Picture-in-Picture Presenter

Presenter in a window

by openfilm · MIT license · ★ 975 on GitHub

Presenter in a window cover

Make a talking-head video where the content fills the screen and the presenter speaks from a small window: slides, screens, charts and pictures that follow what they say, their face always there. Use when the person gives a recording of…

---
name: presenter-pip
description: Make a talking-head video where the content fills the screen and the presenter speaks from a small window: slides, screens, charts and pictures that follow what they say, their face always there. Use when the person gives a recording of themselves talking (with slides, a screen, data or notes) and asks for a lesson, a knowledge video, a business review, a commentary or a presentation with them in a corner (小窗人物口播, 画中画, 口播, 讲解视频).
---

# Presenter in a window

Make the video that keeps the person present and puts what they talk about in charge: each thing they mention fills
the screen the moment they say it, and they speak from a window that never gets in its way.

## What has to be true

- **The picture follows their words.** Each slide, screen, chart, number or picture arrives as they say it and leaves
  when they move on. Check it in the frames against the transcript's word times, and say how close it is.
- **The window never covers what matters**: the content, the subtitles, or a phone app's buttons; their face is never
  cut badly.
- **Their words and meaning.** Cuts fall in pauses; ums and false starts go only when they ask; nothing appears that they
  did not say or give. A number on screen is the one they say, from their data or its source.

Overview

Make the video that keeps the person present and puts what they talk about in charge: each thing they mention fills the screen the moment they say it, and they speak from a window that never gets in its way.

Manual install (npx, ZIP)
Claude Codenpx skills add openfilm/openfilm --skill presenter-pip -g -a claude-code
Claude Code (degit)npx degit openfilm/openfilm/packages/openfilm/skills/presenter-pip#main ~/.claude/skills/presenter-pip
Claude.ai / desktopZip the folder and upload it under Settings → Capabilities → Skills.
Codex, Cursor, other agentsPoint them at SKILL.md. The other files are plain text.

Connects with Agent Alley MCP · agentalley.io/mcp

Files included

  • SKILL.md2.4 KB
Read SKILL.md here
openfilm/main1 file shown
SKILL.md
Show the full text35 lines
presenter-pip/SKILL.md35 lines · 2.4 KB

Presenter in a window

Make the video that keeps the person present and puts what they talk about in charge: each thing they mention fills the screen the moment they say it, and they speak from a window that never gets in its way.

What has to be true

  • The picture follows their words. Each slide, screen, chart, number or picture arrives as they say it and leaves when they move on. Check it in the frames against the transcript's word times, and say how close it is.
  • The window never covers what matters: the content, the subtitles, or a phone app's buttons; their face is never cut badly.
  • Their words and meaning. Cuts fall in pauses; ums and false starts go only when they ask; nothing appears that they did not say or give. A number on screen is the one they say, from their data or its source.
  • It holds up next to the best explainers you know. Watch the whole film as a viewer would, and keep refining until it does.

What you have

Their recording, cut by its in and out points, never re-encoded, its window shaped with CSS; its transcript with every word's time (npx openfilm get asr), from which Studio draws the subtitles. A version you treat (its background removed, its sound cleaned) is a new file under assets/, and the cuts stay trims of it. Their slides as pictures or rebuilt as pages, anything you can make charts, pictures and motion with, images and sound from get, your own built-in tools, and services the person has keys for. Ask before downloading anything large onto the person's computer.

With the person

What they ask for comes before anything here. The recording, their slides, data or notes, where it will be watched, and whether ums and pauses should go. Ask without holding up the work. When done, tell them in a few lines what you made, how close the content is to their words, and what of theirs would make it better.

1---
2name: presenter-pip
3description: Make a talking-head video where the content fills the screen and the presenter speaks from a small window: slides, screens, charts and pictures that follow what they say, their face always there. Use when the person gives a recording of themselves talking (with slides, a screen, data or notes) and asks for a lesson, a knowledge video, a business review, a commentary or a presentation with them in a corner (小窗人物口播, 画中画, 口播, 讲解视频).
4---
5 
6# Presenter in a window
7 
8Make the video that keeps the person present and puts what they talk about in charge: each thing they mention fills
9the screen the moment they say it, and they speak from a window that never gets in its way.
10 
11## What has to be true
12 
13- **The picture follows their words.** Each slide, screen, chart, number or picture arrives as they say it and leaves
14 when they move on. Check it in the frames against the transcript's word times, and say how close it is.
15- **The window never covers what matters**: the content, the subtitles, or a phone app's buttons; their face is never
16 cut badly.
17- **Their words and meaning.** Cuts fall in pauses; ums and false starts go only when they ask; nothing appears that they
18 did not say or give. A number on screen is the one they say, from their data or its source.
19- **It holds up next to the best explainers you know.** Watch the whole film as a viewer would, and keep refining until
20 it does.
21 
22## What you have
23 
24Their recording, cut by its in and out points, never re-encoded, its window shaped with CSS; its transcript with every
25word's time (`npx openfilm get asr`), from which Studio draws the subtitles. A version you treat (its background
26removed, its sound cleaned) is a new file under `assets/`, and the cuts stay trims of it. Their slides as pictures or
27rebuilt as pages, anything you can make charts, pictures and motion with, images and sound from `get`, your own
28built-in tools, and services the person has keys for. Ask before downloading anything large onto the person's computer.
29 
30## With the person
31 
32What they ask for comes before anything here. The recording, their slides, data or notes, where it will be watched, and
33whether ums and pauses should go. Ask without holding up the work. When done, tell them in a few lines what you made,
34how close the content is to their words, and what of theirs would make it better.
35 
View all files on GitHub →

Details

License
MIT
Stars on the repo
★ 975
Checked
Source
openfilm/openfilm ↗
Creator
openfilm · 8 listings
View creator

Similar skills

See all →

More from openfilm

SkillCut-Paper Collage Motionopenfilm/openfilm · by openfilmMake motion graphics in a cut-paper collage style, with photos, paper and print cut out and put together by hand. Use for explainers, promos, stories, intros and social posts, or when the person asks for a collage, cut-paper, scrapbook, zine or mixed-media look (剪贴风, 拼贴, 手账风, 杂志拼贴).★ 975 GitHub starsSkillTech Keynote Film Builderopenfilm/openfilm · by openfilmMake a product film in the style of a big tech launch keynote: the real product on stage, one confident claim at a time. Use for app, website, software and gadget launches, feature announcements and product intros, or when the person asks for a keynote, launch event, premium or minimal tech style (发布会, 产品发布, 新品, 高级感).★ 975 GitHub starsSkillVertical Clip Creatoropenfilm/openfilm · by openfilmTurn a long video (a talk, podcast, interview, lesson, livestream or vlog) into short vertical clips that each stand alone and hold attention from the first second. Use when the person gives a long recording and asks for clips, shorts, highlights or cut-downs for Douyin, Reels, Shorts or similar (长视频转短视频, 切片, 剪短视频, 高光, 拆条).★ 975 GitHub starsSkillLyric Video Creatoropenfilm/openfilm · by openfilmMake a lyric video for a song that the artist would release as the official one — a piece made for this one song, every word landing exactly on the voice. Use when the person gives a song (with its lyrics or an .lrc file) and asks for a lyric video, a music video with words, karaoke-style lyrics or a song post (歌词视频, 动态歌词, 歌词 MV, 卡拉OK).★ 975 GitHub starsSkillVideo Recreation Toolopenfilm/openfilm · by openfilmRemake a reference video in OpenFilm, as close as the person wants it, from frame for frame to its structure, rhythm, look and motion carrying their own content. Use when the person gives a video, as a file or a link, and asks for an exact copy of it, one like it, its style or pacing, or a remake for their product or topic (复刻, 照着做, 做个同款, 一模一样, 仿这个视频).★ 975 GitHub starsSkillScreen Demo Creatoropenfilm/openfilm · by openfilmTurn a screen recording into a polished product demo or tutorial: the view follows the action, dead time is gone, every step is easy to see, and the product looks as good as it is. Use when the person gives a screen recording (or a product to capture) and asks for a demo, walkthrough, tutorial, feature video or how-to (录屏演示, 产品演示, 教程, 操作视频, 功能介绍).★ 975 GitHub stars