One of the biggest problems in the stock photo business is search. As the most popular search engines gobble up more and more images it has become harder and harder for clients to find the best image for their projects. No buyer has the time to review even a fraction of the returns from most keyword searches.
Many sellers think the answer is better keywording, but there is a limit as to how well one can describe the look and feel of an image with keywords alone. In most cases 99% of the thumbnails delivered as a result of a keyword search are not even close to what the customer needs. Usually the same keywords will return a set of images that vary widely in visual characteristics, quality and usability.
Traditionally, experienced editors have searched collections and prioritized images they believe are most likely to be in demand into a “Best Match” category. But this is costly in human time and difficult to keep up with given the volume of new images to consider.
The microstock companies have always offered a “Most Downloads” search option. In this way they use their customers as image editors. Based on downloads, rather than digging through a collection to find something no one else has ever seen, most customers seem to be perfectly happy to use images found by those who went before them. But, this method tends to return a lot of the same images over and over again. Images that would be just right for the customer are often hidden.
Getty recently added a “
Most Popular” option to its search. We assume it is based to a large extent on sales. Most other traditional sites still present images to customers in a single default order.
In addition, Getty has recently selected images from various brands and moved them into its Brand X collection. Then they pushed this collection higher in the search return order. This seems to have improved sales for some contributors. But most companies are looking for technological solutions to the problem.
Shutterstock Labs is testing several new search techniques with buyers.
With
Shutterstock Instant the user not only gets a look at the subject described in keywords, but a small selections of ten different sub-categories of the main category searched. Choose any one of the sub-categories and the searcher gets another 10 sub-categories. This enables searchers to quickly navigate the collection in a very different order from the standard search.
Shutterstock Spectrum starts with a keyword search and then allows customers to look for images within that search that fit a particular spectrum of color. The customer uses a slider to find the particular color needed.
Shutterstock Palette allows the user to choose up to five colors in a color palette, and show only the images with specified keywords that include colors in the palette.
With
Shutterstock People users can narrow their search by specifying the number of people, 15 different ethnicities, age and gender. These four categories are fairly common on many sites, but Shutterstock’s version may go into a little more depth in terms of options. Of course, this relies entirely on keywording. If the appropriate keyword hasn’t been added then the search will not find the image.
Last month Shutterstock introduced
Sequence an in-browser editing tool designed to provide an easy way for anyone to create and quickly share videos with colleagues and clients. Sequence allows users to seamlessly integrate their own footage and music with anything in Shutterstock's 2 million video clips and music tracks collection without downloading anything. Thus, customers can test the usability of a particular clip before they buy.
Pixolution, a German company, has developed a variety of visual search tools that allow customers to narrow their searches based on the visual characteristics of the image. The customer starts with keywords and gets a best match result. Then the customer chooses an image that has some of the look and feel desired. By clicking on a “Find Similars” and adjusting the slider in the direction of images that are more or less “visually similar” the customer can quickly find all the visually similar variations in a large collection. This is particularly useful when lots of similars from the same shoot have been uploaded.
Pixolution’s demonstration search (http://demo.pixolution.de/) looks at the Fotolia collection. Among the other sites that are currently using pixolution tools to some degree are:
living4media.com; framepool.de; Yayimages.com; clipdealer.com; imaginechina.com; mainstreamdata.com; belgaimage.be; Newscom.com; graphicobsession.fr; photononstop.com; stockfood.com; peopleimages.com; f1online.de; motionelements.com and shotshop.com. The
living4media site is particularly interesting in the way it uses these tools.
The various features Pixolution offers are:
Visual similarity - Arrange hundreds of images according to their visual similarity on the fly.
Similarity Weighting – Makes it possible to fine tune balance between textual and visual relevance of an image search.
Multi color - Find images that best match one or several colors.
Visual arrangement – Arranges hundreds of images on the fly according to their visual simularity
Auto Context - Automatically find images of the correct semantic context. Keywords associated to the query image are automatically involved in a combined textual and visual search.
Text space – Finds images with available Copy Space
Query with web images – Use any image found on the Internet, or other source, to locate a similar image in the agency’s collection.