One of the biggest problems in the photo world today is that we are being buried in photos.
InfoTrend estimates that consumers will take 1.2 trillion photos worldwide in 2017. The compound annual growth rate (CAGR) is 9%. This year 3,934,500,000,000 will be stored on hard drives and other formats worldwide.
Mylio.com has estimated that 5 billion of the 7.5 billion people in the world have mobile phones and that roughly 80% of those phones have a built-in camera. Thus, if each person took an average of 984 photos a year, or 2.7 a day they would reach the InfoTrend number.
Photographers trying to license rights to their photos can leave most of these huge numbers to the consumers to worry about. If they can’t find their own photos that’s their problem. However,
PicturEngine.com tells us that they have collected over 900,000,000 unique photos from various stock photo agencies and individual photographers that are available for licensing. How does anyone find an image that can work for their project among such a huge number?
Of course, part of the answer is “accurate keywording.” But, even if “totally accurate” keywording is possible, it is difficult to find enough unique words to apply to any image to easily separate it from hundreds to thousands of others with the same words. The Oxford English Dictionary contains full entries for 171,476 words in current use. It is estimated that a college-educated English speaker has a vocabulary of between 15,000 to 20,000 words. That leaves, on average, between 45,000 and 60,000 images with the same word or words.
And even then, if you’ve keyworded and image with a bunch of esoteric, but accurate, words that people seldom use, the image may not be found because the searcher didn’t use one of the words you chose.
On top of this it is now common to submit 10, 20 and sometimes 50 to 100 very similar images of the same situation to collections. This forces the person looking for images to scroll through a lot of images that are of no interest to get to the next image that is really different. No one has time for this.
Apple Acquire Regaind Photo
Apple has acquired Regaind Photo.
Regaind proposes to use “state-of- the-art artificial intelligence to analyze and sort” photos. They point out that the end users problem, “is often not reduced to finding a photo of a « sailboat » or of a « lion »: it’s also about finding the right one for your specific use, among many others. Regaind enables you to understand the content of an image, as well as to assess its technical and aesthetical values, so as to maximize your impact with high quality photos.”
Among the feature they will offer are:
Face Analysis: Detect faces on the photo, as well as the gender, age and emotion of the people that appear.
Main semantic class and label detection: Understand what the photo is essentially about (portrait, landscape, animal, object…) and get detailed labels on what is depicted: environment, objects, action, emotion (+3500 labels).
Region of interest detection: Detect where the attention goes on the image and generate automatic smart crops that preserve the main subject
Aesthetical and technical attributes evaluation: Assess the technical quality (exposure, dominant colors, sharpness of main subject) and the aesthetical quality of the image.
These steps may help the average consumer to analyze his or her own relatively small collection, but they are unlikely to do much to assist in dealing with large professional collections.
Changes Needed
First, the industry should make better use of the knowledge of image buyers who are constantly reviewing professional collections. Data on images actually licensed needs to be shared.
Searchers are making daily decisions about which images work best for their projects. Often the exact same images used by others are exactly what someone else might find useful for their project. For many customers being able to search a collection of images that others had found useful for their projects would greatly improve the efficiency of the new customer’s search.
Second, a system should be devised to show only one of a collection of images from the same situation created by the same photographer. A notation beside the primary image would indicate the number of similars available. If the customers is interested in a particular subject they could click on the notation and see all the similars. This would speed the customer’s search process and allow him/her to quickly see a greater variety of options.
Agencies do track “clicks” and “views” and organize future searches based on such interaction, but this information can be misleading. Customer may click on a complex image simply because they can not adequately review the detail in a thumbnail size. Once reviewed they recognize the weaknesses and move on to something else.
When customers don’t have to pay for each image downloaded (subscriptions) they may also download images just for reference purposes with no intention of ever using the image in a final product.
It is much more important and useful to know which images other customer have been willing to pay money to acquire.
Some will argue that revealing sales information gives proprietary information to competitors. But the benefits it supplies to customers far outweighs any loss of proprietary information.
When collections were much smaller it may have been wise to protect the information about which images are actually selling. But given the size of collections today, and where they are headed, much more needs to be done to make it easier for customers to find useful images.
No matter how cheap the license, if customers can’t find what they need in a reasonable period of time they are going to move on to some other way to get what they need.