Uploaded April 2026 | Updated September 2026, 2 weeks ago
A-Eye uses local AI vision models to analyse your photos and automatically rename them based on what it actually sees. It also writes descriptions, generates tags, pulls out location data, and makes your whole library searchable - all running privately on your own hardware through Ollama.
No cloud services, no subscriptions.
It works with your existing folder structure, your photos stay exactly where they are, it just gives the files inside proper descriptive names. And if you don't want anything renamed at all, you can run it in catalogue mode where it analyses everything and builds a searchable database without touching a single file. That means you can layer it on top of Immich, Lightroom, or whatever you're already using.
In this video I walk through the full setup on Unraid — installing Ollama, configuring A-Eye, running your first scan, reviewing AI-generated filenames, using context to teach the AI things like your pet's name, LLM-powered search, the temporary workspace, and more.
🔗 A-Eye on GitHub: github.com/SpaceinvaderOne/a-eye
⏱️ Timestamps:
0:00 - Introduction
1:44 - What hardware do you need? GPU vs CPU
2:24 - Installing Ollama on Unraid
3:15 - Installing A-Eye from Community Apps
3:58 - The Onboarding Wizard
4:52 - CPU vs GPU model recommendations
6:04 - Choosing your processing mode
8:28 - Watch mode and security setup
10:09 - The Dashboard explained
11:48 - Upload to Library and Temporary Workspace
12:59 - First scan - watching it process
14:03 - Reviewing AI-generated filenames
15:02 - All Images view and confidence scores
15:42 - Reverting a rename
17:00 - History tab and export
17:52 - Reprocessing with context (teaching AI your dog's name)
20:07 - Search — keyword vs LLM-powered
22:25 - Photo Showcase and full-screen mosaic
26:02 - Temporary Workspace walkthrough
28:18 - Settings deep dive
29:01 - Catalogue mode
30:27 - Read-only mode for maximum safety
31:32 - Wrap up
What will you build with Unraid?
Get Started with Unraid in 15 minutes or less:
unraid.net/getting-started
A-Eye uses local AI vision models to analyse your photos and automatically rename them based on what it actually sees. It also writes descriptions, generates tags, pulls out location data, and makes your whole library searchable - all running privately on your own hardware through Ollama.
No cloud services, no subscriptions.
It works with your existing folder structure, your photos stay exactly where they are, it just gives the files inside proper descriptive names. And if you don't want anything renamed at all, you can run it in catalogue mode where it analyses everything and builds a searchable database without touching a single file. That means you can layer it on top of Immich, Lightroom, or whatever you're already using.
In this video I walk through the full setup on Unraid — installing Ollama, configuring A-Eye, running your first scan, reviewing AI-generated filenames, using context to teach the AI things like your pet's name, LLM-powered search, the temporary workspace, and more.
🔗 A-Eye on GitHub: github.com/SpaceinvaderOne/a-eye
⏱️ Timestamps:
0:00 - Introduction
1:44 - What hardware do you need? GPU vs CPU
2:24 - Installing Ollama on Unraid
3:15 - Installing A-Eye from Community Apps
3:58 - The Onboarding Wizard
4:52 - CPU vs GPU model recommendations
6:04 - Choosing your processing mode
8:28 - Watch mode and security setup
10:09 - The Dashboard explained
11:48 - Upload to Library and Temporary Workspace
12:59 - First scan - watching it process
14:03 - Reviewing AI-generated filenames
15:02 - All Images view and confidence scores
15:42 - Reverting a rename
17:00 - History tab and export
17:52 - Reprocessing with context (teaching AI your dog's name)
20:07 - Search — keyword vs LLM-powered
22:25 - Photo Showcase and full-screen mosaic
26:02 - Temporary Workspace walkthrough
28:18 - Settings deep dive
29:01 - Catalogue mode
30:27 - Read-only mode for maximum safety
31:32 - Wrap up
What will you build with Unraid?
Get Started with Unraid in 15 minutes or less:
unraid.net/getting-started
