I am trying to figure out a reasonable naming convention for files and folders for my digital photos. Most of the discussion I have seen seems to center on date based conventions, but I don’t see the value of that, just yet…at least not for me. Usually, the strategy being discussed is a pro or semi-pro photographer who might be relating a set of images to a wedding or some other occasion, where I am sure the date is extremely relevant.

I started off taking snapshots with an Olympus C-2020 and the volume of photos was on the order of hundreds a year, certainly not thousands. Over time, I had maybe twenty high level folders with names like, Trips, Holidays or Family. Under Holidays, I might have Easter 2006. I didn’t much care what date Easter fell on that year, the holiday and the year pretty much summed it up. In addition, I did not really need to tweak the photo names beyond the camera assigned name since most folders only had maybe fifty pictures in them and they were all related. Finally, I did not really process the photos to any great degree. I had access to various tools to handle simple problems like red eye or cropping small imperfections, but I figured I’d get some decent software “someday” and I could always deal with any photos I really wanted to massage later on.

This sort of mechanism allowed me to find most pictures relatively quickly, but I imagine it has limitations when the volume increases (which it has). That being said, knowing the date from a folder name doesn’t seem particularly useful to me…to find my photos in the distant future, I would have to go figure out what day Easter fell on in 2006 to determine where to find the photos….

Skipping forward a bit, I have now graduated to a Canon EOS 20D, have a (newly acquired) copy of Photoshop CS3 and am taking a LOT more photos. After spending only a minimal amount of time playing with Bridge and Photoshop, I can see I have a lot more power at my disposal and I’d like to get things set up “properly” so I don’t have to go back and change my naming conventions a number of times while I work things out. However, I don’t really see that any one naming convention is going to be a lot more powerful than another. The power in these tools seems to be in using keywords, ranking and perhaps the creation date to help me find things.

While it would be cumbersome, it almost seems like I could stick everything in one big folder and rely on the tool to find a given set of photos. I understand this would not be optimal, I am just making a point about the power of the search tools and how little import the folder name seems to have. While the use of dates in a folder name would be slightly more useful than a generic sequence number, it doesn’t scream ideal.

I guess what I am ultimately saying is that when you do not have a reasonable way to search your images, a hierarchical set of names for folders is helpful, but when you have access to metadata with your software, the naming conventions seem more of a way to enforce uniqueness and the structure serves a much smaller purpose. So, perhaps I am spending too much time worrying over something that really doesn’t matter?

At this point, I am not particularly concerned about what to do with processed images. I have seen quite a bit written about either using subfolders or tweaking filenames. I think I can work that part out easily enough, I am more interested in the higher level folder structure. If anyone has any commentary or suggestions, it would be greatly appreciated.

sf

I have found that most software tends to be somewhat intolerant of changes made to the folder structure outside the software - Adobe in particular. I also find that entering keywords is more than a little onerous. With that in mind, my current strategy is:

Dates are available in the EXIF and any competent search program ought to be able to select by EXIF date. Therefore I do not depend on dates in folder names. I may use them to provide uniqueness and because import programs like to also. I dump from card to a folder called cameras and let Bridge create a folder with the date as a name. Or I will do that manually if not converting to DNG.

I use folders as do you, your trips example is good. The problem with this approach is when an image logically fits in several folders and we, of course, only want one copy of the image. Another problem arises when we have an especially good image; one that we work intensively in PS and then print in multiple sizes and/or have a B&W version. Then there are the web & email copies and so on. Most of that can be handled with filename tweaks of one sort or another especially since the PS stuff gets saved as either PSD or TIF.

I don't much like Bridge as an image DB manager. I use (and have used since the DOS days) ThumbsPlus (www.cerious.com). It isn't perfect, but I haven't found anything better for the price for what I need and want. I want to try Lightroom, but there is other software I need more first.

To conclude this ramble, I have two suggestions:

1) - Stick with what you are doing until you have sufficient reason to change.
2) - Make sure that any software you use will survive radical reorganizations without having to start over from scratch. Reentering keywords and/or ratings is no fun.

I am approaching 20K images that occupy appx 100GB of space.

The value thereof is the fact that this system is simple, reliable, easy to back-up (and to restore in case of calamity), and endlessly scalable---that is, it will fit small, medium, large, and huge amounts of data equally well.



Actually no, the date as such hardly ever is particularly relevant. Instead, it just provides a simple, natural, and unambiguous sorting and discrimination criterion, that's all.



Almost everybody starts out with a system like that---a hierarchical system of folders, named after categories, topics, or events. The problem is: all these systems sooner or later will fail, no matter how careful you've worked your hierarchy out. They work pretty well for several hundred or a few thousand images, but they don't for tens or hundreds of thousands of images. The main reason is the fact that no hierarchy can be orthogonal---that is, it's impossible to build a hierarchy of folders so that every image will fall into exactly one single possible place within it. For example, if you shot a photo of your kid at Easter while on vacation with your family, will the image go into 'Holiday,' 'Trips,' or 'Family'? This kind of problem will grow exponentially with the number of images. So at some point in time, you will have to reorganize your system, and that's gonna be a lot of work.



Exactly.



No. You'd search for "Easter" and "2006." In order for this to work, you'd need to add these keywords to your image files beforehand. With a tool like Adobe Bridge, this can be done in a blink of the eye. You download your image file from the memory card to the hard disk, remove the obvious failures (optional), add some basic captions and keywords to them, and save them in a folder named by the date (and by the event, optionally, like e. g. "2006-04-16_EasterWithAuntKate"). To retrieve the images later, simply search for the keywords.

In essence this means, your "hierarchy" will be purely virtual. This has many advantages. First, every image file has exactly one place to go: the folder with the appropriate date. Second, you can have any arbitrary number of hierarchies at the same time rather than just one. Third, moving all the data to another (usually larger) physical storage as well as back-up and restore are very easy and straightforward.



In an ideal world---yes indeed. Unfortunately, the world isn't ideal; we have to deal with things like computer operating systems which allow only so many files and subfolders per folder, with hard disk crashes, with user errors etc.



That's right. The physical hierarchy of folders is basically meaningless. Instead, captions and keywords create virtual structure.

For further in-depth discussion of this topic, I strongly recommend "The DAM Book" by Peter Krogh. There, the author develops a viable system to store, to back-up, and to work with large amounts of images (and other kinds of files which he calls digital assets), and explains in great detail all the hows and whys. I suggest to adopt his system and to stick to it---that's what I did, and boy I'm glad I did. By the way, DAM stands for Digital Asset Management. Highly recommended reading for every digital photographer who has more than, say, 10,000 images to handle!



Sorry, but in my opinion that's particularly bad advice! I'd suggest to switch to a competent storage system as soon as possible, because the later you do it, the harder it will be. And sooner or later you will be forced to do it.



Now that's good advice. The point with Krogh's system is: it will survive radical reorganisations. It's desiged this way. And while adding keywords to thousands of legacy image files is no fun indeed, once you did it, you will never be required to re-do it. And switching to another set of keywords doesn't require deleting the old ones.

You may keep your current folder system for the time being (albeit not recommended)---but whatever you do, start adding meaningful keywords to your image files as soon as possible. If you don't want to go through this chore for your old images, at least start doing it for all new images.

-- Olaf

o1AF -- great reply. Your explanation of why a hierarchical system will eventually fail is simple, clear, and to the point. I will second the advice to read The DAM book.

sf --

I am one of those who archives everything by date. The date is part of my filenaming convention, all my raw files are in folders by date. Do I ever search for something by the date it was shot? Only rarely. This organizational tool is just an easy and functional way to store my files, and has nothing to do with how I organize and search for individual photos or shoots.

One overwhelming advantage of digital photography is the ability to add reams of data to each photo -- data that can be sorted, searched, read, added to, deleted, etc. Think about the limited data one can add to a page of 35mm b+w negatives -- there is a little strip of white plastic at the top of the page, where one can write a few words, maybe a date, but it's nigh on impossible to fully caption each frame. Digital images can be individually captioned and keyworded, making it child's play to find any or all images from a particular shoot, a single individual or location, or even a concept ("happy").

Make the conceptual break now, while you only have a few thousand images. At the very least, create a keyword list and apply it to your photos. (Adding captions too would be even better.) IN the long run, you'll find it easier to file, archive, and locate the photos that you want.

--Ken

Olaf's analysis is right on. The only thing I would add is that I find Breeze Software's Downloader Pro and Breeze Browser Pro to be excellent tools for building the appropriate file structure and for quickly finding individual images.

There was another fairly recent thread on the forum in which numerous posters discussed their preferred file-naming conventions. A search for "file naming" will probably turn it up.

Thanks to everyone for their input, so far.

The DAM book is already on my list of things to investigate, I have seen it referenced a couple of different places. However, I was concerned that it would be geared more toward organizations (with all sorts of digital assets) that might be well beyond the scope of my needs. I'll definitely give it a look, though.

I work with databases for a living, so I can appreciate the desire to let go of my own hierarchy and allow the keywords to do the work, that is pretty much where I was heading. I guess my confusion has centered on why people are so attached to dates, since this seems to be just another style of hierarchy. Why not just let go entirely and use random numbers for your folders? I remember years ago, Macs had a Hypercard stack that would give you a seed of some sort for filing documents in your drawer so you never filed things by name, you just let them get evenly distributed by the software and when you wanted to look things up, you let the search engine tell you where to find it. I'm not sure I'm ready to go to that extreme, but that is where you end up if you are serious about letting go of hierarchy.

But, I'll certainly concede that a structure based on dates provides a predictable number of files and folders at any one layer and one that is manageable to the file system of any operating system I can imagine. Since that seems to be its primary purpose, it makes a lot more sense why many folks go that route. And, I suspect using dates gives most folks just enough of a hierarchical structure to feel comfortable, while giving the appearance that we are not tied to a hierarchical structure. :-)

Of course, you know what my next thread is going to be about, right? Now that my folder hierarchy is irrelevant for finding a photo, the structure of my keywords takes on an exciting level of importance....playing around with some keywords last night, things started to look suspiciously like my original folder hierarchy. I imagine there is a method to the madness, but I will do a few searches and save it for the topic of another thread, should it come to that.


thanks again,
sf

just another vote for date !

080307_smith_smm1.tif

where

08 is year, 03 is month 07 is date, smith is subject and smm is me

----------

My latest crazy idea for archiving (I am not claiming originality BTW) is to get a gmail account, and every day e-mail yourself a description of the photos taken

then you get a connection to that date and google search power over your descriptions,

attach a jpg of the best shots to the e-mail and you have an offline backup of sorts too

S

Just to drift the thread a little..... I agree that (for now) keyworded images that are stored in virtual sets/folders is the way to go - at least in an ideal world. I've not yet read Krogh's book - I suppose I need to up the priority on that.

My complaint with all the applications, of which I am aware (only a few), is that they "own" the keywords; i.e., they are not written to the image file, or, at least, not in an application independent fashion. When the application dies some years down the road, then the hundreds or thousands of hours spent keywording are gone. If images depended on the application for the underlying folder structure / file system structure finding images could be a long term nightmare. Even Polaroid died.

My question is: Does there exist a program(s) for under $300US that meets these criteria?

1) - A modular, open source component for reading new file formats; i.e., plug-ins available from anyone. Don't Want to wait 3 years to read CR4 files or buy an upgrade.

2) - implements all we know about adding keywords and virtual filing structures while working on top of standard OS file systems..

3) - Stores keywords and any other search metadata with each image file (it can maintain its own database as long as the data is duplicated in an industry standard fashion with each image.)

4) - Is able to handle file renames and folder re-orgs done outside the program.

Item 4 is probably asking too much but everything I've seen so far is lousy at straight OS style file management. But still, the issue of splitting the image base across 3 physical drives and reorganizing in the process needs to be addressed.

My apologies if this has already been asked and answered. If so, a pointer would be appreciated.

A coule of notes about the above post

JPEG and Tiff images can have the text embedded (file info)

So recataloging may not be an issue

The downside of this is that with large quantities of images searching becomes heavy

Photomechanic is less than $300US with great faciity for adding File Info

As above searching file info can get heavy

So there is a trade between attaching the DATA to the image or storing seperately

Incedentally FILE INFO does not seem to be 100% stable,

FotoStation 4.5 (old now I imagine) doesnt see file info added in CS2

I have seperated my image text data from the image for years

I started off holding it in a TXT file, then Exel then various MSAccess databases

I have found that it has been fairly striwghtforward (for a bit of a tech head) to import tho older records into the newer systems

What scares me is software like Lightroom where (I dont use it) the data seems to be held in a very proprierty manner

S

Mr. Moore,

Please note that I see some metadata as needing to be duplicated. The data that is used for search, ratings, sorting, etc. ought, IMO, to live both in an index and in the image file. That way the index can be built or rebuilt by any piece of software as required.

Given the size of images now, a few to a few hundred extra bytes won't hurt.

Yeah all my edits have file info added - that info is then exported to a text database

I used to do this with fotostation pro - (now I use homemade JAVA scripting part of my 'stillsmonkey' software that generates thumbs, displays online etc in one hit but thats a different story..)

(there are issues that the two sets of data can drift apart - ie edits on one dont automatically pass to the other)

S

I name my files in with the following discriptor:

yymmdd-Info1-Info2-Sequencenumbers.Filetype

So I could have

080308-Studio-Me-01.NEF

These would all be in a folder 080308-Studio-Me.


Versions of images would then become;

080308-Studio-Me-01-Master.tiff
080308-Studio-Me-01-Master-10x8r-Web800.jpg


Using a mac and the search facility it becomes easy to search for files if they become misplaced.


However then all the files go into aperture and are tagged with keywords and some with captions which provides a very powerful search asset.

I use a slightly different naming convention. Filenames are a simple YYYY-MM-DD-XXXX with Y = year, M = month, D = day, and X is a 4-digit shot counter for the day. The folder the shots live in is named YYYY-MM-DD + a description of the shoot, such as "2005-10-15 Devereaux Wedding" or 2001-05-12 Grand Canyin North Rim" or "2004-01-01 New Year Party" or similar. So instead of keywording each image, which is a big hassle, I merely put the keywords in the folder name. All I have to do is do a file/folder name search for "Grand Canyon" or "Devereaux" or "Party" and I can find the shoot very quickly. And I don't have to worry about losing the keyword database if I switch to a different app or OS or whatever.

Then the amount of keywords you can add is restricted to the folder name lenght

'Keywording' is different from 'describing' too..

'Keywording' is formalising the choice of word - maybe 'categorisation' is that formalisation

Either way 'formalising' the structure can be very helpful

(Or not - that is a different debate)

Formalising allows for attaching a series of alternative words to be attached to the keyword/category and then produce a result when included in a free text search

Eg

KEYWORD:FEMALE
SEARCH STRING: woman girl female

KEYWORD:BEACH
SEARCH STRING: coast beach seaside

You will see that some image search systems fail

ie a search for

'female at coast' will not bring results where a search for 'woman on beach' will produce reults

It is my experience that only 'database driven text software' as opposed to 'picture software' will produce robust reults of that nature

I am therefore of the opinion that the most robust method of storage/archiving is to

-use date folder structure
-add basic info to each file in file info
-export the information to a text DB
-add further desciption, keywords, sale history etc in that DB

the first and second build in a certain reduncancy against OS changes etc

I can view my images and generate text reports of images by..
Date
Commissioning customer
keyword
text search
most viewed
most downloaded
most purchased
etc



S

OTOH, my system requires perhaps 30 seconds per shoot to name the folder intelligently, no database creation or maintenance is required, and so far I haven't had issues finding images. But I only have about 130,000 images in my archives.

indeed the system needs to scale for the organisation

I have had staff for the last decade - staff who have left and new staff who have arrived - they dont know what is there or what they are looking for !

S

I manage about 30,000 active files, and my backup (archive) systems are the same as the active system, i.e. direct copies updated daily. I never allow any software to manage my files, only to present them for editing etc. When any current folder's files need renaming or other reorganization, I use command-line "batch" files plus a variety of text tools to automate the tasks. For example, if I have a series of jpeg's named 'Bird_Robin....' and 'Bird_Sparrow....' and I need to change the format to 'Bird_Perching_Robin....' and so on, the text tools simply make a left-right list of filenames and insert the '_Perching' text on the right, then output the result as a batch file that renames the appropriate files. Getting used to doing things this way doesn't appeal to many people, but it works 100 percent and eliminates all those dependencies and conflicts.