The tricky part about finding a Google Photos replacement isn’t storing photos or working within its ecosystem. Often it’s just finding an alternative that makes photos just as easy to search, organize, and share as Google does.
Immich already covers much of that, which made me wonder whether Noodle Gallery could improve on it.
Noodle Gallery is a self-hosted Immich fork with loads of extra features for managing and sharing a photo library. I installed it on one of my Docker VMs and uploaded just over 120 photos to see if the features made it a good alternative to Google Photos.
The familiarity of Immich made a good impression, but the extras have me well and truly hooked.
The photo library made a strong first impression
Search found specific cars, and my collection was easy to browse on both screens
I installed Noodle Gallery v5.7.1 through Docker Compose on a Proxmox virtual machine with two CPU cores, 8 GB of RAM, and a 50 GB disk. That was my test setup, and it had enough room and resources to deal with background processing tasks.
Once the photos were uploaded, I could browse them by date, open up individual images, and inspect the image EXIF data. I also gathered all the coastal photographs and popped them into an album called “Ocean.” I then installed the Noodle Gallery Android app and connected it to my server, and could view the collection from either screen.
All of this is normal behavior I expected from an image organization app, but Search was where I started having fun.
One of the best features of Google Photos is its image search. You can search for the most granular terms, and somehow it generates a collection of your own images that match them with scary accuracy. Immich did a good job of replicating this, but its limitations became obvious once you started using more specific search terms.
Naturally, this is one of the first things I tested with Noodle Gallery. Searching for a photo of a “cat” would have been an easy first test, but I wanted to know how specific I could get without manual labeling:
- “Seal point cat” brought up pictures of my cat and omitted cats with other coloration patterns.
- “Lada Niva” and “Suzuki Jimny” brought up images of those precise models of cars.
- “Snowy mountains” picked up the winter scenes specifically with mountains only.
- “Seagull” picked out all the seagulls and omitted pictures of parrots and pigeons.
I hadn’t tagged a single photo before I started. Being able to type what I remembered about a picture made the Noodle Gallery pretty close in terms of Google Photos’ search functionality.
Noodle does inherit this capability and control from Immich, so credit goes where credit is due. However, the UI itself was much snappier and faster than my previous experience with Immich running on the same resources and handling similar images.
It recognized my cat before he grew up
The kitten photo made pet recognition feel genuinely useful
It matched my cat Jarvis’ kitten photo and kept the other cats out. That one feature caught my attention more than any other in the app’s description. Jarvis the cat has since passed away, so seeing these pictures brought together different stages of his life and gave the collection of images that Noodle surfaced some personal weight.
I’d enabled its pet detection feature, which works like facial recognition, only for animals. I let the processing finish, looked under Explore, and named the resulting pet entry “Jarvis.” Opening it gave me a collection of his photos without ever needing to select each one individually and manually add it to an album.
Searching for “cat” would’ve found all the cats in my photos (and there are a lot of them!), but Noodle’s pet detection feature recognized this one particular cat across visibly different ages.
It matched my cat Jarvis’ kitten photo with his adult pictures and kept the other cats out.
Now, Jarvis was a very unique-looking cat, so I can’t promise Noodle’s pet detection will distinguish every pet equally well. Still, the correctly grouped photos showed why I and other pet owners would want the feature. It helped me revisit photos with personal meaning without having to manually sift through a messy image library.
Shared Spaces brought two libraries together
The other account’s sunsets appeared immediately alongside mine
Combining photos from different accounts is awkward when I only want to share part of each collection with other family members or friends. I wanted to see whether Noodle Gallery’s Shared Spaces could bring pictures together while still leaving the rest of each library separate.
This feature is handy for when you have multiple people taking photos at an event, say, a wedding, and you want to combine everyone’s photos into a single album.
I created a Space called “Sunset Photography,” added photos from my main account, and then gave a second test account Editor access. From that test account, I uploaded another selection of sunset photographs and added them to the Space.
From the main account’s perspective, the images appeared immediately. Better still, they slotted neatly into my existing photos ordered by date taken. I didn’t need to download the other account’s photos, nor upload another copy. The shared collection was already there, arranged into a single unified timeline.
Permissions gave that shared space some structure:
- The Owner manages the space and its membership.
- An Editor can contribute and manage photos in the collection.
- A Viewer can browse without the ability to change the photos.
I could also check the activity log to see which account had added which photos. That made it far easier to track changes, especially compared to doing it by memory, which left room for too many mistakes.
Immich already supports shared albums, so again, sharing isn’t a novelty that’s exclusive to Noodle. What did appeal to me, though, was the combination of a shared timeline, member roles, and trackable activity.
Even with just two test accounts, I could see myself persuading friends and family to join my Noodle server so we could collect everyone’s photos from shared trips and celebrations in one place.
Automatic sorting needed a little supervision
It found my screenshots, but photographs of screens fooled it too
Noodle’s search functionality was an instant hit for me, but automatic classification still needed a bit of the human touch to work properly. Gallery lets me describe a category in plain language, then uses OpenAI’s open-source CLIP (Contrastive Language-Image Pre-training) to tag photos that fit. This feature is unique to Noodle Gallery, and one of the main reasons I wanted to try it. It can also archive matches automatically, though I chose Tag only so I could at least inspect its decisions first.
My initial descriptions covered screenshots of phone screens, computer applications, and websites. By default, the similarity threshold is set to 0.28, and using this, the Gallery found only two Windows-related images and missed the screenshots I uploaded specifically for the test.
Lowering that threshold to 0.24 brought in all the intended screenshots. Unfortunately, the results also included photographs of monitors and my car’s stereo LCD screen.
|
Configuration tested |
Result in my collection |
|
Initial prompts, 0.28 similarity |
Two matches, missing the test screenshots |
|
Expanded prompts, 0.24 similarity |
All test screenshots, plus photographs of screens |
|
Expanded prompts, 0.25 similarity |
Most test screenshots were missed |
Increasing the threshold by 0.01 to 0.25 made the matching too strict for my collection. I stopped there and kept the broader results for a manual review. The classifier recognized screen-like content, but unfortunately, it couldn’t differentiate between a photo of a display and a screenshot.
Tagging still helped gather potential candidates, but automatic archiving is just too hard to justify enabling when ordinary photos could get swept up too.
Enable Tags under Account Settings → Features to browse generated tags. Start with Tag only, then inspect the results. If you’re happy with its selection, turn on automatic archiving.
More features are only good when they solve problems I can actually see
Immich has a much larger community and a longer track record. Those are two very good reasons to stick with it, especially if you already have an established photo library. Noodle Gallery builds on the good work Immich has done, but adding features doesn’t automatically make the fork worth switching to.
My test also had its limits. I only used a small collection of test photos, and I still needed to review the classification results. I also hadn’t moved my entire photo library of thousands of images across or tested whether I’d fully trust Noodle Gallery with years of backups.
In saying that, my old image management solution, Google Photos, isn’t a backup tool either. I would comfortably move my own images over to either Immich or Noodle, provided a secondary image backup plan was in place.
Still, the additions gave me really useful results. My cat’s kitten and adult photos all came together, the shared sunsets appeared immediately, and the classifier gathered my screenshots once I’d tweaked the settings. Every new feature really did something I could appreciate while browsing my own pictures.
Honestly, that’s more than enough for me to keep using Noodle Gallery and see how it holds up when transferring a truly massive photo collection. If add-on features like pet detection and prompt-based automatic classification don’t really have a use for you, then Immich may already be everything you need.
In my opinion, Open-Noodle has made the fork worth spending more time on, and in the meantime, created an absolutely viable alternative to Google Photos.