A new search engine called
PicturEngine is getting a lot of attention in the blogosphere. It has recently been released in Beta after four years in development. The founder, Justin Brinson, also owns two agencies: Corner House Stock Photos and Picturesque.
The aim of the search engine is to provide image buyers with one-stop shopping to all the professional images available for licensing, worldwide. PicturEngine is NOT an agency, rather it is a place where photographers and agencies can advertise their work. When a customer finds an image they like they are referred directly to the agency or individual photographer who placed the image on the site. The customer deals directly with the image supplier, the supplier sets the price for usage and retains 100% of the fee collected.
Advertisers pay a modest annual fee to have their images included in the service. There are several levels of participation. While the site is in photographer Beta individuals will pay $20 per month, and no additional storage fees, to place an unlimited number of images on the site. There is no indication at the moment when the site will go live and start marketing directly to customers. For those who sign up during the Beta period this monthly fee rate is locked in for the future. Agencies with more than 1 million images may participate for free. Agencies with less than 1 million images should contact PicturEngine to determine the fee they would be charged.
To date Brinson claims to have over 300 million images in his database from more than 50 major distributors that include: Getty Images, Corbis, Alamy, AGE, Masterfile, Superstock, ImageSource, Fotosearch, Veer, Aurora, iStockphoto, Shutterstock, Dreamstime, Fotolia, Bigstock, Jupiterimages, Thinkstock, Depositphotos, Yay, 123RF and pixmac. After sorting and filtering out duplicates (many images are represented by multiple sites) there are about 200 million unique images in the database.
Among the key advantages from the individual photographer’s point of view are:
1 - PicturEngine allows each contributor to set the price for use of his/her images, deal directly with the customer and retain 100% of the fee collected.
2 – The contributor can use PLUS Packs to offer customers online pricing. Sales can be transacted without the necessity of the contributor being constantly available.
3 – When an image is represented by both a contributing agency and one of PicturEngine’s individual contributors PicturEngine favors the contributor and sends the customer to the contributor rather than the agency. Thus, photographers whose images are represented by one or more agencies non-exclusively can place the same images on PicturEngine and eliminate the agent commission in those cases when a PicturEngine customer chooses one of their images. Potentially, this could be of great benefit to photographers who are unhappy with the prices their agencies are charging for use of their imagess.
There are a host of other issues photographers need to consider before jumping on the PicturEngine band wagon. One is the size of the collection (200 million images and counting), and the idea that it is possible to order search returns so the first images that come up will always be the best for the buyer’s purposes. (Every photographer believes that will always be his image.)
The number of images available for licensing is growing at a dramatic pace. Buyers must sort through all these options to find the one that works best for their project. Brinson says that in the last few years he has received over 16,000 responses from buyers to surveys and their chief complaint is that they see the same images on all the sites.
PicturEngine uses a visual search algorithm that identifies images in its collection that are exactly the same. When it finds the same image on two or more sites it will only present the image that was uploaded to the site first to its customers in search returns. The customer will have no way of knowing that the image is also available on many other sites. As we discussed earlier, the only exception to this is when the individual creator posts the image on PicturEngine. Then the customer will be directed to the individual creator and not to any of the creator’s agencies or distributors.
While eliminating duplicates is an important first step this still leaves over 200 million images. To narrow the search the customers must use keywords or phrases. Unfortunately, words alone usually won’t define all the unique characteristics that cause a customer to choose one image over another. Even if it were possible to define all the unique characteristics the customer would have to use all the same words in his query to sort out the one image from the pack.
If it were possible to insure that the best would come up first, or even in the first 10 or first 100 then this system might be a boom to both buyers and photographers. But given the number of images in the collection it seems unlikely it will work that way.
Let me illustrate the problem with an example. Suppose the customer wants a picture of a “Lioness and cubs with a fresh kill.” I searched Alamy.com using this phrase. Alamy has a large database, but it only totals 30 million images, not 200 million. Alamy delivered 143 returns from this search.
Three images have the following keyword:
East Africa, Masai, Lion, cubs, feeding, antelope, kill
Lion, lioness, panthera, leo, cub, kitten, zebra
Lioness, eating, Zebra, prey, carcass, safari
Based on this subject matter and these keywords how would you decide which image is best? Some of the images in this return showed a male, not a female lion. However, the key to any successful photo is how the elements fit together in the image. That can not be easily described.
Another dilemma is with variations on a particular scene. In the scene above each photographer might have shot 20 or more frames from different angles using different lenses. All images in the sequence would have been keyworded using the same words. How could a search engine possibly know which image in any sequence would be the one to satisfy a given customer’s need? From the creators point of view he’ll want to show all the images in the sequence in hopes that one will appeal to the customer. If there is no editing he’ll put them all on PicturEngine increasing the total number of images to be searched. As more creators do this it will become less likely that the customer will quickly find the right image. Customers may decide that it is more efficient to use sites that do some editing.
Another factor to consider is that we don’t know how the images from a given contributor will be ordered in any particular search return. Will all the images from individual contributors to PicturEngine be shown before agency images are shown? Will RM images be shown before RF and both shown before microstock? Some photographer’s call for a “level playing field” but that is impossible. We can never know what the customer will want until he/she sees the image. With so many images to choose from on virtually every subject someone’s images must be shown first. At some point the customer will get tired of reviewing “all” the options and settle for something he/she has already seen. Those shown early in the search return will always have an advantage.
Will the total number of images a photographer has in the collection have any bearing on how early his images are shown on any given search relative to those of other photographers? Will all of one photographer’s images be shown before those of any other photographer? Is the image shown first based on how many times a preview of that image has been opened? (This can produce false results because the preview of an image that has a lot of detail may be opened many times because it is difficult to “read” in thumbnail size image and yet no customer may have ever purchased that image. It also gives an advantage to older images that have been in the database for a while, relative to new images.)
Brinson said, “Finding the "best" image is no easy task, and YES it is somewhat subjective. Some of the factors we use for determining what a client is looking for is that client's individual searching patterns and habits.” If customers log in before they begin a search then PicturEngine will track their searching habits.
He continued, “Our smart system remembers your settings including images you regularly search, rate, place in lightboxes and purchase, to provide you the best search experience possible. With PicturEngine, the more you search the better your results. Just as Google records your search patterns and habits to better serve you targeted advertising, we do the same, but instead of side posted advertising, we rearrange your results according to what you like seeing, sharing and BUYING. Our image results are individually rated, ranked and weighted by overall quality, quantity, concentration, image hovers, click throughs and sales (if sales are known or licensed on our platform). Each of these actions has a different weight. I completely understand that some of these benefit older images (image hovers, click throughs and sales), so we have made the appropriate resets and adjustments for time. The more a user searches, the more we understand their individual needs and the better their individual results become.” Over time, customers might see an advantage if they always search for the same thing, but one wonders if customers will stick around long enough to see this advantage.
“Criteria we take into consideration include, but are not limited to:
Agency and agency sims;
Photographer and photographer sims;
Individual image overall quality, quantity, image hovers and click throughs;
Groups of visually similar images industry wide;
Groups of keyword similar images industry wide;
Individual search terms are also compared to keyword, caption, description when determining relevance;
AND MUCH, MUCH MORE!!!!” Brinson added.
The dilemma is that given the way search engines work the more images in a collection the less likely any individual’s images will be seen and the less likely the customer will find the right image quickly. Technology still hasn’t devised a way to solve the problem of too many images.
For more information check out:
http://support.picturengine.com/