Finding moments by shot and by story

Answered on 5 October 2026: all three recommendations accepted. Keep Shot Types, make Shots visibly discoverable, and separate camera framing from activity and story descriptions. The recorded decisions preserve the reasoning. Exact vocabulary, individual merges and implementation remain open.

What you approved

Choice	My recommendation
Keep a home for shot-based browsing	Keep the existing Shot Types group and its Dialogue and Establishing shelves. Do not adopt the older proposal to remove their group without a replacement.
Make that route discoverable	Show a clear Shots entrance alongside the other ways into moments. Opening it includes matching shots without requiring a hidden catalogue switch.
Untangle “Framing”	Separate camera framing from what happens in a scene and what it does in the story. Review the exact filter choices afterwards.

The original alternatives below remain as the review record. No new answer is needed for these three choices. Your examples of product shots, demonstrations, reveals, action and medium shots inform the definition review; they do not approve a complete filter list or removal of Talking Head.

Already decided — no new answer needed

A moment is a meaningful interval in a parent video. The same moment can fit several shelves, with one main shelf. A shelf is a named kind of moment that people browse; its group is a navigation heading. Each shelf has one group in the current model. Other matching descriptions provide additional ways to find the same moment.

You already said that production teams need establishing shots and close-ups, while other researchers need logo animations, product reveals and interface moments. Neither audience is being removed. A visibility label does not decide whether a shot has creative value. A brief logo animation also does not make its whole parent video a logo-animation piece.

The answered choices below concern the structure and presentation of those paths. The recorded decisions and Moments explanation remain the starting point.

1. Keep the shot group, or replace it with filters alone?

Recommendation: keep Shot Types as the group for Dialogue and Establishing. People looking for these references have an understandable starting point. The group is a heading, not an additional claim the analyzer must make about every moment.

The checked example is the taxonomy itself: Dialogue and Establishing are both recorded under Shot Types. The July model and current Moments page show that arrangement. An August proposal removes Shot Types but keeps those two shelves, without giving them another group. That would leave the agreed structure incomplete.

Counterexample: a logo animation is not forced into Shot Types because it uses a camera-like move. It keeps its appropriate shelf and can also carry a camera description when the evidence supports one. Keeping a shot group does not make all moment classifications mutually exclusive.

Alternative: remove the group and offer Dialogue and Establishing only through searchable filters. This could simplify the group list, but it also changes their role as shelves. It needs an explicit replacement for their existing memberships and for how people reach them. Merely deleting the heading would not do that.

What this affects: navigation, shelf membership and saved queries. It does not require duplicating a moment or moving it into a new video. Existing terms and links must remain interpretable during any later migration.

2. Should shot browsing have a visible entrance?

Recommendation: show a Shots entrance alongside the other moment paths. Its results include matching production shots. People should not have to discover an internal “catalogue” switch to see them. This does not mean putting every talking-head frame on the first screen or expanding every section by default.

The checked example is the current browse rule: Dialogue and Establishing are tagged as catalogue entries in the registry. The older frontend contract says catalogue entries never headline the shelf wall. That is a presentation choice, not evidence that they are unhelpful. Your later feedback explicitly values production-shot research.

Counterexample: someone who chooses Logo Animations should see matching logo moments, not all dialogue shots as well. A visible entrance expands discoverability; it does not weaken the chosen filter or mix unrelated results into every view.

Alternative: keep a secondary, collapsed catalogue section as the only browse entrance for these shots. Search can still find them, but newcomers are less likely to learn that shot research is available.

What this affects: discovery pages, filter explanations and query scope. A saved search for establishing shots must keep that meaning when opened as a tracker or a collection. Featured selections and quality review can remain separate from whether a matching moment is eligible to appear.

3. Should “Framing” contain several different kinds of description?

Recommendation: separate the questions. Camera framing describes how the view is composed, such as a close-up or point of view. A reaction describes what the shot presents. An interview describes a situation or presentation. A demonstration describes showing something working. They can overlap, so selecting one should not erase the others.

The checked example is the existing Framing list: it offers Establishing, Hero, Product Detail, Macro, Lifestyle, Reaction, Reveal, Demonstration, Interview, Point of View, B-Roll and Transition in one list. The inspected classifier instruction asks for exactly one of those twelve. That cannot faithfully describe every combination those words allow.

Illustrative counterexample: a close-up can show a person's reaction during an interview. “Close-up,” “reaction” and “interview” answer different questions about the same interval. They should not compete for one answer. This is an illustration of the boundary, not a new classification of a library clip.

Adobe's camera-shot guide distinguishes camera distance and angle from shots used to establish a setting or show a response. It supports checking those meanings separately; it does not determine BrandTrackers' final vocabulary or storage design.

Alternative: rename the mixed list “Shot descriptions” and allow several defined choices from it. This is a smaller conceptual change, but each value would still need a clear meaning and rules for which combinations make sense. Renaming alone would not fix the current one-choice instruction.

What this answer approves: either separating the questions or retaining one explicitly mixed, multi-choice description list. It does not approve the final choices, fields or automatic conversion of existing classifications. The proposed removals of Reaction, Lifestyle and Product Detail from other lists stay unresolved until their meanings are compared. Two entries sharing a word are not sufficient evidence that they are duplicates.

How this connects to the rest of the system

Step	Required consequence of the final decisions
Human meaning	Definitions explain what each question asks, what qualifies and what does not.
Relationships	One moment stays linked to its parent video and to every applicable, evidenced description. Navigation does not invent new content.
Classification	Instructions must ask the intended questions and permit legitimate combinations. A browse preference must not become a claim about what is visible.
Storage and correction	Preserve prior meanings and human corrections. Review compatibility and conversion before changing stored values.
Search and trackers	A selected description has the same meaning in browsing, saved searches and ongoing monitoring.
Research interface	Show why the result matched, open the relevant interval and retain access to its parent and collection.

The technical evidence note separates live checks, local code findings, earlier rulings and remaining audit work. That review found reasons to challenge the old proposal; it did not establish that the entire deployed pipeline already behaves correctly.

No answer is needed again for multiple shelves, a main shelf, useful production shots or the whole-piece versus moment boundary. Pricing, calls to action, broad video names and sponsorship were also answered on 5 October; their review record preserves the accepted meanings. Exact descriptions, merge candidates and implementation checks remain in the review plan.

Prepared 4 October; answered 5 October 2026. The recommendations below are the accepted direction; alternatives are historical context.

Last reviewed 5 October 2026.