Organize and keep content
Cloud drives: connect, import, and back up
Import drive images into personal materials or back up BANANAI images. These are explicit copies, not automatic synchronization or complete canvas archives.
Open featureManage external connectionsOn this page
Connect a cloud drive
When the site disables a drive, its entry points are hidden and new connections, browsing, imports, and backups are blocked. Existing connections are then disconnected; active transfers finish first. Site images and drive backups are not deleted. Users must authorize again when the drive is re-enabled.
- In Settings โ External connections, choose an enabled and configured drive. Check account, permissions, and accessible directories in its authorization window and confirm yourself.
- Check connected status and account, then browse Cloud drive in Asset Library. Another account does not open the old account's files, and access need not cover the entire drive.
- If popups are blocked, permit the site window and retry. Reconnect in Settings after expiry or access errors. Repeated clicks do not enable unsupported platforms.
Browse, import, download, or back up?
Separate actions let you choose when, where, and which storage to use. Authorization is not full import, and viewing a remote thumbnail is not backup.
Import and backup are not continuous two-way sync or a canvas archive containing nodes, connections, prompts, tags, and permissions. Images alone cannot restore the graph.
Backup does not change site ownership or folders or free site storage. Drive, site, and device files are separate copies; later edits or deletion do not automatically synchronize them.
Baidu Netdisk currently supports only the BANANAI application folder /apps/BANANAI, found under My application data / BANANAI in the drive. Browsing, importing, and backing up are limited to this folder and its subfolders. To import images from elsewhere, first place them in this folder through Baidu Netdisk. Approval to publish the app does not grant full-drive access.
| Goal | Action | Where the copy is saved |
|---|---|---|
| See drive contents | Browse directories and thumbnails. | Browsing does not save remote originals into materials. |
| Use drive images in creation | Drive โ BANANAI: add selected images to personal materials. | Independent saved files using BANANAI storage; drive originals stay. |
| Keep BANANAI originals on a device | Download in details or batch tools. | Device downloads, as separate image files rather than ZIP. |
| Save BANANAI images to a drive | BANANAI โ Drive: Back up to cloud drive. | The chosen drive directory; site images stay. |
Import selected images into a personal folder
The site copy does not disappear when a drive original moves, is deleted, or disconnects, nor update automatically with remote changes.
Estimated size uses remote metadata; each file still needs validity, single-file-size, and real-capacity checks. A readable thumbnail does not guarantee an available valid original.
Drive import does not offer local-upload compression confirmation or silently compress oversized files. Adjust oversized originals first. Follow live upload rules, not screenshot capacity numbers.
- Choose a connected drive and open its image folder. Load more as needed. Current-folder search filters loaded entries, not the whole drive recursively.
- Select images and choose Add selected items. Directories open navigation; ordinary non-image files cannot be imported here. Recheck selection after drive/folder changes. Google Drive's picker joins the same flow.
- Check count, estimated bytes, and available site capacity, then choose a personal-material folder, not a copied remote directory tree. Resolve folder-load failures first.
- Choose Start adding and wait for final results. Selecting, opening a dialog, or Preparing is not saved content. Estimated bytes are not necessarily additional storage: repeated imports may reuse site images. Per-item results determine whether capacity is sufficient.
- Check successful items and full images in the destination materials folder. Use these saved files through normal reference selection later.
Repeated imports from the same drive
Repeated imports of the same file from the same drive account first check its content checksum or version. If unchanged and the site image is complete, it is reused without downloading or image processing. Changed or unverifiable content is downloaded: unused materials are updated; images referenced by a canvas, conversation, inspiration, or other content are preserved with their material, and a new material is created. The selected destination still applies. Reusing an in-use image in another folder creates a material reference without moving the old material. Later imports target the latest material. Different files, accounts, or drives remain separate even with matching names. A matching filename, file ID, size, or modification time alone does not establish identical content.
Check target, version, and capacity before backup
| Option | Use | Remember |
|---|---|---|
| Original | Delivery, editing, and complete saved image files. | Larger files; cannot recover pre-upload compressed device originals. |
| Preview | Smaller files for quick viewing. | Not an equal-quality substitute or sole full backup. |
| Source subfolders on | Group In use by session and materials by folder. | Check source/target hints; unassigned items use an unassigned directory. Not a complete canvas or folder-tree export. |
| Source subfolders off | Write directly to the confirmed directory. | Organize later yourself. |
- Select image-bearing items in In use or Personal materials and choose Back up to cloud drive. Text-only items are not exported; an all-text selection disables backup.
- Check selected count, connected drive, and account. Connect in Settings if necessary, then return and recheck.
- Open Choose directory, navigate, and confirm Use this directory. Opening a folder alone is not confirmation. Google Drive uses its folder picker. Cancel keeps the old target; switching drive requires rechecking.
- Choose whether to create source-based subfolders and whether to use Original or Preview. Check target, estimate, and drive capacity.
- Start backup and wait for item results. Check actual files, versions, and dimensions in the drive. Closing the window or task acceptance is not successful writing.
Google Drive least privilege
Use drive.file to access only files selected through Google's official picker or created by BANANAI. Selecting a folder does not grant access to every file inside.
The following Google Drive replacement and folder-reuse rules apply only to items authorized for BANANAI. Unshared items with matching names are not read or overwritten; new files or folders with the same name may be created.
Names, repeated backups, and separate batches
Filenames use image identity, version, and real extension, such as an identity followed by -original.png or -preview.jpg. They are not all named original.png and do not necessarily use material titles or PNG.
Source names label subdirectories, with unsuitable path characters adjusted. Distinct names may sanitize to the same folder; verify paths.
All four drives (Dropbox, OneDrive, Google Drive, and Baidu Netdisk) use same-name replacement for backups. Backing up the same image and version to the same directory updates or overwrites the same-name destination file without comparing contents or asking for confirmation. No dated history copy is created automatically; changes you made to that backup in the drive will also be overwritten. Choose a different directory to keep the old file or separate batches.
With source-based subfolders enabled, all four drives create missing subfolders under the selected parent and reuse existing same-name subfolders. Repeated backups do not clear or delete the entire folder or create another numbered folder. Same-name image files are overwritten as described above; other files remain. This is BANANAI's backup rule, not whole-folder replacement.
Use different parent directories for separate batches. Changing a site title does not create a new filename, and renaming source folders or sessions does not move old drive files.
Repeated references in selection can produce fewer distinct files than selected count. Inspect actual content rather than compare folder count with success count alone.
Completion, partial failures, and retries
| Status | Meaning | Next step |
|---|---|---|
| Preparing/processing | Accepted or transferring, not all successful. | Wait rather than resubmit. |
| Processed count/bytes | Progress, not proof every file succeeded. | Inspect final per-item results. |
| All successful | All selected items were handled successfully; repeated imports may reuse images, update materials, or create new copies. | Check actual destination content and file versions. |
| Partial failure | Successful items stay; missing items are not automatically filled in. | Record failures, reselect only missing needed items, and start another task. |
| Drive full | Written files stay; unprocessed items are failures. | Free capacity or change target and check successes first. |
| Expired connection, missing/invalid/oversized source | Item conditions failed, not successful import/backup. | Reauthorize, locate, or adjust files rather than blindly resend everything. |
- One active drive transfer per user at a time. Asset operations pause during transfer. Refreshing and returning resumes status display, not task cancellation.
- There is no active-task Cancel or one-click failed-item retry. Done acknowledges results; reselect failures manually.
- Partial success is not rolled back: updated or newly created materials and written drive files stay. Do not delete successful files to retry, and do not treat closing the page or disconnecting as undo.
- Estimated or processed bytes need not equal original size, especially for reuse and failures. Judge by final status and real files.
