For a
photographer who works with photo banks, high-quality images are half the
success. In order for a photo to be commercially successful on stocks, it must
be properly “packaged”: the correct title must be selected for it and relevant
keywords must be invented. In the language of drains, this is called
"attribution" of the image. Without proper attribution, your picture,
no matter how professional and fresh it may be, is simply lost in a huge mass
of stock images.
You need to
know a few simple rules, practice in order to "get your hand", and
gain time and patience, because this work is painstaking and time-consuming. In
addition, today there are many services that help automate this work. But
ultimately, it's up to you to work with content on your own or to give a
routine to neural networks.
Explore Buyer Behavior
To feel and
understand the behavior of the buyer of the photobank, try yourself to look for
something there. Imagine that you need a wallpaper with a landscape on your
desktop or a cute pattern for printing on a t-shirt. Now formulate the request
and see what the algorithm gives you. Then change it and track what has changed
in the extradition. When you find the very image that you imagined in your
head, pay attention to its title and description. Perform this exercise
periodically to train your client’s thinking and better understand how search
works.
Correct title
Now let's
move on to the description of the content. The title of the photo is the most
creative and at the same time the most difficult part of attribution, because
you must express the whole idea of your work, the main idea in 3-7 words, not
counting the articles and prepositions. If there is a point of interest in the
picture or we recognize the city in which you shot, then indicate this in the
name. But no need to rush to the extreme and write the exact location address.
Good
examples: Driving along the Golden Gate Bridge // Girl on the seashore //
Keywords
Next, we
select the keywords. This process is also called "keying" or
"keyword". Your goal is to describe the image as relevant as
possible. It’s best to use between 10 and 35 keywords.
First, list
the objects that are shown in the photo, and their properties. Feel free to use
synonyms: different people may come up with different names for the same
phenomenon. But don’t overdo it: if it’s “palm,” then “brush”, “hand”, but not
“hand”, are perfect as keywords. Do not neglect the numerals and words that
describe groups of people (family, crowd, group, class, team); names of
professions, occupation, mention gender, age, nationality, social status, if
attention is paid to this in your photo. Describe the circumstances in which the
photo was taken: background, posture, color; this room or open space; time of
day or time of year. It is very important to describe the action if it is
present in the image, so do not avoid the verbs: if the person is in the photo,
Use
adjectives, including those with a negative meaning. Most of the photos on the
stock portray positive emotions, but users are looking for a variety of
content! And let you have fewer words, but they will be in essence.
Description
Example
Title:
Shopping
Description:
Girl, shopping, shopaholic, packages, shopping, trade
If we talk
about illustrations or texture photographs, patterns, then there is a trend:
most authors of photobanks really like the word "abstract", and also
"grunge" and "bokeh". Do you think the buyer will find your
work among several hundred thousand images with such a description?
In the case
of abstract images, the advice is simple: pay attention to the details,
determine the style of the picture, the prevailing colors and shades, the presence
of halftones or gradients.
Description
Example
Title: Wood
texture
Description:
Aged surface, wood, trunk, bark, blue and brown, beige, pattern, daylight,
centuries-old tree
Man vs Neural Network
There are
many tools that can speed up the attribution of content at times. The simplest
thing is when this process is outsourced and given to specially trained people.
The next level is the selection of keywords by the semantic core. You enter one
word, and the system selects a series of related by meaning or synonyms.
But in the
XXI century, it looks a little outdated, and the trend is still towards process
automation. Today there are services that generate words using artificial
intelligence. For example, the mykeyworder service , which can be used both in
the web version and as a plugin for Lightroom. Auto keyvordinga there and the
bank Shutterstock , and in late May the service of the company has launched a
content attribution Everypixel .
Surely there
will be ardent supporters of the human mind who will say that a soulless
machine will still not see what a living person sees. May be so. But the
numbers suggest the opposite: as a result of the experiment, which was carried
out over the course of a month, it turned out that the photos that were processed
by artificial intelligence sold better . 100 photographs processed by a man and
100 photographs processed using automatic keyword loading were uploaded to
photobanks. For the purity of the experiment, similar shootings were taken. Of
the first sample, we managed to sell 23 photos, of the second - 36. And another
important argument in favor of artificial intelligence is time. A person can
describe 1 photograph in 1 minute, and a neural network - 1000.
Given the
fact that technology is developing, there is no doubt that these services will
work better every day, so even if you still think that you are better at
attribution, do not neglect automatic solutions. In the end, instead of
spending a thousand minutes describing the images, it’s better to do more
creative and inspiring tasks - to shoot and process, for example.
Comments
Post a Comment