> ## Documentation Index
> Fetch the complete documentation index at: https://aspect.inc/docs/llms.txt
> Use this file to discover all available pages before exploring further.
# Metadata
> Configure project metadata, edit values, generate fields with AI, and manage people and object recognition.
Every asset in Aspect carries metadata: information about the file itself and information your team adds to support its workflow.
Use this page as a reference for metadata settings and controls. For help deciding which fields to create, start with [Custom Metadata](/docs/workflows/metadata).
Aspect extracts technical details automatically. You can add custom fields for project-specific information such as status, campaign, category, rights, or QA.
## General metadata
The General tab shows information about the file. Aspect extracts technical details automatically during upload processing; use custom metadata fields for information your team creates or updates.
## Automatic technical metadata
When you upload a file, Aspect reads the technical metadata embedded in it. Depending on the file type, this can include:
* Resolution
* Duration
* Codec
* Frame rate
* Camera information
* GPS location, when the file contains it
* EXIF and IPTC data, when available
Open an asset and use its details panel to review the metadata Aspect extracted. Extraction happens automatically as part of processing a new upload.
## Custom fields
Custom fields let your team track information that is specific to a project or workflow.
Fields are defined at the project level, so each project can use its own structure. Common examples include:
* Approval or QA status
* Campaign or client name
* Usage rights
## Supported field types
Choose the field type that matches the information you want to store:
| Data type | Use it for |
| - | - |
| **Text** | Notes, names, identifiers, and other free-form information. |
| **Integer** | Whole numbers such as counts or ratings. |
| **Decimal** | Numeric values that need fractional precision. |
| **Boolean** | True-or-false values such as Cleared. |
| **Date/Time** | Shoot dates, deadlines, and expiration dates. |
| **Single Select** | One option, such as a production status. |
| **Multi Select** | Several applicable options, such as categories or talent. |
| **User** | A user associated with the asset. |
Single-select and multi-select fields can use colored options, making values easier to scan in the library.
## Create a custom field
1. Open **Settings**.
2. Select the project you want to configure.
3. Open **Metadata**.
4. Under **General Metadata**, click **Create attribute**.
5. Enter a name and choose its data type. For a select field, add the options and their colors.
6. Leave **Generate with AI** off to enter values manually, or configure it as described below.
7. Click **Create**.
The **Custom** tab lists your project fields; the **System** tab lists built-in attributes.
The new field is available for assets in that project.
Changing or removing a field can affect every asset that uses it. Confirm the field name, type, and options before applying a structural change.
## Add or update metadata values
Open an asset and use the metadata controls in its details panel to add or change values. You can also edit values directly in **List** view.
Creating or changing a field defines the structure for the project. Editing a value changes the information on the selected asset or assets.
For bulk work, select the relevant assets before applying the value, or use Aspect Agent to update many assets as one workflow.
Useful values are consistent and specific. For example, use the same status options across a project instead of entering several variations of the same term.
## Filter and search with metadata
Metadata can help narrow a large project before you browse or search.
Use filters to find assets based on technical or custom metadata, such as:
* Videos with a specific resolution or frame rate
* Assets marked Approved
* Files associated with a campaign
You can combine metadata filters with [AI Search](/docs/asset-intelligence/ai-search). This lets you match what an asset contains and the structured information your team has added.
For example, search for an interview quote while limiting the results to assets whose Status field is Approved.
## Use Aspect Agent for bulk metadata work
Aspect Agent can create custom fields and update values across multiple assets.
For changes that affect your library, review the selected assets and proposed values before approving the action.
Find, organize, and update project content with Aspect Agent.
## AI-generated values
Turn on **Generate with AI** when creating or editing a custom attribute to populate it from asset content. AI generation supports Text, Integer, Decimal, Boolean, Date/Time, Single Select, and Multi Select fields. User fields use manual values; people and objects have separate recognition controls.
### Choose the inputs
* **Visual** uses visual content from videos, images, and vector files.
* **Transcript** uses spoken content from video and audio files.
* Select both when the answer depends on what appears and what is said.
Write a **Prompt** that describes the desired value. For a select field, explain what each option means and when it should apply. For a text field, specify the level of detail you need, such as a one-sentence summary.
Prompts can contain up to 8,000 characters. The selected sources are supplied automatically; you can use `{{transcript}}` if you want to place the transcript at a particular point in your prompt.
### Watch new assets
Turn on **Watch** to run automatically on new assets added to the folders you choose. Set **Folders** and **Asset types** to limit its scope. The coverage count describes existing compatible assets in that scope; turning on Watch does not process them retroactively.
Click **Create** or **Save** to save the setup. For an AI-generated custom field with Watch enabled, the split-button menu also offers **Create & run on watch** or **Save & run on watch** to explicitly run on existing assets in that scope.
You can leave Watch off and run a field on a folder or an individual asset when needed.
## People recognition
Under **People Recognition**, click **Create person**. Enter the person's name and add reference photos, cropping each to the face you want to recognize. An optional description can help AI Search describe the person's appearance.
Wait for reference photos to finish processing before running recognition. A person needs an accepted reference photo to be recognized. Aspect compares references with faces in images and videos and records matches in the project's **People** field.
Enable **Watch** to check new assets in selected folders, or explicitly run recognition on existing assets. Saving a new person alone does not backfill the project. For an existing person with Watch enabled, **Save & run on watch** is available from the save-button menu once its references are ready.
You can also mark a person present or absent on an asset yourself. The bulk re-run controls described below let you choose whether to preserve those manual answers.
## Object recognition
Under **Object Recognition**, click **Create object**. Add a name, a description, and reference photos of the specific object you want to recognize. The description is required and helps Aspect understand the object.
The **Recognition policy** lets you configure four traits:
* Color / colorway
* Labels & serial numbers
* Accessories & attachments
* Wear & condition
For each trait, choose **Auto** to infer acceptable variation from the references, **Can vary** to allow differences, or **Must match** to require a match. Use **Additional notes** for guidance beyond those options.
Matches appear in the project's **Objects** field. As with people, wait for an accepted reference, then explicitly run on existing assets or enable **Watch** for new ones. You can correct an object's presence on an asset manually.
## Run on existing assets
AI-generated fields, people, and objects share run controls. The available actions depend on your access and whether the field or recognition setup is ready to run.
| Scope | How to reach it |
| - | - |
| **Current folder** | Open the folder in **List** view and open the metadata column's header menu. Choose a run action for that folder. For recognition, choose a person or object, or all people or objects. |
| **Watch folders** | Use the field, person, or object's run controls in Metadata settings, or the **Watch folders** group in the column menu. This scope is available when Watch is enabled. |
| **One asset** | Right-click its metadata value and choose **Run**. For People or Objects, select which enrolled person or object to check. |
A folder run can include subfolders using **Include subfolders**. Explicit folder and single-asset runs do not require Watch and do not inherit its folder or asset-type restrictions; they still run only on file types the selected field or recognition setup supports.
### Run remaining or re-run all
* **Run remaining** processes eligible assets that have not been answered yet. An earlier AI result of no value or no match counts as an answer, even if the cell looks empty.
* **Re-run all** revisits earlier results in the chosen scope. Review the confirmation's scope and counts before starting it.
* **Replace manual values** is off by default for bulk re-runs. Leave it off to preserve values or recognition answers a person entered.
Choosing **Run** on a single cell explicitly re-runs that value and can replace a manual answer. This differs from the default protection on a bulk **Re-run all**.
## Metadata on share links
Share-link settings control whether recipients can view metadata and, separately, edit it.
Enable metadata viewing, then use **Choose visible attributes...** to select the fields recipients can see. Drag selected attributes to change their order. These fields also seed the metadata shown on share-view asset cards.
Keep internal information such as budgets, rights notes, or production details out of the visible list when recipients do not need it. Showing a field in your internal library view does not make it visible on a share link, and allowing metadata viewing does not grant editing permission.
Configure access, metadata visibility, and other controls for share links.
## Metadata in archived projects
Archiving a project moves it to the **Archived** section on the workspace homepage. Its assets retain their metadata, previews, search access, and sharing behavior.
You can archive projects in either Active Storage or Cold Storage. Archiving leaves their storage tier unchanged, and you can continue to find and update metadata according to your permissions. See [archiving a project](/docs/concepts/core-concepts#archiving-a-project).
## Metadata best practices
* Keep select options short and mutually exclusive when possible.
* Use single select for one status and multi select for several applicable categories.
* Review automated values during ingest or QA.
* Avoid exposing internal fields on external share links.
* Keep metadata attached to the source asset instead of repeating the same information in filenames.
## Related guides
Create fields and update metadata across multiple assets.
Choose which metadata appears in the project library.
Control access and metadata visibility on share links.
This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.