> ## Documentation Index
> Fetch the complete documentation index at: https://kino.ai/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Search & Logging

> Kino watches your footage, writes logs, and makes every moment searchable

Without an index, a big library is write-only: footage goes in and never comes out. Kino watches everything during ingest and writes logs you can search — each clip gets a title, description, transcript, detected people, and highlight markers. Chat and connected agents read the same logs.

![Kino search interface showing massive crowds search](https://mintcdn.com/kinoai/ewcknApqdW8_9YSq/images/screenshots/massive-crowds-search.png?fit=max\&auto=format\&n=ewcknApqdW8_9YSq\&q=85\&s=e9eb199bccbb3621c42dd94ad3733cc9)

Natural language search for 'massive crowds' finds relevant scenes

## Search modes

**Transcript**

Spoken words become searchable after transcription. Find dialogue without making transcripts by hand.

**Visual**

Kino recognizes objects, scenes, actions, and visual elements in your footage. Search for "beach sunset", "two people walking", or "drone shot of city".

Visual search can find shots that filenames and folder names miss. Chapter highlights from logging surface in visual results too.

**Files**

Matches filenames, folder paths, and AI-generated titles. This tab appears when your query looks like a file — useful when you know a clip is called `A012_C004` but not what's in it.

Press `Shift+Tab` to cycle between search modes.

> [!TIP]
>
> Combine search methods:
>
> * Find all clips where "@Sarah mentions quarterly results near whiteboard"
> * Locate "exterior shots of building with people entering during daytime"
> * Filter for "closeup shots with dramatic lighting and minimal dialogue"

## Search by people

Kino detects people in your footage. Use people filters to find shots containing `@Ted`, or what was said while `@Amy` was on screen.

![Searching for Brian in Kino interface](https://mintcdn.com/kinoai/ewcknApqdW8_9YSq/images/screenshots/brian-ethernet-search.png?fit=max\&auto=format\&n=ewcknApqdW8_9YSq\&q=85\&s=caa03913cb293e3ebf337b786ca4d514)

@'ing Brian returns all footage containing him, plus 'ethernet'

### Teach Kino about a new person

**1. Navigate to People page**

Go to the `People` page and click **Create Person**.

**2. Add a photo**

Add a clear, high-quality picture from your clipboard with `Cmd + V`, or select an image from your computer.

**3. Submit and process**

Submit the photo. Kino will populate the page with footage containing that person.

Use well-lit photos for better detection. Once configured, Kino's logs incorporate the names of people in your library, and you can browse the whole library by person.

## Logging

Video logging — analyzing footage and writing detailed descriptions of its content — is traditionally done manually by professional loggers who watch every second of footage, often multiple times. Kino does that work during ingest, so editors can skim a library without opening every file.

![AI video logging interface with automatically generated descriptions](https://mintcdn.com/kinoai/ewcknApqdW8_9YSq/images/screenshots/logging-brief-1.png?fit=max\&auto=format\&n=ewcknApqdW8_9YSq\&q=85\&s=acd8aaf91b4bd38a630a3b8823a7a6e8)

**Automatic titles**

Whether it's raw footage, clips from YouTube, or a mix of both, every clip receives a readable title.

Filenames like `IMG_3923.MOV` can become `Brendan Backstage Interview Before Keynote`, which makes grid browsing much faster.

![List of video clips with AI-generated titles](https://mintcdn.com/kinoai/ewcknApqdW8_9YSq/images/screenshots/titles-showcase-2.png?fit=max\&auto=format\&n=ewcknApqdW8_9YSq\&q=85\&s=f978274fc16647a1b9865eb7f6fd8472)

AI transforms raw filenames like 'IMG\_3923.MOV' into descriptive titles

**Automatic descriptions**

Kino descriptions can include:

* Visual elements and actions happening on screen
* Camera movement and shot type
* Dialogue context and quotes
* Setting and environment details
* Emotional tone and atmosphere of the scene

Use descriptions when a transcript alone is not enough to find the shot.

![List of video clips with AI-generated descriptions](https://mintcdn.com/kinoai/ewcknApqdW8_9YSq/images/screenshots/descriptions-showcase-1.png?fit=max\&auto=format\&n=ewcknApqdW8_9YSq\&q=85\&s=d96d5822db813bd411850e95520b91f2)

Descriptions capture visual details alongside transcript context

**Person identification**

Once you teach Kino the people in your library, logs incorporate their names, which makes logs easier to read and search.

![Video timeline with person identification](https://mintcdn.com/kinoai/ewcknApqdW8_9YSq/images/screenshots/chris-log.png?fit=max\&auto=format\&n=ewcknApqdW8_9YSq\&q=85\&s=1eff31ffb87b3476cd5d392e47a82b9c)

Person identification makes logs more specific and useful

**Highlight generation**

Kino can mark notable moments in footage:

* Creates a chapter-like experience similar to YouTube markers
* Makes footage review faster
* Adds another search target
* Provides edit points for rough cuts

![Video timeline with AI-generated highlight markers](https://mintcdn.com/kinoai/ewcknApqdW8_9YSq/images/screenshots/montage-logs.png?fit=max\&auto=format\&n=ewcknApqdW8_9YSq\&q=85\&s=73874b0e334ee9f45f5718b8cb50f888)

Auto-generated highlights create chapter-like markers for easy navigation

## Example use cases

**Documentary organization**

For verité documentary footage with multiple characters:

1. Ingest 10+ hours of footage from your shoot
2. Kino identifies every person and describes every scene
3. Review a searchable database of logged moments
4. Face detection automatically groups scenes by character
5. Natural language descriptions allow you to search conceptually ("emotional conversation" or "urban landscape")

A 10-hour shoot becomes easier to scan without a manual first pass.

**Event coverage**

For multi-camera event coverage:

1. Ingest 20+ hours of footage from your shoot, across multiple cameras
2. Descriptions note what each camera captured, including audience reactions
3. Highlight generation marks the strongest moments
4. Person detection can group appearances across cameras

Editors can jump straight to the strongest angles and reactions.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.