Building a Photo Archive: Finding Tens of Thousands of Railway Photos Again

Updated: August 2026 – Anyone wanting to build a photo archive doesn’t struggle with storage space after a few years, but with the question: where was that picture of that one locomotive in Hannover again? Tens of thousands of railway photos aren’t a problem – tens of thousands of badly named railway photos are. This guide shows a naming scheme, a keyword strategy and a backup structure that will still work in twenty years’ time.

How do you build a photo archive you can actually find things in again?

A working photo archive consists of three separate layers: folder structure, file name and metadata. The folder structure sorts by date and stays unambiguous forever as a result. The file name carries the key identifying details, so a file is recognisable even without any software. The metadata – keywords, title, description – handles everything thematic and makes the archive searchable. Museums such as the Cité du Train work on the same principle, just on a larger scale: without clean tagging, a collection isn’t usable.

The most common mistake is squeezing all three tasks into folder names: a folder per class, inside that a folder per location, inside that by year. That works until the first picture that belongs in two categories – so, by about week three. The rule is therefore: sort folders, categorise via metadata. What a well-kept archive is worth after decades is shown by the railway archive with more than 96,000 historical photos.

The naming scheme

A good file name answers four questions without you having to open the file: when, where, which service, which vehicle. A proven pattern looks like this:

2026-08-19_Koeln-Hbf_RE1_146-xxx_001.jpg

The individual parts:

• Date in YYYY-MM-DD format. It goes at the front, because that makes alphabetical sorting automatically chronological – on every system, in every program.
• Location or operating point. Short, unambiguous, no umlauts. “Koeln-Hbf” instead of “Cologne Hauptbahnhof”.
• Service. The train category and line number, or the train number itself. That identifies the journey.
• Vehicle. The number in a fixed notation – the same one you also use in your sighting list.
• Running number. Three digits, so a series from one moment stays together.

Three technical rules matter more than they sound: no spaces, no umlauts, no special characters. Spaces break scripts and URLs, umlauts occasionally get mangled when moving between operating systems and cloud storage, and characters such as a forward slash or colon are simply forbidden in file names.

Layer	Task	Concrete implementation
Folder structure	Sort unambiguously	Archive/2026/2026-08-19/ – year, then shooting date inside it
File name	Make it identifiable without software	Date, location, service, vehicle number, running number
EXIF	Provide technical shooting data	Comes from the camera – the time must be correct
IPTC/XMP	Make it thematically searchable	Keywords, title, description, author
Rating	Control selection	Stars or colour labels for selection, editing, publication
Output folder	Separate finished versions	Export/ – can be regenerated at any time, not part of the archive

Keywords: the real search tool

The file name answers “which picture is this”, the keywords answer “which pictures do I have on this topic”. For railway photography, a hierarchical structure has proven itself, where each layer stays searchable on its own:

• Vehicle: class, specific number, type, traction
• Operator: railway undertaking, and where relevant the owner – for older shots, the name that was valid at the time, so in Sweden Banverket rather than today’s successor organisation
• Location: operating point, town, federal state
• Route: line name or timetable route number
• Shot type: side-on, three-quarter, panning shot, night shot, detail
• Occasion: special working, diversion, replacement service, engineering works – for society-run trains, also the organiser, such as the ÖGEG

Consistency is what matters: a controlled keyword list you choose from, rather than typing freely each time. Otherwise, after two years, you’ll have “class 146”, “BR146”, “Baureihe 146” and “146er” side by side – and every search only finds part of them. Most photo management programs support such controlled lists; use them from the start.

The workflow after every trip

The fifth step is the one most people skip – and it’s exactly the one that links photography and documentation. The guide to using EXIF data for spotting describes how to reconstruct missing details from your image data afterwards.

Backup: three copies, two media, one other location

The 3-2-1 rule isn’t a recommendation for professionals; it’s the minimum kit for anyone whose pictures mean something to them:

• Three copies – the original and two backups. Two copies mean that a single failure can set you back to zero.
• Two media types – say, an internal hard drive plus an external drive or network storage. Identical media age similarly and tend to fail around the same time.
• One off-site location – cloud storage or a drive kept at a relative’s house. Fire, water damage and theft hit every device in one room at the same time.

Two additions that are often missing in practice: versioning and testing. A syncing cloud folder isn’t a full backup, because an accidental deletion gets synced everywhere immediately. And a backup that’s never been restored is just a hope. Twice a year, take a few files from your backup and actually open them.

Sorting out existing chaos afterwards

Most people don’t start from zero, but already have years of unsorted pictures. The mistake would be trying to tidy everything at once – nobody manages that, and it gets abandoned after two evenings. A two-track approach works better:

• Clean from today onwards. The new scheme applies to every new trip from now on. That way the organised part grows on its own.
• Work backwards in stages. Set aside a year or a month per week. Start with the most recent years, because you’ll still remember the details there.
• Automate where you can. Date and time sit in the EXIF data; photo management and renaming tools can generate folders and file names from it. What’s left to do manually is only what the program doesn’t know: location, service, vehicle number.
• Prioritise. Pictures you want to publish or attach to sightings come first. The rest can wait. For publishing itself, the guide to bahnbilder.de with workflow and legal questions offers the right pointers, complemented by the Flickr guide for railway photography with tags and groups.

Anyone wanting to work on photographic quality at the same time will find the basics in the train photography guide – an archive full of good pictures is a lot more motivating to maintain.

Self-test: how robust is your archive?

Common mistakes when building an archive

• Keeping camera file names. After your second camera body, or the first counter overflow, you’ll get duplicate file names – and data loss when merging folders.
• Creating thematic folders. By the first picture that belongs to two topics at the latest, the copying starts – and with it, the chaos.
• Using umlauts and spaces. Works locally, causes trouble when moving to other systems, cloud storage and web interfaces.
• Only maintaining metadata in the catalogue. When you switch software, all that work is gone if nothing was written into the files.
• Confusing backup with syncing. Whatever you delete gets deleted everywhere immediately. Without versioning, that’s no protection.
• Keeping sightings and pictures separate. Anyone maintaining both systems without a shared key ends up searching twice – the comparison between forum and app for live sightings shows why structured fields make the difference here. For systematically working through a class, the guide to completing your vehicle collection is also worth reading.

Conclusion

Building a photo archive is an investment of a few minutes per trip that pays off more with every year. Folders by date, file names with date, location, service and vehicle number, keywords from a controlled list, and a 3-2-1 backup – that’s all it takes. The real payoff comes where your archive and your sighting data use the same notation: then ten thousand files become a collection where you can answer any question. The kind of finds this can uncover is shown by the guide to Drehscheibe Online RailView.