Independent visual research / Public editionReviewed 04 Sep 2026
Motion fieldnotesOpenAI film study
About this study

A field guide to product filmmaking

Study the
moving parts.

How OpenAI films frame a screen, reveal a result, and move between ideas. A source-linked library of observed examples—not a mood board or a list of guessed timings.

100official source videos
1,075timecoded shot groups
637app / browser occurrences
37close motion studies

The film library.

Search across the studies or choose a format. A screen tutorial, an output reel, and a live-action story solve different editorial problems.

Representative frame from Introducing ChatGPT Atlas
001 / Product film01:37

Introducing ChatGPT Atlas

A repeatable product-film structure: short white chapter cards, a recognizable browser window, then tightly cropped evidence of one action. The shopping sequence stays with a single task through permission, activity, and a visible cart.

Representative frame from Introducing ChatGPT agent
003 / Human + product02:07

Introducing ChatGPT agent

Interview coverage and physical laptop shots surround short, highly cropped software examples. Compact browser activity cards lead to concrete files and tables. This is a contrast to the product-only films, not the default layout for a separate presentation video.

Representative frame from ChatGPT Images
005 / Output montage00:55

ChatGPT Images

Most of this film is an escalating montage of image variations, not an app walkthrough. A recurring subject keeps the collage coherent. Only near the end does a phone show the Images interface.

Representative frame from This is ChatGPT Images 2.0
006 / Output montage01:49

This is ChatGPT Images 2.0

A materially rich output film: words appear on physical-looking surfaces, then museum, print, diagram, comic, editorial, and photo examples change rapidly. Device mockups show imagery, not a verified OpenAI app workflow in the inspected samples.

Representative frame from Introducing Sites in Codex
010 / Presenter demo02:13

Introducing Sites in Codex

A presenter-led tour puts finished interactive sites near the start, then shows several different output layouts and finally sharing controls. The strongest lesson is demonstrating what was built, not merely showing the request.

Representative frame from Meet ChatGPT Work
011 / Product film01:40

Meet ChatGPT Work

A product-only film carries one lamp-project story across voice input, connected work, a spreadsheet, imagery, a deck, a chart revision, and mobile follow-up. The coherent task makes many surfaces feel connected.

Representative frame from 10 years.
095 / Retrospective01:07

10 years.

Archival-looking photos, research imagery, documents, and later product footage are arranged as small objects on white before an anniversary ending.

Reference frame

Enlarged reference frame

Watch this moment on the official channel ↗