Uploading assets
Click Upload in the asset sidebar to add files from your device. You can upload multiple files at once — up to 50 files per batch, with a maximum of 2 GB per file. Supported file types:| Category | Formats |
|---|---|
| Images | JPEG, PNG, WebP, GIF, AVIF, HEIC |
| Videos | MP4, WebM, QuickTime |
| Audio | MP3, WAV, OGG, M4A |
- Deduplicates files so the same image is never stored twice
- Extracts EXIF metadata from photos (camera, lens, GPS location, exposure settings)
- Analyzes visual content for similarity grouping
- Clusters similar images and assigns AI-generated tags
Tags
Tags replace traditional folders with a flexible, multi-label system. Each asset can have multiple tags, making it easy to find files that belong to more than one category.AI tags
When you upload images, TalkCut clusters visually similar files and generates descriptive tags automatically. These AI tags appear alongside your manual tags and can be used in filters.User tags
Create your own tags to organize assets the way you think about them. Select one or more assets and assign tags from the sidebar.Tag groups
Organize related tags into groups for cleaner filtering. For example, group tags like “outdoor”, “studio”, and “street” under a “Location” group.Filtering assets
The filter bar lets you narrow down your library using multiple dimensions at once:- Tags — Filter by AI-generated or user-created tags, with nested tag group menus
- Type — Show only images, videos, or audio files
- Status — Filter by processing status (uploaded, processing, classified)
- Date — Filter by upload date or original capture date
EXIF filters
For photos with embedded metadata, additional filter dimensions are available:- Camera brand and model — Find all photos from a specific camera
- Lens — Filter by lens model
- Country and city — Browse by shooting location (reverse-geocoded from GPS coordinates)
Using assets on the canvas
Open the asset panel from the canvas toolbar by clicking Assets. Browse or search your library, then drag assets onto the canvas to use them as:- Reference images for AI generation — including reference video for visual consistency
- Style guides that the AI agent can analyze and match
Auto Classify
TalkCut automatically analyzes and groups your assets by visual similarity when you upload files. You can also trigger classification manually for more control.Automatic classification
Assets are processed automatically after upload. The system analyzes visual content, groups similar assets into clusters, and generates AI tags for each group.Manual classification
Click Auto Classify in the asset sidebar to open the Process & Group dialog:- Grouping granularity — Adjust the slider from Very Fine (many small, tightly matched groups) to Very Coarse (fewer, broader groups). Default: Balanced.
- Reset all groups — Toggle this to delete existing groups and regroup from scratch. When off, only untagged assets are grouped.
Reviewing classified tags
After classification, AI-generated tags appear in the sidebar under the AI section, marked with a sparkles icon. These tags have a purple indicator to distinguish them from your custom tags. To keep an AI tag permanently, click the tag menu and select Claim. Claimed tags move to your custom tag groups and behave like any user-created tag.Auto Name
Click Auto Name in the asset sidebar to generate descriptive names for your asset groups. You can provide custom instructions — for example, “Use Chinese” or “Prefer film terminology” — to control the naming style.Auto Name costs 5 credits per group. Auto Classify costs vary by the number of assets processed.
Managing assets
Select one or more assets in the library to:- Add or remove tags in bulk
- Delete assets from your workspace
- View metadata including EXIF details, dimensions, and file size