Daniel sent us this one — he's been roaming Jerusalem with a camera, discovering that the best stock photo is the one you shoot yourself, because stock libraries never have that one specific alley in golden hour. But he's hit the classic wall: building the library is joyful, retrieving anything from it six months later is a nightmare. He wants a practical walkthrough — capture strategies, pro tagging tips, and a real comparison of tools, from Google Photos up to self-hosted DAMs.
This is one of those problems where the fun part and the useful part are completely at odds. Shooting stock feels ridiculous in the moment. You're standing in an alley photographing a dumpster lid because the rust pattern is interesting, and every passerby thinks you've lost your mind.
The trash can lid that becomes your next thumbnail background. There's a sentence that captures the entire creative condition.
And then six months later you're editing a video at two in the morning, you need a texture that says "urban decay but make it warm," and that rust pattern saves your project. The problem is you can't remember what you named it, where you put it, or whether you even backed it up. So let's structure this properly. There are really two separate challenges here. One is capture — developing the instinct for what's worth shooting even when it feels pointless. The other is retrieval — building a system where you can find that rust pattern at two in the morning without wanting to throw your laptop out the window.
The retrieval part is where most people fail, because the tools that make capture easy don't necessarily make search easy, and the tools that make search powerful often make capture a chore. So where do we start? The shooting itself?
Let's start with the psychology of stock creation, because I think that's where most creators get tripped up before they even begin. When you shoot for a specific project, you know exactly what you need. You frame it, you light it, you get the shot, you use it. But shooting for stock is the opposite. You're shooting with no immediate purpose, which means your brain keeps asking, why am I wasting storage on a fire hydrant?
The fire hydrant that later becomes the perfect thumbnail for your episode on municipal infrastructure. Of course there are.
That's the shift you have to make. The best stock photographers — and I'm using "stock photographer" loosely here, I mean anyone building a personal library — they develop what I'd call a texture instinct. They see the world in terms of backgrounds, patterns, negative space, and color palettes. A brick wall isn't just a brick wall. It's a warm red texture with horizontal lines that could anchor a title card. A patch of sky between buildings isn't just sky. It's negative space with a specific shade of blue that matches your channel's brand palette.
You're training yourself to see in terms of design elements rather than subjects. Which actually makes the shooting easier, because you're not looking for a perfect photograph. You're collecting raw materials.
And that brings me to the first practical tip, which is about volume and variety. When you do a deliberate stock shoot — and I think deliberate is the key word here — you should aim for quantity over perfection. Set your camera to burst mode, walk through a neighborhood, and just capture everything that catches your eye. Reflections in puddles. Shadows on walls. The way light hits a particular window. A cat sitting on a parked car. None of these are portfolio shots. All of them are potentially useful.
The eighty-twenty rule applies here, doesn't it? Most of your usable stock comes from a small number of shooting sessions.
And the sessions that produce the most usable stock are the ones where you deliberately go hunting for textures, patterns, and negative space. Not the ones where you happen to have your camera out while doing something else. There's a case study here that's worth walking through. Daniel mentioned shooting around Jerusalem. Let's say you dedicate a single afternoon — two hours, one neighborhood, no specific agenda. You're just walking and shooting. In that session you might capture fifty or sixty frames. Of those, maybe forty are usable as stock — they're properly exposed, sharp enough, and have compositional elements that could work in a layout. Of those forty, you'll probably use twelve to fifteen across various projects over the next two years.
Those twelve to fifteen uses might include a thumbnail background for a podcast episode, a texture overlay for a video intro, a placeholder image for a blog post, a background for a quote card on social media. One afternoon of shooting feeds eight different projects.
Here's the thing — you don't know which twelve of those forty will be useful when you're shooting. You can't know. That's the whole point. So you shoot all forty and trust that your future self will be grateful.
Like adopting a feral cat. You don't know which one's going to curl up on your keyboard while you're trying to work, but one of them definitely will.
That's an image. Let's talk about technical capture tips, because there are a few things that make a real difference in post-production flexibility. First, shoot RAW plus JPEG. RAW gives you maximum editing latitude — you can recover shadows, adjust white balance, pull detail out of highlights. JPEG gives you an immediately usable file for quick projects where you don't want to spend time editing. Storage is cheap.
For video, stabilization is everything. Nothing ruins stock footage faster than shaky handheld shots. If you're shooting with a gimbal, great. If you're shooting handheld, use a wider lens and keep your shutter speed at least double your frame rate. For thirty frames per second, that's one-sixtieth of a second minimum. And shoot in the highest resolution your camera supports. You can always downscale four-K to ten-eighty-P, but you can't go the other direction. Also, shoot clips longer than you think you need. A ten-second clip of a street scene gives you room to find a stable section, apply a slow zoom in post, or use it as a looping background.
For handheld street photography specifically, what settings do you recommend?
For stills, I'd keep your ISO on auto with a minimum shutter speed of one two-hundred-fiftieth of a second. That'll freeze most motion and compensate for hand shake. Aperture priority mode, somewhere between F four and F eight, depending on how much depth of field you want. For street scenes where you want context, F eight gives you sharpness throughout the frame. For isolating a texture or detail, open up to F four or wider. And if you're shooting in harsh midday sun — which Jerusalem gets plenty of — underexpose by about a third of a stop. You can recover shadows in post, but blown highlights are gone forever.
The midday sun problem is real. You're walking through the shuk at noon and half your frame is in blazing sunlight and the other half is in deep shadow. That's a nightmare for exposure.
That's exactly where RAW saves you. A modern camera sensor can capture twelve to fourteen stops of dynamic range in RAW. JPEG throws away about half of that. So that shadow detail that looks completely black in your JPEG preview? It's there in the RAW file, waiting to be recovered. But let's move on to the second phase, because capturing stock is only half the battle. The real challenge is finding that one photo six months later when you're on deadline and can't remember what you named it.
This is where I see creators hit the wall. They've got thousands of photos sitting in a folder called "Stock" with filenames like IMG underscore four seven two one dot CR two. That's not a library. That's a landfill.
A landfill with treasures buried somewhere inside it, and no map. So let's talk metadata. Metadata is the difference between a pile of files and a searchable library. And I want to break this down into two categories. Structural metadata is the stuff your camera captures automatically — date, time, camera model, lens, focal length, GPS coordinates, exposure settings. Descriptive metadata is the stuff you add — tags, keywords, captions, ratings, color labels. Structural metadata answers the question "when and how was this taken?" Descriptive metadata answers the question "what is this and where would I use it?
Most people rely entirely on structural metadata without realizing it. They scroll through their timeline view, find the trip they took in March, and hope the photo they need is somewhere in those two weeks. That works until you've taken twenty trips and can't remember which March.
Or until you need a photo of a specific thing — a doorway, a texture, a color palette — and you have no idea when you shot it. That's where descriptive metadata becomes essential. And I want to propose a tagging system that I think works particularly well for creator stock libraries. It's a three-tier hierarchy. Tier one is subject — what is this a photo of? Tier two is mood or color — how does this photo feel, what's the dominant color palette? Tier three is use case — where would I actually put this?
Give me an example. Walk me through tagging a specific photo.
Let's take that Jerusalem alley Daniel mentioned. Tier one, subject: "Jerusalem alley, stone walls, arched doorway." Tier two, mood and color: "warm tones, golden hour, shadows, texture." Tier three, use case: "thumbnail background, title card, video overlay." So the full tag string might look like "Jerusalem alley, stone walls, arched doorway, warm tones, golden hour, shadows, texture, thumbnail background, title card, video overlay." Now when you search for "warm tones thumbnail background," that photo surfaces immediately.
You can batch-apply these tags using something like ExifTool, right? You're not sitting there typing all of that into each individual photo's metadata panel.
ExifTool is a free command-line tool that's been maintained by Phil Harvey since two thousand three. It reads, writes, and edits metadata across basically every image format. And you can batch-process entire folders with a single command. So if you've just come back from a shoot and you've got forty photos of Jerusalem alleys, you can apply the base tags to all forty in one go, then go back and add specific tags to individual standouts. The command might look something like "exiftool hyphen keywords plus equals Jerusalem hyphen keywords plus equals alley hyphen keywords plus equals warm underscore tones star dot ARW." That's a simplified example, but you get the idea.
For the non-command-line crowd, there are GUI wrappers for ExifTool, but honestly, learning a few basic commands is worth the hour it takes. The alternative is manually tagging four thousand photos in Lightroom and losing your mind.
Losing your metadata if you ever leave Lightroom, which brings me to the tooling conversation. Let's walk through the four tiers of stock library management, starting with the one everyone already has.
The tool Daniel says doesn't get enough love, followed immediately by his grievances about backupability. So let's be honest about what it is and isn't.
Google Photos is genuinely excellent at what it does well. The AI search is remarkable. You can type "brick wall" and it'll find every photo with a brick wall in it, even if you never tagged anything. You can search by location, by date, by object, by face. The auto-curation features — the "this day last year" memories, the automatic albums, the suggested collages — those are useful for browsing. And the storage situation, as of mid two thousand twenty-six, is this: if you're on a legacy account that had unlimited high-quality storage before the policy changed in June twenty twenty-one, you may still have it. Pixel device users also get some unlimited storage at reduced quality. But for most new users, Google Photos storage counts against your Google account quota, and you get fifteen gigabytes free across Gmail, Drive, and Photos.
The compressed "storage saver" tier caps photos at sixteen megapixels and videos at ten-eighty-P. Which for most stock purposes is actually fine — you're not printing billboards, you're using these as backgrounds and overlays.
But here's the grievance, and it's a big one. Google Photos has no native metadata export. You cannot download your photos with the tags, albums, and organizational structure you've built inside the platform. If you use Google Takeout to export your library, you get a pile of files with no folder structure and a separate JSON file with your metadata. Reconstructing your library from that export is a nightmare. It's essentially a one-way trip. You put your photos in, and they live there forever, but you can't easily take them somewhere else.
Which means Google Photos is a single point of failure for your stock library. If Google changes the pricing, or discontinues features, or you just decide you want to move to something else, you're stuck. Your library isn't portable.
That's the fundamental tension. Google Photos is the best browsing and discovery tool on the market, and it's also the worst for long-term archival and portability. So my recommendation is to use it, but not rely on it exclusively. Use it for casual browsing, for AI search, for rediscovering old photos. But maintain a separate, properly structured archive of your best stock photos somewhere else.
Which brings us to tier two — Apple Photos. Similar story, different tradeoffs.
Apple Photos has better metadata handling than Google Photos, particularly if you're exporting. It supports IPTC metadata, which is the standard for descriptive tags, and it can export sidecar XMP files alongside your images. That means your tags, ratings, and keywords can travel with your photos if you move to another platform. But it's still a walled garden. iCloud storage limits apply, and if you're shooting high-resolution RAW plus four-K video, you'll blow through even the two-terabyte iCloud plan faster than you'd think. Apple Photos is good for casual users, good for people who live entirely in the Apple ecosystem, but for a serious stock library at scale, it starts to creak.
Both Google Photos and Apple Photos share the same fundamental limitation — they're designed for personal photo management, not for asset management. They organize around chronology and faces and locations, not around use cases and color palettes and project needs. They don't know what a thumbnail background is.
Which brings us to tier three — self-hosted digital asset management systems, or DAMs. And this is where things get interesting for creators who are willing to put in a little setup work. There are three main options worth talking about. Immich, Photoprism, and Piwigo. Immich is the one that's gotten the most attention recently. It's an open-source, self-hosted alternative to Google Photos, and its version one point nine four release in March two thousand twenty-six added facial recognition and significantly improved EXIF extraction. It has a web UI that rivals Google Photos in terms of polish, it supports automatic backup from your phone, and because it's self-hosted, you control your data completely.
"self-hosted" for the non-technical listener means what exactly?
It means you run the software on your own hardware — a home server, a NAS, an old computer you've repurposed, or a virtual private server in the cloud. Immich has a one-click Docker install, which means you can get it running in about fifteen minutes if you're comfortable with basic command-line stuff. If you're not, there are step-by-step guides that walk you through it. Once it's running, you access it through a web browser or a mobile app, just like Google Photos. The difference is that your photos live on your hard drive, in a folder structure you control, with metadata that you can export or migrate at any time.
Because there are always tradeoffs.
Three main ones. First, you're responsible for backups. If your server dies and you don't have a backup, your library is gone. Second, you need to maintain the software — updates, security patches, occasional troubleshooting. Third, the AI search isn't as sophisticated as Google's. Immich has object detection and facial recognition, but it's not going to match Google's ability to search for "sunset over water with a sailboat in the left third of the frame." You're trading some search intelligence for complete data sovereignty.
Photoprism and Piwigo are similar but with different strengths?
Photoprism is more focused on AI-powered tagging and organization. It uses TensorFlow to automatically classify and tag your photos, which means it can do a lot of the metadata work for you. Piwigo is older and more established, with a huge plugin ecosystem, but its interface feels dated compared to Immich and Photoprism. For most individual creators building a personal stock library, Immich is probably the sweet spot right now — it's modern, it's actively developed, and it balances ease of use with control.
Then tier four is the enterprise stuff — Bynder, Widen, Cloudinary. The kind of tools that large media organizations use to manage millions of assets across teams.
And for an individual creator, these are mostly overkill. But they're worth mentioning because they offer features that the consumer tools don't. AI auto-tagging that actually works at scale. Version control so you can track edits. API access so you can integrate your stock library directly into your content management system or video editor. Cloudinary in particular is interesting because it has a free tier — up to twenty-five gigabytes of storage and twenty-five gigabytes of monthly bandwidth. That's enough for a curated library of your best stock assets, and Cloudinary's image transformation API means you can dynamically resize, crop, and optimize images on the fly.
You upload a full-resolution RAW file, and when you need a ten-eighty-P JPEG for a thumbnail, you just request it at that size through the URL. You're not storing multiple versions of the same image.
It's a completely different paradigm from the folder-based approach. But it requires thinking of your stock library as an API rather than a file cabinet, which is a mental shift most creators aren't ready to make.
If no single tool is perfect, what do you actually do? What's the hybrid strategy that actually works?
After two years of trial and error, here's what I've settled on. Use Google Photos as your ingestion layer and casual browsing tool. It's the best at getting photos off your phone and into a searchable library with zero friction. But don't treat it as your archive. Once a month, export your best new stock photos — the ones you actually think you'll use — to a self-hosted Immich instance or even just a structured folder on Dropbox or a NAS. Apply your three-tier tags using ExifTool or directly in Immich. Keep a simple spreadsheet as your index of last resort — just a CSV with filename, tags, and a one-sentence description. That spreadsheet is your insurance policy. If every tool you use disappears tomorrow, you can rebuild from that spreadsheet.
The spreadsheet approach sounds low-tech, but it's actually the most future-proof part of the whole system. A CSV file will be readable in fifty years. Your Google Photos library might not exist in five.
That's the digital dark age problem in a nutshell. We talked about media format obsolescence in a previous episode — the idea that we're generating more data than ever but losing the ability to read it as formats and platforms die. Your personal stock library has the same vulnerability. If your entire library lives inside a single platform, you're betting that platform will exist and maintain compatibility for the rest of your career. That's a bad bet.
Folders are brittle too. I've seen people organize their entire stock library into nested folders — "Travel, Europe, France, Paris, Streets, Alleys" — and then they can't find anything because they don't remember which folder they put it in, or the photo belongs in three folders at once. Folders impose a single hierarchy on data that's inherently multi-dimensional.
Metadata is forever. Folders are a snapshot of how you were thinking on the day you filed the photo. Tags let you find the same photo through multiple paths. Subject, mood, use case, color, location, project — any of those can be the entry point that gets you to the right image.
Let's get concrete about the tagging system. You mentioned the three-tier hierarchy earlier. Walk me through implementing it in practice, for someone who's never tagged a photo in their life.
Start with the photos you already use. Don't try to tag your entire library — that's a recipe for burnout. Pick the fifty or hundred photos you actually reach for regularly. Open them in whatever tool you use — Lightroom, Apple Photos, Immich, even just the file properties panel in Windows or MacOS. For each photo, ask yourself three questions. What is this? How does it feel? Where would I use it?
"What is this?" seems almost too obvious. But the trick is being specific enough to be useful without being so specific that the tag only applies to one photo.
"Building" is useless — you have five hundred photos of buildings. "Jerusalem stone building with arched windows" is useful. Think like a search engine. What terms would your future self type to find this photo? Use those as your tags. And don't be afraid to use multiple subject tags. A photo of a cat sitting on a Jerusalem windowsill gets tagged "cat," "Jerusalem," "windowsill," "stone texture," "animal," "urban." Each of those is a potential search path.
The mood and color tags — those feel subjective. "Warm tones" means something different depending on the photo.
Develop a consistent vocabulary and stick to it. I use a limited palette of mood tags: warm tones, cool tones, bright, moody, high contrast, soft, natural light, golden hour, blue hour, overcast. For colors, I tag the dominant one or two: red, blue, green, warm orange, cool blue, neutral, monochrome. The key is consistency. If you sometimes tag "golden hour" and sometimes tag "warm sunset light," you've created two separate tags for the same thing, and your search results will be incomplete.
Use case tags — those are the ones that save you at two in the morning. "Thumbnail background," "title card," "video overlay," "social media banner," "website hero," "podcast cover art." Anything that describes the specific creative context where the photo works.
Here's a pro tip that took me way too long to figure out. Tag your photos for the projects they've already been used in. If a photo worked perfectly as a thumbnail background for an episode about urban planning, tag it "urban planning thumbnail." Next time you need a thumbnail for a similar topic, that photo surfaces immediately, and you don't have to reinvent the wheel.
That's clever. You're building a usage history into your metadata, so your library gets smarter the more you use it.
And this is where I think the next frontier is. Automated metadata generation based on actual usage patterns. Imagine a tool that watches your photo library, sees which photos you use for thumbnails, which ones you use for video overlays, which color palettes you gravitate toward, and then suggests tags and surfaces similar unused photos. That's not science fiction — it's a fairly straightforward machine learning problem. The training data is your own behavior.
The algorithmic version of "you might also like," but for your own photos. That's actually compelling. And it solves the problem of the photos you forgot you even took.
Which is most of them. The average person takes thousands of photos a year and never looks at ninety percent of them again. A smart enough system could surface the hidden gems — the perfectly composed shot you took three years ago that matches the color grade of your current project. That's the promise. But we're not quite there yet with consumer tools.
In the meantime, we tag manually and hope our future selves appreciate the effort. Let's talk about the one thing people should absolutely avoid doing.
Relying solely on folder structures. I've seen people build elaborate folder hierarchies — year, month, event, location — and then discover that they can't find a photo of a doorway because they don't remember which event they were attending when they took it. Folders are a single axis of organization. Photos are inherently multi-dimensional. A photo has a date, a location, a subject, a mood, a color palette, a use case, and associations with projects and people. You can't capture all of that in a folder name. You need metadata.
The folder approach fails catastrophically when you reorganize. If you decide that your current system isn't working and you want to restructure, you're moving thousands of files around and breaking any external references. With metadata, the files can live anywhere. The tags are inside the file.
That's the key insight. Embed your metadata in the file itself using industry-standard formats like IPTC and XMP. That way, your organizational system travels with the photo, regardless of what tool or platform you're using. ExifTool can write IPTC keywords directly into your RAW and JPEG files. Lightroom can do it on export. Immich and Photoprism both support it. Make your metadata portable.
To pull this all together — the hybrid strategy is Google Photos for ingestion and casual browsing, a self-hosted DAM or structured cloud folder for your curated stock archive, three-tier metadata tagging applied consistently, and a CSV spreadsheet as your disaster recovery plan.
A monthly stock shoot. Thirty minutes, no agenda, just hunting for textures and backgrounds and negative space. That's the habit that feeds the whole system. Without new material coming in, you're just organizing an increasingly stale collection.
The "shoot something ridiculous this week" principle. Your future self will thank you, even if the people watching you photograph a dumpster lid right now think you've lost your mind.
That's the joy of it, honestly. There's something liberating about shooting without a brief. No client, no deadline, no specific deliverable. Just you and a camera, paying attention to the visual texture of the world. It's a creative practice as much as a practical one.
The practical payoff is real. One afternoon of shooting, forty usable photos, twelve video clips, eight projects over two years. That's a return on investment that most creative workflows can't match.
The one thing I'd add is to not overthink the tooling. It's easy to spend more time configuring your DAM than actually using your photos. Start with what you have. If that's Google Photos and a spreadsheet, great. If it's Apple Photos and manual tagging, great. The best stock library is the one you actually use. Everything else is optimization.
The actionable takeaways. One — do a dedicated stock shoot once a month. Thirty minutes, textures and backgrounds, no people, no specific subjects. Two — implement the three-tier tagging system immediately on your top fifty photos. Subject, mood and color, use case. Three — use a hybrid tooling strategy. Google Photos for browsing, a self-hosted or cloud archive for your curated collection, and a CSV spreadsheet as your index of last resort. And the one thing to avoid — relying solely on folders. Metadata is forever. Folders are brittle.
If you want to go deeper on the self-hosted side, Immich version one point nine four is the current release as of March two thousand twenty-six, with facial recognition and improved EXIF extraction. ExifTool is free and maintained by Phil Harvey since two thousand three. And Cloudinary's free tier gives you twenty-five gigabytes to experiment with if you want to try the API-based approach.
The open question I keep coming back to is this. As AI-generated stock becomes indistinguishable from real photos — and it's getting close — does a personal stock library become more valuable or less? On one hand, authentic photography becomes a differentiator. On the other hand, if an AI can generate a perfect "Jerusalem alley at golden hour" in seconds, why spend an afternoon shooting one?
I think it becomes more valuable, but for a different reason. AI-generated stock is generic by definition — it's trained on the aggregate of what exists. Your personal stock library is specific to your eye, your aesthetic, your projects. It has a visual coherence that no prompt-engineered image can replicate. When someone sees your thumbnail and recognizes it as your style, that's branding. That's value. AI can mimic a style, but it can't originate your specific way of seeing.
The personal stock library is ultimately an extension of creative voice. It's not just a convenience. It's a signature.
That's worth the effort of building it right.
Now: Hilbert's daily fun fact.
Hilbert: In nineteen forty-seven, Mauritius issued a stamp depicting a dodo that was so poorly engraved, philatelists nicknamed it the "drunken dodo." The stamp's face value was twenty cents. At auction in two thousand twenty-one — adjusted for inflation, that twenty-cent stamp would be about three dollars today — it sold for over ninety thousand dollars.
The drunken dodo. There's a metaphor in there somewhere.
I'm not sure I want to find it.
This has been My Weird Prompts. Our producer is Hilbert Flumingtop. If you enjoyed this episode, leave us a review wherever you get your podcasts — it helps more people find the show. Now go shoot something ridiculous this week. It might be your next thumbnail.
