视频·图像·音频技能

Game audio principles. Sound design, music integration, adaptive audio systems.

中文简介正在准备中。当前显示原文说明。

Use when the user asks to generate, remix, poll, list, download, or delete Sora videos via OpenAI’s video API using the bundled CLI (`scripts/sora.py`), including requests like “generate AI video,” “Sora,” “video remix,” “download video/thumbnail/spritesheet,” and batch video generation; requires `OPENAI_API_KEY` and Sora API access.

中文简介正在准备中。当前显示原文说明。

Process multimedia files with FFmpeg (video/audio encoding, conversion, streaming, filtering, hardware acceleration) and ImageMagick (image manipulation, format conversion, batch processing, effects, composition). Use when converting media formats, encoding videos with specific codecs (H.264, H.265, VP9), resizing/cropping images, extracting audio from video, applying filters and effects, optimizing file sizes, creating streaming manifests (HLS/DASH), generating thumbnails, batch processing images, creating composite images, or implementing media processing pipelines. Supports 100+ formats, hardware acceleration (NVENC, QSV), and complex filtergraphs.

中文简介正在准备中。当前显示原文说明。

Build a HyperFrames video composition from a brief/script autonomously. Uses separated builder and critic subagents — builders never review their own work. Quality decisions are grounded in project context (identity, audience, goals) and established visual/motion design principles. Measurable criteria (text size, contrast, overflow) are verified with tooling, not visual judgment. Use when the user says "make this video", "implement this brief", "build this composition", or provides a video script/brief to implement.

中文简介正在准备中。当前显示原文说明。

Use when pulling or downloading a new llamacpp model. The active ROCm image (kyuz0/amd-strix-halo-toolboxes) fails to start in the ephemeral pull container without ROCm device access. Must temporarily switch to the standard CPU image.

中文简介正在准备中。当前显示原文说明。

Generate high-CTR YouTube thumbnails using Nano Banana Pro via the Novoads API. Handles reference image upload, character likeness alignment, proven CTR-tested prompt formulas, and bounded batch generation. Use when the user asks to create a YouTube thumbnail, video thumbnail, A/B test thumbnail variations, or refers to thumbnail design with their face, brand assets, or product photos.

中文简介正在准备中。当前显示原文说明。

Analyze a reference video and reverse-engineer its style into a reusable Seedance 2.0 prompting template for the Novoads API. The output is a new formula file — like seedance-2-ugc.md — that captures the video's structure, pacing, camera work, edit style and tone so it can be recreated with any product, any person, in any setting. Use this whenever someone provides a video they want to use as a style reference, says "I want to make videos like this", "deconstruct this video", "turn this into a template", "analyze this style", or drops a video file and wants to recreate that format repeatedly.

中文简介正在准备中。当前显示原文说明。

Clones a VIDEO ad end to end for a different product or offer, on the Novoads API. The deliverable is a finished rendered MP4 — the WHOLE ad, not its opening, and not a template. Reads the source's style, pacing, camera work, dialogue and tone, adapts it into a Seedance 2.0 prompt, prices it, renders it, files the source beside the clones and diffs each clip's transcript against the approved script. Use for "clone this video ad", "recreate this video for my brand", "make this ad but for my product" when the source is a video, "clone my competitors' VIDEO ads", or a video ad handed over with a product photo. **Among cloners the source's medium decides, not the wording**: "clone this ad" plus a video lands here; a still, or nothing attached, is clone-image-ad, and animating a still as-is is image-to-motion. This pack has no hook-only skill: for the opening alone, clone the whole ad and cut. If it does not move, wrong skill.

中文简介正在准备中。当前显示原文说明。

Generate one or more standalone Meta image-ad creatives via ChatGPT Image 2 (gpt-image-2) through the Novoads API. Locks the model, auto-strips platform chrome, enforces edge-safe layouts and glyph-safety inside body text. Use when the user asks for a "gpt-image-2 ad", "ChatGPT Image ad", "Image 2 ad creative", "make a static image ad with GPT", or anchors on a need for typography-heavy / dense-text / UI-mimicry ad creatives (chat threads, comparison tables, fake search results, iOS dialogs, Slack snapshots, ChatGPT-conversation ads, Apple Notes lists). Does NOT trigger on Nano Banana cues — use nano-banana-image-ad for those.

中文简介正在准备中。当前显示原文说明。

Turn any still image into a professional motion graphic through the Novoads API on Seedance 2.5. Covers how to read a reference image, motion vocabulary for any image type — UI, marketing hero, flat-lay, key art, collage, product, character — the prompt clauses that control camera, timing, frozen regions and text fidelity, and the upload mechanics. Use whenever the user supplies an image and wants it animated, moving, or turned into video — including "animate this", "turn this into a motion graphic", "make this move", "この画像動かして", "モーショングラフィックにして", "I want the cards to pop in", "I want the bars to slide up", or when they describe motion beats for a static image. Also use when asked to write a Seedance prompt for an existing image. Do NOT use for text-to-video with no source image, restyling existing footage (this skill takes no video input), cloning a reference video (use clone-video-ad), or a product photo that should seed a new scene rather than move as-is (use novoads-api).

中文简介正在准备中。当前显示原文说明。

Lays a music bed under a finished video — the music is trimmed and faded to the video's exact length and ducks beneath the voice, while the video stream is copied untouched. Use when the user asks to add music, add a background track, add a music bed or soundtrack, or asks for something like "make a lofi mix of this": "put music under this ad", "add a background track", "give me a couple of music versions". Entered from a finished cut — captions already burned is the preferred input, not a problem. It never edits the picture and never changes the video's length.

中文简介正在准备中。当前显示原文说明。

Clones a STATIC IMAGE ad into finished ads for your own product, into a reusable prompt TEMPLATE, or both — the template is what makes the ads and the ads are what prove the template. Round-trips the frame through gpt-image-2, nano-banana-pro or reve-2.1 until it reproduces, replaces every brand-specific with a {placeholder}, then fills it with your product pinned to a real photo. Use for "clone this ad for my product", "rebuild this competitor's ad for us", "make me ads like this", "reverse engineer this ad", "make this ad reusable", "add this to my prompt library", or "clone my competitors' ads" with NO file attached — it sources one via spy-competitor-ads rather than asking which. **Among cloners the source's medium decides**: a still, or nothing attached, lands here; "clone this ad" plus a VIDEO is clone-video-ad. NOT for generating an ad from a brief (chatgpt-image-ad, nano-banana-image-ad), nor animating a still as-is (image-to-motion). Refuses to render your product without a real photo of it.

中文简介正在准备中。当前显示原文说明。

Burn timed, styled captions into a finished video on the local machine, without re-rendering the source. This is the manual fallback to the API's captioning endpoint, POST /v1/captions, which stays the default path: come here when the style the user wants is outside the API's presets, when the wording has to be hand-corrected before it is burned into the pixels, or when the source was rendered with audio disabled and the API refuses it. Transcribes the audio with Whisper, groups words into reading phrases, renders the captions in a real browser via HyperFrames, and composites them over the source with ffmpeg using a real alpha channel. Use when the user has an MP4 with speech and asks to add captions, subtitles, burned-in text, TikTok-style captions or word-by-word text to it, and the API path has been ruled out. Runs out of band, so it costs no credits and works on any video file whoever generated it. Not for generating video (use the video skills) and not for a separate .srt/.vtt sidecar file.

中文简介正在准备中。当前显示原文说明。

Overlays silent b-roll cutaway clips onto a finished base video — the base audio keeps running underneath while the picture cuts away and returns, and the final duration always equals the base duration. Use when the user asks to add b-roll, add cutaways, or overlay clips on a video: "add b-roll to this", "cut away to a product shot here", "overlay these clips on my UGC ad". Entered from an approved base video plus its transcript, which comes from POST /v1/transcripts and needs no local install; it never generates the base itself and it never extends a video.

中文简介正在准备中。当前显示原文说明。

Generate one or more standalone Meta image-ad creatives via Nano Banana Pro through the Novoads API. Locks the model, auto-strips platform chrome, enforces edge-safe layouts. Use when the user asks for a "Nano Banana ad", "Gemini image ad", "nano-banana-pro ad creative", "make a static image ad with Gemini", or anchors on a need for photoreal / lifestyle / multi-reference / handheld-object / clay-texture ad creatives (sticky-note flatlays, held-whiteboard signs, lifestyle portraits, ingredient collages, OOH photography). Does NOT trigger on ChatGPT Image cues — use chatgpt-image-ad for those.

中文简介正在准备中。当前显示原文说明。

MATLAB workflow for signal processing, DSP, image processing, computer vision, lidar, medical imaging, deep learning, and AI-assisted reproducible experiments. Use for filters, spectrograms, wavelets, object detection, segmentation, classification, training, inference, and generated figures.

中文简介正在准备中。当前显示原文说明。

Create a new Remotion video

中文简介正在准备中。当前显示原文说明。

Preview a Remotion video

中文简介正在准备中。当前显示原文说明。

Export a Remotion video

中文简介正在准备中。当前显示原文说明。

Create colorful infographic PNGs from a conversation, topic, file, skill, or workflow. Use when: インフォグラフィック, 図解, 挿絵, 章扉, X投稿用画像, OGP, HTML→PNG, visualize session, poster image. Produces 2-3 self-contained HTML patterns, renders static PNGs, and verifies every image visually before reporting. Not for animated GIFs, architecture/sequence/flow diagrams (editable .drawio or interactive HTML), or HTML slide decks.

中文简介正在准备中。当前显示原文说明。

GPU-optimized OCR using Surya. Use when extracting text from images/screenshots, recovering image-only content in PDFs or slides, or inspecting multilingual document layouts. Preserve source locations and report recognition gaps separately from verified text.

中文简介正在准备中。当前显示原文说明。

Prepare video URLs or local video files for GitHub Copilot analysis by extracting captions, sampled frames, contact sheets, and a prompt packet. Use when the user asks Copilot to watch, inspect, summarize, or diagnose a video, screen recording, demo, webinar, YouTube/Loom/TikTok/X/Vimeo URL, or .mp4/.mov/.mkv/.webm file.

中文简介正在准备中。当前显示原文说明。

Turn any context (blog URL, pasted article, PR diff, meeting notes, release notes, raw prompt) into a narrated explainer mp4 with slides, subtitles, and optionally a talking-head avatar. Uses a free local stack (Pillow + edge-tts + ffmpeg) with optional SadTalker for the avatar. Use when the user says "動画化", "explainer video", "解説動画作って", "ブログを動画に", "記事を動画化", "PRを動画で説明", "議事録を動画化", "context to video", or wants an mp4 from arbitrary text/source. Output is mp4 + srt under the user's chosen project workspace.

中文简介正在准备中。当前显示原文说明。

Generate a single self-contained HTML file — a horizontal slide deck, a vertical-scroll explainer document, or a fixed-canvas summary image — with zero external runtime dependencies, then verify it mechanically. Use when the user asks for HTMLスライド, 単一HTMLスライド, single-file HTML presentation, ブラウザーで開く説明資料, HTML 説明資料, HTMLサマリ画像, or self-contained HTML, or wants to hand someone a deck or explainer that opens anywhere without PowerPoint. Also use to embed images into such a file or to re-check an existing one. Does not output PPTX.

中文简介正在准备中。当前显示原文说明。