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

Human + product / 01:56 / Published 2025-02-04

OpenAI deep research in practice.

A customer-story format uses environment, interview, industry B-roll, and intermittent UI evidence. The most useful software details are the mode selector, research activity, a long report, and the close source list.

Watch official film ↗7 screen examplesAll 100 films
Representative frame from OpenAI deep research in practice.
OpenAI · representative sample 01:30.000 · Source moment ↗

Useful to borrow

Link the work to a real question, retain a little research activity, and end with readable sources. A result's provenance can be a strong proof shot.

Do not infer

Do not reuse the customer identity, endorsement, industry footage, or imply a research duration from the film's compressed runtime.

19 shot groups.

Read across from the observed content to its treatment and possible adaptation. Enlarge the frame for detail.

Approximate study intervals; horizontal scrolling reveals all columns on small screens.

Moment / sourceReference frameWhat is on screenFraming / editUseful adaptation
Shot group 01
≈ 00:00–00:07
People / environmentWatch from here ↗
Sample 00:04.000
Environment

Window and office shots establish a working setting.

Wide and medium environmental coverage.Use context only when it explains the work.
Shot group 02
≈ 00:07–00:10
App / browser screenWatch from here ↗
Sample 00:08.000
Composer / tools

The deep-research choice is visible in the composer.

Centered application fragment.Show the mode entry clearly.
Shot group 03
≈ 00:10–00:17
People / environmentWatch from here ↗
Sample 00:14.000
Interview / title

A person at a laptop leads into name and role coverage.

Live action with title overlays.A customer-story reference, not an all-hands default.
Shot group 04
≈ 00:17–00:25
People / environmentWatch from here ↗
Sample 00:20.000
Environment / portrait

Portrait strips and office reception coverage establish context.

Collage and live-action cuts.Do not borrow customer branding or endorsement.
Shot group 05
≈ 00:25–00:32
People / environmentWatch from here ↗
Sample 00:28.000
Research materials

Paper, a hand pointing at a screen, and files create work-context inserts.

Close tactile details.A physical screen is not necessarily a product demo.
Shot group 06
≈ 00:32–00:38
People / environmentWatch from here ↗
Sample 00:36.000
Physical laptop

Laptop use appears in varied shot sizes.

Small framed shots and wider coverage.Keep the relevant action, not ornamental B-roll.
Shot group 07
≈ 00:38–00:42
App / browser screenWatch from here ↗
Sample 00:40.000
Composer / tools

The research mode becomes visibly selected.

Stable composer crop.Show the state change that starts the task.
Shot group 08
≈ 00:42–00:45
People / environmentWatch from here ↗
Sample 00:44.000
Physical laptop

Hands and a wide laptop shot bridge into the task.

Short live-action inserts.The user's separate recording can supply the human context.
Shot group 09
≈ 00:45–00:49
App / browser screenWatch from here ↗
Sample 00:48.000
Research activity

The conversation shows a research sidebar, then a close reading-status detail.

Whole response into activity crop.Make work-in-progress specific rather than decorative.
Shot group 10
≈ 00:49–01:06
People / environmentWatch from here ↗
Sample 01:00.000
Industry B-roll

Interview and city coverage lead to semiconductor manufacturing imagery.

Multi-shot contextual montage.Only use footage that genuinely belongs to the topic.
Shot group 11
≈ 01:06–01:08
People / environmentWatch from here ↗
Sample 01:06.000
Interview

The speaker returns before another screen example.

Tight interview shot.Avoid a redundant second presenter in the companion video.
Shot group 12
≈ 01:08–01:12
App / browser screenWatch from here ↗
Sample 01:10.000
Physical laptop / composer

An over-the-shoulder laptop shows a research request.

Oblique screen framing with the user in foreground.Useful for context, weaker for small-text readability.
Shot group 13
≈ 01:12–01:19
People / environmentWatch from here ↗
Sample 01:16.000
Prompt overlay

Large prompt-like text overlays the person and office.

Text over live action, not a literal app screen.Treat editorial annotations as annotations, not UI evidence.
Shot group 14
≈ 01:19–01:23
App / browser screenWatch from here ↗
Sample 01:22.000
Research report

A long research response fills the software view.

Direct screen view with visible report progression.Show report structure rather than every paragraph.
Shot group 15
≈ 01:23–01:25
App / browser screenWatch from here ↗
Sample 01:24.000
Physical laptop / report

The report is seen on the user's laptop.

Over-the-shoulder result insert.Use it for ownership/context, not detailed reading.
Shot group 16
≈ 01:25–01:27
People / environmentWatch from here ↗
Sample 01:26.000
Interview

A short speaker beat separates report details.

Medium camera coverage.The external talking-head stream can perform this role.
Shot group 17
≈ 01:27–01:31
App / browser screenWatch from here ↗
Sample 01:30.000
Citations / sources

The sources section becomes a readable close-up.

Report overview into source-list detail.Proof can be traceability, not just a polished result.
Shot group 18
≈ 01:31–01:52
People / environmentWatch from here ↗
Sample 01:44.000
Interview / environment

Interview, stairs, and outdoor portraits conclude the story.

Multiple live-action and framed-portrait cuts.Study the breathing room, not the specific footage.
Shot group 19
≈ 01:52–01:56
Title / identityWatch from here ↗
Sample 01:54.000
End card

A source wordmark sits over the office view.

Quiet environmental ending.A settled ending need not be a complicated animation.

Reference frame

Enlarged reference frame

Watch this moment on the official channel ↗