Add Media to a Repository

在应用中找到它

SidebarManage

The Manage page has two areas: an upload zone at the top and repository cards below. The upload zone ingests files you select in the browser into the current repository; the cards are used to scan files that are already on the server, create repositories, and run maintenance operations.

Pick the right import method first

There is one deciding question: can the machine running Lumilio read the files directly?

Where the files are now	Method	Where they end up	Best for
Current computer, phone, or tablet, selectable in the browser	Upload	inbox/, archived by the repository policy	Small batches, importing across devices
A location authorized in Desktop, or a repository directory mounted in Server	Scan	stays at its current free-zone path	Large batches, when you want to manage the folders yourself
A connected cloud-service credential	Cloud import	inbox/, archived by the repository policy	Importing from the cloud; iCloud is still experimental

Selecting files in the browser only sends their contents to Lumilio; it does not authorize the server to read arbitrary directories on the browser device. For Storage Locations, repositories, and Desktop/Server differences, read Storage Locations and Repositories first.

Before you start: confirm the upload target

1. Open the Manage page.
2. Choose the repository in the Upload target dropdown above the upload zone.
3. Confirm the target again before clicking Upload; the batch goes to the repository selected here.

If the dropdown is empty, or a repository shows “offline” or “needs attention”, do not import yet. Desktop users check the Control Panel Storage Locations; Server users check Compose mounts, directory permissions, and disk state. An offline repository refuses writes; it does not quietly create a same-named directory elsewhere.

Method 1: upload files from the browser

Steps

1. Click Supported formats next to the page title to confirm the extensions are supported.
2. Drag files onto the upload zone or click Add files to select them. Images (including camera RAW), videos, and audio are supported.
3. Review the file list and the upload target, then click Upload (count).
4. Keep the page open until the queue items become Completed, Duplicate, or Failed. Large files automatically use resumable chunked upload; no manual splitting is needed.

The browser filters by extension first; the server validates again. Unsupported files do not enter the upload queue — remove or convert them and retry.

Reading upload results

• Completed: the file passed server validation and finished ingestion.
• Processing: the original file was received; metadata, thumbnails, video/audio web versions, and optional analysis are still being generated in the background.
• Duplicate: the content fingerprint matches existing media in the target repository; Lumilio skipped the second copy and does not use extra storage.
• Failed: ingestion did not finish. The failed file stays in the upload queue; fix the network, permission, or disk issue and click upload again.

After a successful upload, the original file first goes through the repository's staging pipeline, then into inbox/ according to the repository policy. Repository creation uses the Server's safe defaults instead of asking for layout details in the primary task. After the upload is complete, inbox/ is an ordinary user-visible landing area: you may move or rename its originals inside the same repository, then rescan. Only .lumilio/ is application-private.

Validate with a small batch first

When importing from a new camera, a phone export folder, or an external drive, select a few files first. Confirm dates, orientation, duplicate handling, and thumbnails before submitting the full set.

Method 2: scan files already on the host

Scanning does not copy files and does not make the browser choose server paths; it registers files that already exist in the target repository.

Preparation

• Put the files anywhere in the user-visible media tree under the repository root, including inbox/, but never under .lumilio/.
• Desktop: authorize the Storage Location in the Desktop Control Panel first.
• Server: complete the bind mount per repository mounting and creation; the in-container path must be readable.

Do not put files into the private directory

.lumilio/ holds staging files, derived resources, and other application-private state. The scanner skips it, and manual changes can make the catalog and files disagree. inbox/ is not private and is included in scans.

Running a scan

1. Put media into the repository and wait until copying finishes.
2. Open the ⋯ menu on the repository card and choose Rescan.
3. To scan every repository, click Scan all repositories above the cards.
4. Wait for the operation to finish, then check the repository or the Server Monitor page for new media and background tasks.

The scanner only registers supported media extensions. For already-registered files, including files in inbox/, it tries to recognize “the same file moved to a new path” and updates the record instead of creating a second media record. A missing path is confirmed only after a complete authoritative scan; partial reads and ambiguous matches do not prove deletion.

Method 3: cloud import

Cloud import needs a connected cloud credential. First complete the login and any required verification on the Cloud import tab of Settings. Repository creation and cloud authorization are separate tasks: create or open the destination repository first, then choose Import from cloud in its card menu and select the connected credential.

Cloud-imported files enter inbox/ like uploads and follow the repository's storage policy. The card shows the last import status, imported count, and failed count; while the task is “queued” or “running”, the trigger button is disabled.

iCloud is still experimental

iCloud import depends on Apple's unofficial network service behavior, may require extra verification, and can stop working at any time. Use it with care.

Creating a repository (administrator)

The Create repository button in the card area is for administrators. Its primary path asks for:

1. Storage Location: Desktop picks from locations registered in the Control Panel; Server uses the mounted Storage Location from Compose.
2. Repository name: the user-facing display name. It is independent of the folder name.
3. Storage folder: one portable direct-child directory name. Only letters, digits, full-width text, half-width spaces, -, and _ are allowed, 1–80 characters, no leading/trailing spaces; case is preserved.

Creation validates the name, direct-child topology, directory state, and write permissions and fails fast; an incomplete repository is never registered. It applies the Server's safe layout and filename-conflict defaults. If the target already contains a valid .lumiliorepo, creation stops and directs you to the explicit Open Existing Repository task; it never attaches that identity implicitly. An ordinary non-empty unmarked directory cannot be initialized as a new repository.

Desktop can also start Add Storage Location or Open Existing Repository from Web. The request appears under Desktop Settings → Storage → Requests from Web; a person at that computer must review it and choose the directory locally. Web never submits an arbitrary Desktop path. Standalone and Docker Server instead list classified direct-child candidates from the configured default Storage Location.

For Docker, create a new folder by entering a directory name, or mount an existing empty host directory at that exact direct-child path before creation. On Linux, an existing empty target must be verified as an actual mount point. Follow the long-syntax Compose example in Storage Locations and Repositories.

Maintenance actions on a repository card

Open the ⋯ menu on a card for the current repository's actions:

• Rescan: scan this repository's user-visible media tree, including inbox/; use after copying, moving, or renaming a batch of files.
• Detect stacks: run automatic stack detection; review results in Browse, filter & batch.
• Scan duplicates: run duplicate analysis; results are handled in Duplicates, likes & trash.
• Rebuild location clusters: recompute media location clusters; never moves originals.
• Import from cloud: only for repositories bound to a cloud credential.
• Remove from Lumilio: unregisters a non-primary, idle repository after showing its impact. .lumiliorepo, .lumilio/, and every media file remain on disk; this is not physical deletion.

The counts and “offline / needs attention” marks on the cards come from the server's current state. When a Storage Location is unavailable, write and maintenance buttons are disabled; restore the mount or permissions first, then retry.

What to check after importing

Ingestion and derived processing are two stages: the original file may already be in the repository while thumbnails, web versions, people, or semantic results are still queued. When in doubt:

• Queue still running and the completed count is growing: wait; do not rescan or re-upload;
• Media shows as duplicate: confirm the target repository; usually no second import is needed;
• Failures keep growing or nothing progresses: confirm the disk is online, writable, and has space, then check task errors on the Server Monitor page;
• Scan found 0: confirm files are not under .lumilio/, the extension is supported, and the files really are under the current repository's mounted path.