Desmond Wong · Transcript · · Updated
This AI Turned 10 Years of Photos & Videos Into a Searchable Library | MUZIM by OpenSoul – Transcript
Video by Desmond Wong · Watch the original on YouTube
Summary
AI-generated from the transcript. Verify details against the source video.
- Museum is a local-first AI file agent that indexes selected folders and drives across photos, videos, audio, and documents so users can search by meaning rather than file names.
- The software processes files locally without uploading data to the cloud or moving them, though users can also connect cloud services such as iCloud and Google Drive.
- Museum includes AI agents that take action on libraries by organizing footage, returning highlights, and summarizing activity, alongside bringing your own key options for models like Claude or GPT.
Use this video with Pro
Reading, copying and transcript downloads are free. Create saved AI results for this video with a signed-in Pro account ($9/month).
Transcript
Search ignores accents. All transcript text stays visible.
All right, so this is an AI tool that you are actually going to use as a content creator, a photographer, as a videographer. This thing will help you find just about anything in your massive archive of files. This is Museum. Let's get right into it.
>> >> Hey folks, welcome back to the channel. My name is Desmond and we are talking about a brand new AI tool that actually think you're going to use. I've been using it. Before we get started, this video is sponsored by Open Soul, the creators of Museum, the software we'll be talking about. Now, if you've seen this channel before, you know I travel a ton. I do a lot of photography, videography, brand deals. I'm creating terabytes and terabytes of data, different files all over the place. Now,
I just got back from a two-week trip to Hawaii. a family vacation, some content work, photos of the kids, hiking, beaches, sightseeing. I also brought my Nikon ZR for cinematic shots, my DJI Osmo Pocket 4K for vlog work, my iPhone, >> >> and I recorded a lot of audio just on the DJI mics because sometimes it's just a little bit easier. After Hawaii, I then went to Shenzhen for a brand trip.
So, I went to China with the same kit, more photos, more videos, different cities, but ultimately the same workflow. In between these two trips, I've got almost two terabytes of just overall raw footage sitting on my drives right now. Plus, thousands of photos, dozens of wave files just from the last month, and that is ultimately just a small part of the story because I also
have a 16 terabyte backup drive of everything from all my trips from Athens, Paris, Tokyo, London, and just decades worth of family photos. And I'm just going to stop right there. Does this sound like you?
Do Do have multiple drives that you've been backing up for years? Uh just like me, the truth of being a creator is I've got a real solid archive, and I've got a lot of storage. The problem though is finding everything. My folder organization, I will be very honest with you, is not very, very good. I'm just dumping things into folders to get them off of my different SD cards, and you know, there there's just so much.
And I think this is a problem that a lot of creators or just photographers, videographers, people who just like to create, take photos, these are problems that we all run into, and this is ultimately the challenge that Museum has helped me solve. So, Museum is a local-first AI file agent that indexes the folders and drives I choose across supported photos, videos,
audio, and documents. So, I can search by meaning instead of file names. So, Museum is available now for macOS and Windows in beta. The way the team describes it is your local AI and personal digital museum. Uh which sounds abstract until you see it work, right?
So, here's what's actually happening. I'm connecting it to my external SSD hard drives, and it goes where your drives are. You're not uploading any data. It's not leaving your computer or the safety of your drives. Museum does not move them. It doesn't upload them anywhere. It actually builds a local index and then analyzes what's in them.
So, I've connected it to two of my local SSD drives where I've got backup of photos. I've got YouTube video footage, lots of content on all of those. I can also connect it to my 16 TB uh backup drive. The other nice thing I like about this so far is I can actually connect this to my cloud drives as well. So, I can connect it to my photo album, which is connected to my iCloud account. So,
I've got about 43,000 files in iCloud just collected across years and years of iPhones. Museum will import all of that and then analyze it all locally. So, let's head over to my computer and I can give you guys a look at Museum. So, first of all, this is my Museum library. I've got a ton of photos in here, video files, wave files,
>> >> just tons of them. You can see here I've got about 8,000 files currently loaded into this Museum library. And the easiest way to do that, again, you just go to import. You can import from your computer. So, for me, what I did is I picked my two external SSD drives right here and I loaded all of these files in.
Also, you can import from cloud services like Google Drive, Google Photos, Dropbox, iCloud Photos, and your iCloud Drive. I've actually imported a bunch of photos from my iCloud Photos, and I'll just give you a quick look as well.
And here we are in our iCloud photo album. I can just choose all of my photos, you know, quick shift click, and I can just add or, again, if you are someone like me who's got about 43,000 files in iCloud, you can kind of add them as you see fit. But, right now I've added maybe about a thousand or a couple thousand photos from my iCloud library and also my
drives. Um but, here it is. All of my photos, all of my videos. We can go here, we can sort by photos. We can sort by videos. We've also got some of my R3D files in here as well.
That's one thing I will call out is Museum, at the moment, it doesn't support red raw files. You know, not a problem for a lot of people, but just just so you know, MP4s, normal MOVs, and other video files, no problem. And then here all of my audio files. So, I film a lot of independent audio on my DJI mics, so they are also here.
Documents, >> >> I don't have any documents loaded here, but for now, this is what you're seeing. But for now, this is what you're seeing. The other thing that you're going to notice, too, is over here, we have people. So, all of these faces it's identified. I can go through and start to apply names to this, but we're not going to do that right now. What I want to show you is how the vibe search functionality works and how this is
really changing my workflow and actually quite a good use of AI as a creator, right? It's not replacing the creative process, it's actually helping me streamline one of the most mundane and >> >> tooth-pullingly painful uh aspects of being a creator, and that's sifting through all of my different files and finding the right things that I need for when I'm working on a project. So, a quick vibe search, let's just do kids at the beach. It's a search I've done
before. And what it's going to do now is it's going to go through, find all of my photos of people and kids at the beach. So, we've got some photos here, but mostly I've actually got a ton of videos. So, it's actually been able to go through, watch all the videos.
You can see here it's using contextual matching. So, it can identify all the different elements, and this is what Museum is using to pull this when I'm searching for something like this.
So, maybe another one is like uh skiing. If I'm doing a video on a ski trip, if I'm pulling up images from my recent trip to Vail with my daughter for her birthday, and I want to put together a montage or some type of video, we can go ahead and find all of that here as well.
And again, contextual matching, it can even identify this is a nice little town of Vail, which is beautiful, doesn't explicitly show skiing, but it does connect to these other files here. Now, more useful as a creator, let's just say I am doing a travel vlog. I want some b-roll of me walking through an airport.
That's something I use quite often so that I can mix it into my b-roll just to show the transition uh of me moving from destination to destination. And here we go. All this footage of me and other people moving through different airports. Again, really, really useful, really, really convenient. And maybe another example is pretty helpful, especially if you are someone like me who has a really big archive of photos from just years
and years ago. You know, if you're putting together a photo album or a collection, pictures of my kids when they were young. My kids are still young, but look, it's pulling up baby photos as well. So, this is something that's really helpful because it's not pulling up any of the recent shots of my kids, which there are plenty while they're older, it's pulling up all
the ones of when they are babies. And also corresponding videos as well. And I'll show you just how I quickly add a collection. Let's do airport transition. And we'll do a smart collection, and here we can just describe any photos or videos of airports, airplanes,
and me moving through a city or even hotel with a suitcase. That way in the future as I'm building travel montages, as I'm looking to create transitions, I can go to this collection and it'll start to pull a lot of this info. So, we've got suitcases.
[snorts] These are just shots of me in hotel rooms, hotel elevators. You can see here we've got shots of kids in the elevators, these photos here of me with bags. This thing has done a really solid job of identifying anything that is relatively travel related. Suitcases as the main callout, airplanes, airports.
We've even got some boats and ferries here. So, a pretty solid job at bringing this collection to life. And on top of the contextual data, it should be able to pull metadata from each of the individual images, maybe even location data. So, let's just say photos from Greece.
And these are accurate. These are photos from my recent trip to Athens. So, again, not only can it get contextual data, but it can pair that with metadata that should be embedded in each of your photos. I would be very impressed if it saw these images and automatically identified them as landmarks in Athens, Greece, but my gut tells me it's a combination of what's happening in the photo plus the metadata and the location
um information from those. So, same thing with videos, right? This is really quite cool. These are all shots from Greece. I can attest to that. They are from all of my different cameras, my Leica Q3. These are DJI Osmo Pocket 4 shots that are in log right now, so the vibe searching here, it's good. And this ultimately is what's