Separate extracted facts, generated descriptions, and team decisions
Technical metadata comes from the file. Aspect shows properties such as resolution, duration, codec, frame rate, and camera information beside the asset. Available details depend on the file and the information embedded in it. A missing camera value is therefore different from an AI description that misunderstood a scene. Generated information comes from analyzing content. Aspect starts AI indexing after upload, making visual content and speech searchable as processing completes. Configured custom fields can also use written guidance to populate values from the asset. A description such as “hands opening a small box on a kitchen counter” belongs in this group. Operational metadata records information supplied by the team. Campaign assignments, approval decisions, and usage rights belong here because their authority comes from a brief, a decision, or an agreement. A model might recognize a product in the box, but seeing that product doesn't establish which campaign owns the clip or whether its release covers paid advertising. Give these groups clear names in your field definitions. This lets an editor understand whether a value describes the footage, came from the file, or records a person’s decision. It also helps reviewers focus on the right evidence when a value looks wrong.
Choose the fields editors need to find and reuse footage
A metadata schema is the agreed set of fields, their types, and their meanings. A taxonomy is the controlled list of categories used within those fields. Build both around the questions your team actually asks when looking for footage. For the next product campaign, the editor needs product-handling shots and permission to use them in paid social ads. Create fields for the visible content, campaign assignment, and rights information:| Field | Type | How the value is supplied |
|---|---|---|
| Campaign | Text | A producer enters the exact campaign name from the brief. |
| Generated content categories | Multi select | AI suggests values from Person speaking, Product handling, Location view, Other, and Unclear. |
| Generated scene description | Text | AI describes visible subjects, actions, and setting. |
| Metadata review | Single select | A reviewer chooses Pending, Checked, or Needs correction. |
| Usage status | Single select | A responsible person chooses Unreviewed, Cleared for stated use, Restricted, or Do not use. |
| Usage scope | Text | The rights owner records permitted channels, territory, and relevant limitations. |
| Rights reference | Text | The rights owner records the release or license identifier and its location. |
| Usage expiry | Date | The rights owner enters the applicable expiration date from the agreement. |
Configure the fields before the next upload
In Aspect, custom fields are defined at the project level. Open Settings, select the project, open Metadata, and add each field with its appropriate type. For select fields, enter the agreed options before saving. Keep the same names and meanings in related projects when editors need to interpret them consistently. Configure content-derived fields with written guidance for automatic population as files are uploaded. Leave the operational fields under the team's control. Although a campaign name might be visible on a slate, the producer still needs to establish whether the asset belongs to that campaign, was licensed from somewhere else, or was uploaded as a reference.
Describe the visible subjects, actions, and setting in one or two factual sentences. Mention a product only when it is visible. Do not infer a product model, a person's identity, the campaign, or permission to use the footage. If a detail is unclear, describe the observable detail without guessing.For the category field, use guidance that matches the exact options:
Select all applicable categories. Use Person speaking when a person visibly speaks, Product handling when someone physically handles the product, and Location view when the clip shows the surrounding place. Use Other when the content is clear but none of these categories applies. Use Unclear alone when there is insufficient visible information to classify the content.Review the generated descriptions and categories against these instructions during ingest. If your categories depend on context outside the frame, such as whether a clip was intended as an advertisement, have a person provide that context instead of asking the description prompt to infer it.
Apply known campaign information in a separate pass
Once you've confirmed which assets belong in an upload batch, you can apply their shared values together. Aspect supports selecting assets and applying metadata values in bulk. Ask Aspect can also create fields and update values across multiple assets, which is useful when the batch is large. For example, first ask:Show the video assets in the Day 02 folder of this project. I want to review which files belong to the Autumn Product Launch campaign.Review the returned files, including whether the folder contains reference material or unrelated pickups. Once the set is correct, continue with a specific instruction:
Set Campaign to Autumn Product Launch for these selected assets. Set Metadata review to Pending and Usage status to Unreviewed.Ask Aspect shows the proposed changes for review before applying them, with approval frequency controlled by your preferences. Review the selected assets and values together before confirming the update. The producer supplies the campaign assignment, and Ask Aspect applies it to those files. Don't bulk-clear an entire shoot merely because one release is attached to it. Different clips may contain different people, products, licensed music, or locations. Apply confirmed usage values only to the assets covered by the decision, with a rights reference that another person can check.
Review clips from each type of footage before tagging the full library
Choose a batch that reflects the material your team uploads each week, including clips that are difficult to describe. For this product library, you might begin with 24 clips covering clear product shots, overlapping speech, brief appearances, dark scenes, and several activities within one recording. Add more clips when a camera setup, subject, or category is missing from that first batch. Wait for processing to complete before judging an absent result. Aspect exposes processing indicators, and different parts of indexing can become available at different times. Then use Display controls to select a list layout and show the generated fields alongside Metadata review. Open the footage to verify each value rather than judging it from the thumbnail alone. Keep a short review record with the filename, original value, correction, and reason. Three failure types suggest different changes:- Wrong label: A person pointing toward a box is tagged as handling it. Tighten the category definition and correct the affected asset.
- Missing label: A demonstration appears halfway through a longer interview, but only Person speaking was selected. Inspect mixed-content clips and consider whether the category is still useful at whole-file level.
- Unclear boundary: Reviewers disagree about whether opening packaging counts as product handling. Agree on the definition before labeling more footage.
Check the workflow by finding a usable shot
After you've reviewed the sample, give another editor a specific request: find a close-up of someone opening the product box for a new paid social edit. Start with the Campaign field and reviewed metadata, then use AI Search to describe the action. Inspect the filters Aspect creates and adjust any interpretation that does not match the request. Generated categories can narrow the search, but they should not be its only route. If a relevant shot is missing, remove the category filter and search the visual content directly. This helps distinguish a categorization problem from footage that is absent. Preview the matching moment and enough surrounding footage to confirm that the action fits the edit. Before adding a shot to the working selects, check its Usage status, Usage scope, Rights reference, and expiry against the proposed placement and publication date. A “Cleared for stated use” value only helps when the stated use is recorded and matches this new request. A person must resolve missing or conflicting information.
Keep the metadata useful through delivery and reuse
Assign one person to maintain field definitions and make new-category decisions. Review repeated corrections with the people who search the library. Add a field when it supports a recurring decision; retire an unused category only after checking the assets and workflows that depend on it. Aspect warns that structural field changes can affect every asset using the field. Treat metadata, approval decisions, and access settings as separate responsibilities. Aspect's approval workflow distinguishes approving a comment from manually updating an asset's review-status field. A rights label likewise does not change a share link or prevent someone from downloading a file. Review metadata visibility and sharing permissions when sending a collection, and expose the information recipients need without including internal rights notes unnecessarily. At the next campaign handoff, ask an editor who did not log the footage to find a specific shot and explain whether it can be used. If they cannot find it, look at the search terms and categories they tried. If they find the clip but cannot establish its permitted use, repair the usage record and rights reference before the next edit begins.Frequently Asked Questions
Automatic video tagging uses analysis of a video's content to add searchable descriptions or categories. This differs from reading embedded technical information such as codec and frame rate. In Aspect, configured custom fields can use written guidance to populate values from uploaded asset content, while AI indexing also supports visual and spoken-content search.
A responsible team member should supply campaign assignments, approval decisions, usage permissions, license restrictions, and expiration dates from the appropriate project or rights records. A visible product or spoken campaign name does not establish permission to reuse footage. Keep these operational fields separate from generated content labels, even when an assistant helps apply a confirmed value to many assets.
Yes. Aspect fills configured custom fields from asset content as files are uploaded. Define the field and its allowed values, provide written guidance or examples, and review the resulting values during ingest. Give reviewers a way to mark ambiguous footage for correction, such as an Unclear category or a Needs correction status.
No. Aspect can search indexed visual content and spoken words without a team first creating a tag for every object or phrase. Custom fields are most useful for repeated categories and operational constraints, such as a campaign or reviewed usage status. Combine those fields with a description of the shot and inspect the matching footage.
First confirm that the asset has finished processing, then inspect the footage and correct the relevant metadata value. Record whether the problem was a wrong label, a missing label, or an unclear category definition. Revise the field guidance when errors repeat, and separately review existing assets rather than assuming that a prompt change will update earlier values.
A custom field records a decision; it does not establish a license or automatically control sharing. A person must check the underlying rights record and maintain the field when circumstances change. Aspect also treats approving a review comment and updating an asset's approval-status field as separate actions.




















