turns visual intentions into a technical profile for each shot, including realism, materials, lighting response, rendering method, and renderer choice.
Skill Claude CodeCodex
A film-design module for creating a complete single-frame image of a character or setting, along with reusable character anchors and world constraints. A single-frame image is one finished picture showing one moment.
Skill Claude CodeCodex
A film-performance module that describes what a character’s body, eyes, and face do on camera, while keeping the character’s identity consistent across shots.
Skill Claude CodeCodex
A Chinese-language module about shot size, camera angle, composition, and lighting for image and video prompts. It explains how these choices control what the viewer sees, where the camera is, and how light behaves in a scene.
Skill Claude CodeCodex
A storyboarding guide for planning a sequence as a chain of shots. A storyboard describes what each camera shot shows, including framing, position, movement, action, lighting, dialogue, and how it connects to the next shot.
Skill Claude CodeCodex
A filmmaking guide for designing camera movement in still images and video. It turns story intent into named, measurable movements such as tracking, zooming, following, or orbiting.
Skill Claude CodeCodex
A Chinese-language module about directing sound for film and video, covering dialogue, sound effects, ambient sound, music, silence, and lip synchronisation. It treats these as separate audio tracks that can be planned in prompts.
Skill Claude CodeCodex
A filmmaking guide for checking how multiple shots join into a continuous sequence. It covers visual continuity, screen direction, transitions, prompt handoffs, and repairing breaks between generated clips.
Skill Claude CodeCodex
A review guide for checking consistency across a generated film project. It compares character identity, locations, visual style, and rendering choices across the different stages of production.
Skill Claude CodeCodex
A translation guide that converts film and rendering instructions into plain visual prompts for the Jimeng image and video platform. It separates the source technical details from the final language given to the model.
Skill Claude CodeCodex
A quality-control guide for reviewing generated video delivery files. It checks whether files meet defined text-to-video, image-to-video, first-frame, and first-and-last-frame requirements, then diagnoses failures.
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: