Photo by Radek Grzybowski / Unsplash

File naming, image tagging and the arrival of AI, machine-learning & Sensei

In this post I tell you a little about my file-naming protocols, the ways I manage my photo data and how I’m looking to the future of AI & machine learning in catalogue searching and image tagging, with Adobe’s new ‘Sensei’ product.

It was 2004 and I was working for my local council in Monmouthshire, Wales, when I learned a new word: taxonomy. At the time I was both running my own web-design and hosting business and working for my local authority, managing the last phase of transition to live of their new website. Local government and the public sector is an awesome place to learn new words and acronyms. Everything has an abbreviation. Everything has a language of its own.

Taxonomy is simply the practice and science of classification of things or concepts, including the principles that underlie such classification. In my life then, it was all about taxonomies for classifying, filing, publishing and building inter and intranet repositories of data. Now, it’s about filing all my own crap so i know where it is and can find it easily when people contact me wanting to use a photo.

To tag or not to tag?

Like all posts about my workflow and working practices, I’ll preface this whole spiel with the usual disclaimer: these are notes about my workflow, about what works for me. If you spot something that you think may work for you; lovely. Bingo. If you don’t, fine. If you think I’m talking a load of bollocks, that’s fine too. Just please don’t start posting comments below telling me your way is better. Because I’m not telling you MY way is better. I’m just telling you what my way IS.

So, tags. I started out in photography a long time ago and when digital came along, I soon developed a workflow that worked for me. That didn’t include Lightroom, which is great for tagging and metadata. It didn’t include Lightroom because Lightroom didn’t exist at the time.

To go back, now, and retro-tag all my work would take me forever. I’m waiting for machine learning and AI to make its way into photo software to the point where it can look through all my data and organise and tag it for me. Such software may exist already. If it does and you know about it, do let me know.  I just haven’t bothered to tag my files. It’s time consuming and although LR allows for tag input on import, I don’t have LR as an integral ‘everyday’ part of my workflow.

I’m not even tagging my new shots because, as I’ll mention at the end of this post, Adobe is on the verge of integrating AI and machine-learning into desktop products to do it for me.

Why don’t I use Lightroom?

Despite having the latest version of Adobe Camera Raw (ACR) installed on my machines, I do such a lot of in-camera styling with my Nikons, which Lightroom doesn’t read, that I have never had a ‘one RAW converter fits all’ solution in my workflow. You may say that using Nikon’s ‘Picture Control’ system to style my files in-camera is backward, given the time-saving I could get from using Lightroom at the centre of my digital workflow. I don’t see it like that. I have another post on this site that details why, despite always shooting RAW,  I prefer to style a lot of my work in-camera. For the full explanation, please refer to that.

So I don’t use Lightroom because it cannot read the Picture Controls I use in my Nikons and therefore when I import RAW files into Lightroom, the software tuns everything back into a ‘standard’ colour setting, losing all the styling I applied to the RAW file in my Nikon DSLR.

I also shoot with the Hasselblad H4D. If any styling is required for those shots, I can only do it whilst tethering the camera to the Phocus software. That allows me a lot of control over the style of the RAW file. Hasselblad doest yet allow me to upload  back to the camera a style I created in Phocus. I’m working on them as regards this functionality. I think being able to shoot in ‘film’ styles is a nice mindset, very similar to the pre-defined aesthetic style of shooting we did with film. I’m hoping that Hasselblad come up with a solution for this. Being an Ambassador for them helps. I have a direct line to the people there. They have a full plate of things to do, so it just takes time.

If I don’t tag my shots, how do I find anything?

Particularly after working in the public sector, I got into the practice of just getting clever with my file-naming protocols. That’s an approach I use still.

My file naming taxonomy starts with a top level: the name of my hard-drives.

Drives that tend only to ever live at home are called AG-0001, AG-0002 etc etc.

Drives that leave the office for one reason or another have a different name: ‘Portable-0001, ‘Portable-0002’ etc etc.

I know my systems. But if someone else is working with me on a particular job, an easier way of identifying what is what is necessary.

Next comes a top-level folder in each hard-drive. There are different names for personal work and client work. Personal work goes into folders with the following protocol:

‘photolibrary-SequentialNumber-StartDate’

Client work goes into folders with this protocol:

‘photolibrary-clients-StartDate’

Inside those top-level folders, the taxonomy is as follows:

Personal work folders go something like this:

General days of shooting, where I didn’t specifically stay in one area or have a rigid theme:

D800E-walk-DayMonthYear

or

H4D-walk-DayMonthYear

If there’s been more of a theme or one specific lens I’ve one out with, then it’s something like this:

D800E-RB67_127mm-DayMonthYear

or

H4D-ShibuyaHikarie-DayMonthYear

Client folders are arranged like this:

ClientName_JobName_DayMonthYear

Inside each client folder, there may be multiple folders, either because the shoot is over multiple days or where different venues are involved.

Typically, there are also multiple cameras used on client shoots. So, inside the job or venue folders there will be folders called ‘Nikon’ and ‘Hasselblad’.

i.e. Heinz_Baked_Beans-Tokyo_20thJune2017 > Day01 > Nikon > RAW >ALF_0001.NEF

How about the filenames and any edits?

Individual filenames are dictated by the camera file naming system, which I have personalised to ALF_0001.NEF or .JPG

If I do a run of batch re-sized (1280pixels on the long side) small JPEG versions, to show the client and find out which shots they want me to edit, it’d go something like this:

heinz-beans_TokyoJun2017_ALF_0001_small1280px.jpg

Once we get to the editing stage, each edit on my side (as there may be a few things I try before I reach the final one I or the client likes) would go like this:

heinz-beans_TokyoJun2017_ALF_0001_a.TIFF

With version two of the edit being heinz-beans_TokyoJun2017_ALF_0001_b.TIFF

But how about if someone finds a shot of yours online and wants a high-resolution copy of it for print?

Most of the stuff I put online, on my own sites or elsewhere on the internet,  is 72dpi and typically no bigger than 1920pixels on the long side.

If someone finds a shot and sends it to me asking for a larger, higher rez version for an article off the shelf, then the file naming protocols I have mean I can find the folder the original files live in pretty quick, usually within 5mins.

All web versions of my work are named something like this:

www_hikarie-ALF_3329.jpg

I’d take the www_ off the front and search and that would lead me to the large edited versions which would, in turn, lead me to the original RAWs.

The blind-spot in all of this?

If someone wants a collection of ‘green’ shots or ‘fish’ shots or ‘shots that have a dog or bicycle in, that’s tougher. The folder I have locally (and on my iPhone as a sort of mobile portfolio), that contains all the stuff I have ever put online on my websites or elsewhere, could be searched for clues. That folder contains sort of ‘the best of the best’ of my online work. Out of that there would be great ‘green’ shots, great ‘bicycle’ shots, great ‘train’ shots. I find one, which leads me to a specific folder, for a specific day in the past when I was shooting trains or bicycles. It’s not a perfect system but it’s a good start.

Once I’ve found one folder that has some candidate images in, I can remember the sorts of things I was seeing or shooting on other days near that day.

For me it’s all about balance. Balancing time doing one thing versus another. I can spend all my spare time tagging or I can spend my time shooting, editing, blogging and just have a decent file-naming protocol to get me 90% of the way to finding any image I need to find, in a sensible amount of time.

Plus, with the advances in AI and machine learning, spending loads of time tagging now would be a waste because something cool is on the way to take care of the problem for me.

Adobe Sensei.

Artificial Intelligence and machine learning are bringing all sorts of innovations to life at an accelerated pace. I’ve heard key people at Facebook, Google, Microsoft and Adobe speak this year about some of the things they are doing in this realm. The next 12months is going to be very exciting.

One product that could, I hope, be helping me catalogue my images and add keywords is Adobe Sensei. Announced in 2016, Sensei is Adobe’s take on the whole AI thing and is something that they say as it the heart of their Creative Cloud and that has already been making its appearance in the world of PDF repositories and has been touted as a way of making your selfies look like they were taken with a better camera or lens. 

The selfie thing doesn’t interest me but it’s exciting to note that Adobe Stock is already using Sensei to search for pictures based on mood and aesthetics.  For me, the exciting part will be when Sensei becomes part of desktop products like Adobe Lightroom or Bridge and can trawl my entire collection of locally stored images, adding keywords based on the colours, content and even the mood of them.

For now, I’ll just keep the taxonomy simple and wait for my Sensei to come to my rescue.

Find out more about Adobe Sensei here at Adobe.

https://www.adobe.io/apis/cloudplatform/sensei.html

Read a press release about Sensei here:

https://www.adobe.io/apis/cloudplatform/sensei.html

Adobe Sensei on Twitter

https://twitter.com/AdobeSensei