Cutsio Blog

Search Your Entire Film Library by Visual Content: Beyond Filename-Only Search

PIX only searches footage by filename. Cutsio's Visual Intelligence indexes every frame by visual content — search for objects, scenes, actions, and text across your entire film library.

How do you search a film or television library when you cannot remember the filename?

Cutsio's Visual Intelligence indexes every frame of uploaded footage by its visual content — objects, scenes, actions, text, and visual characteristics — making any frame searchable by description, not just by filename. PIX offers filename and folder search only. If you cannot remember the exact clip name on PIX, you must scrub through folders manually to find what you need. This is the capability that makes Cutsio a true PIX alternative for film and TV.

The limitation of filename-only search is a constant friction point in post production. Assistant editors, VFX supervisors, and post coordinators spend hours locating specific shots across thousands of clips. The work is manual, repetitive, and prone to error. A clip that was filed under "B-Roll_Day3_V2" is lost unless someone remembers that exact naming convention.

Visual Intelligence solves this by analyzing what the camera actually captured. The search index is created from the visual content of every frame, not from the text someone typed into a filename field. Combined with native raw ingestion and per-minute pricing, Cutsio addresses the three biggest gaps in PIX's platform. For an even deeper look, read the frame-accurate visual search deep dive.

What types of search queries can Visual Intelligence handle?

Visual Intelligence supports search by objects, scenes, actions, visual characteristics, and text visible in the frame. The following examples show the range of queries that return results.

Object search. Find clips containing specific items: "car," "actor wearing red jacket," "table with coffee cup," "tripod in shot."

Scene search. Find clips set in specific environments: "forest," "interior office," "city street at night," "beach during sunset."

Action search. Find clips with specific movements or activities: "actor running," "car driving frame left," "hand reaching for object," "crowd cheering."

Visual characteristic search. Find clips with specific visual properties: "warm lighting," "blue color grade," "shallow depth of field," "aerial establishing shot."

Text search (on-screen). Find clips with visible text in the frame: "sign that says exit," "newspaper headline," "storefront logo."

Composition search. Find clips by framing and camera movement: "close-up," "wide shot," "tracking shot," "handheld camera."

| Query Type | Example | Transcript-Based Search | Visual Intelligence |

| :--- | :--- | :--- | :--- |

| Object | "car" | No (unless mentioned in audio) | Yes |

| Scene | "forest" | No | Yes |

| Action | "actor running" | No | Yes |

| Spoken word | "roll camera" | Yes | Yes |

| On-screen text | "exit sign" | No | Yes |

| Visual style | "warm sunset lighting" | No | Yes |

How does this work for MOS footage with no audio?

MOS footage — action sequences, B-roll, establishing shots, visual effects plates, aerials, underwater footage — has no scratch audio to transcribe. On PIX and other transcript-only platforms, MOS footage is effectively invisible to search. There is no text index to query.

Visual Intelligence indexes MOS footage entirely by visual content. A search for "find the wide shot of the car driving through the forest at sunset" returns results from MOS clips because the visual index captures the car, the forest setting, the wide composition, and the warm sunset lighting — all from the pixel data alone.

For productions where MOS footage makes up 30-50% of the total capture — narrative features, commercials, music videos — this is the difference between being able to search your entire library and only being able to search half of it.

playback-id="IRBqKFllfQTZRgUpvF00DnjqMROLtyclqpWYRLQez6KQ"

title="Cutsio Visual Intelligence — search video by what the camera saw"

poster="https://image.mux.com/IRBqKFllfQTZRgUpvF00DnjqMROLtyclqpWYRLQez6KQ/thumbnail.jpg">

How does Agentic Chat handle natural language queries across the library?

Agentic Chat in Cutsio is a conversational AI interface that understands natural language queries about your footage and returns frame-exact results from the Visual Intelligence index.

Instead of building complex search queries, team members ask questions in plain language:

• "Show me all the master shots from Day 3 where the lighting is warm."
• "Find the close-up where the lead actress says her name."
• "Which takes have the boom mic in the top of frame?"
• "Are there any matching close-ups for this wide shot?"
• "Show me everything from Scene 24 that has a car in it."

Agentic Chat searches both the visual index (for objects, scenes, actions) and the transcript index (for spoken content) simultaneously, returning results from either or both sources in a single query.

For assistant editors and post coordinators who manage large libraries across multiple shoot days, Agentic Chat eliminates the need to remember exact filenames, folder locations, or search syntax. They ask the question in the same words they would use to ask a colleague, and Agentic Chat returns the answer.

How does Visual Intelligence compare to PIX's search capabilities?

| Search Capability | PIX | Cutsio |

| :--- | :--- | :--- |

| Filename search | Yes | Yes |

| Folder navigation | Yes | Yes |

| Transcript search | No | Yes |

| Visual search (objects, scenes, actions) | No | Yes |

| MOS footage search (no audio) | No | Yes |

| Natural language queries (Agentic Chat) | No | Yes |

| Search across mixed camera formats | No | Yes |

| Frame-exact results | No | Yes |

| Search by visual characteristics (lighting, color) | No | Yes |

PIX organizes footage the same way a hard drive does — by filename and folder. If your production generates 500 clips per shoot day across 30 shoot days, you have 15,000 filenames to remember or navigate through. PIX offers no mechanism to find a clip by what it actually shows.

Cutsio's Visual Intelligence creates a searchable index of the actual content. The same 15,000 clips are searchable by any object, scene, or action visible in any frame.

How do Collections complement visual search for organizing footage?

Collections in Cutsio are visual hubs that group clips by project, scene, shoot day, or any custom criteria. They combine with Visual Intelligence to give post teams two ways to find footage: browse visually through Collections or search directly with Visual Intelligence.

A typical narrative feature might have Collections for:

• By shoot day: "Day 1," "Day 2," etc.
• By scene: "Scene 24 — Kitchen," "Scene 25 — Car Interior"
• By camera: "A-Cam Alexa 35," "B-Cam RED V-RAPTOR"
• By selects: "VFX Plates," "Green Screen Shots," "MOS B-Roll"

Each Collection displays video thumbnails from the review stream. Team members browse the Collection visually, then refine their search within the Collection using Visual Intelligence queries.

FAQ

Does Visual Intelligence work on footage with no audio at all?

Yes. Visual Intelligence indexes the visual content of every frame independently of audio. MOS footage, silent clips, and clips with unusable scratch audio are all fully searchable. The visual index does not depend on audio quality or content.

Can I search for specific dialogue or spoken words in Cutsio?

Yes. Cutsio also indexes audio transcripts. You can search for spoken content alongside visual content in the same query. Agentic Chat searches both indexes simultaneously.

How accurate is Visual Intelligence on fast-moving action footage?

Visual Intelligence is most accurate on clearly visible objects, scenes, and actions. Fast motion, extreme close-ups, and heavily obscured subjects may produce less precise results. Accuracy improves with higher resolution source footage and well-lit scenes.

Does PIX offer any form of visual search?

No. PIX does not offer visual search, transcript search, or any form of content-level search. All footage on PIX is searchable only by filename and folder structure.

How does Visual Intelligence handle mixed camera formats in the same library?

Visual Intelligence indexes every frame of every clip regardless of camera format. ARRIRAW, RED R3D, and Blackmagic RAW clips in the same library are all indexed uniformly. A search for "car" returns matching clips from all formats without the user specifying which camera captured them.