EyeEM – making it better!

I wrote earlier about my experiences of what sells on EyeEM, and since then I have been thinking of how it could be improved. Of course, some of these things are only developments that EyeEM can make, but if they are not described then nothing is going to happen! EyeEM sells in two main places – one is their own site, although it has been a while since I got some direct sales there. The main source of income for me has been their partnership with Getty.

The basic workflow is that you can upload your images to EyeEM and their system imports any embedded keywords. The location, if embedded is also added, although you can also type this in directly, as long as the location is recognized by Google. I found that commonly used words like Corfu in Greece were not recognized although the official name of the capital city (Kerkyra) is. I’ll come back to that! There is a bug that applies the same description to all the images in the same upload batch, which EyeEM don’t seem to think is important. Those keywords and descriptions are used in searches on their own site and so are important in getting direct sales.

Their team then examines each image, sometimes asks for model or property releases and selects those that they are going to upload to Getty. You see those in the Partner section of your portfolio. Currently I have 1500 images, 1400 for sale on the EyeEM site and 888 in the Partner collection. What happens next is supposition on my part. I believe that they analyze the images using some sort of Artificial Intelligence program to come up with a title, description and keywords for Getty. They don’t appear to use any of the information you have carefully added, except for that location. So entering that accurately is really important for travel photographers. However, I couldn’t add Corfu, Greece (which I am sure is what most people would search for) so I had to add Corfu, USA!

My guess is that these newly keyworded images also align with Getty’s controlled vocabulary and are then uploaded to that site. As I currently have 406 images in my portfolio there, there is probably both a selection process by Getty, as well as a timing delay before the images are uploaded by EyeEM. However, there are a few images I found on Getty that I uploaded to EyeEM just a month ago, so I think this is more about a selection process than a delay. Their AI keywording is pretty good in some circumstances – it recognized the Acropolis, Parthenon and Lycabettus Hill in Athens for this shot:

But this one below correctly shows the fact it was taken in Athens, but not the neighborhood of Anafiotika, which I think is important in someone actually finding it.

When I search for Anafiotika in the main Getty library I find 32 pictures of this neighborhood. So it is obviously an acceptable word. Interestingly enough, the word is not found in the keywords themselves – but is in the title of the images I checked. What this means is that my carefully uploaded, keyworded and curated image has very little chance of being found under a normal search on Getty. There might be a workaround if the location Anafiotika is acceptable in the upload process – I will have to try that next time!

The other thing I found with this AI approach is that conceptual images are described for what they are:

But the actual concept I was trying to illustrate – the new year and move to 2020 – has been lost. So, again, the chance of finding this for a new year image is zero.

Why am I writing all this? Partly to set your expectations about what will sell on Getty via EyeEM. Partly to help with travel images by carefully setting the location to what you think people might search for, and partly in the hope that the suggestions I am going to make are seen and implemented by EyeEM! Perhaps if we all pointed their support team at this article it might help!

So what could they do? Firstly, they could transfer the description of the image that we carefully write into the uploaded file for Getty. Here we could describe the concept, location and key facts about the photo. It appears that Getty searches this for returning relevant files and so this would be a major step forward. They would have to fix the bug that applies the same description to every file in an upload group though!

Then they could take the first 3 keywords that we enter and add those to the AI generated ones they apply. Perhaps we would have to use the iStock controlled vocabularly in those keywords which would be a pain but better than nothing. In those three keywords I could put the concept – 2020, New Year etc. and be sure it would transfer to Getty.

Finally, they could provide a way to edit keywords (or description) on the Getty site. A friend provided me with a link that appeared to do that. I have an image of San Diego on Getty that is wrongly described as Chicago, but although the provided link appeared to allow me to add up to 5 keywords and delete up to 5, after at least 4 weeks of waiting it hasn’t made a difference to the search for this file. So although it didn’t work for me, it appears it was considered at one point.

Can you think of other changes that would help us out and not add to the EyeEM workload? Please add in the comments below!

Edit – added 29 August: Of course there is one other option – add an FTP import option and tie it into the Stock Submitter application. Then we could use Stock Submitter to correctly write the description, use the iStock controlled vocabulary to add appropriate keywords and then EyeEM could take the description and first three keywords for their own input to Getty. Problem solved!