Saving Your Search Settings

There have been some discussions on iStockphoto.com lately from buyers who wish not to see the results of the new premium Vetta stock image collection in their search return.  A perfectly valid request, but they find they need to take the same steps over and over again.  However, you don’t have to…

In the “Advanced Search” settings drop down box (link at the top right of the iStock page), you can set all kinds of interesting things to help you search.  Including a filter to deal with Vetta: “include, exclude or only”.

Most people, myself included many times, will just set that Vetta setting and then go down to the bottom right of the settings area to click “Search”.

Then, next time they revisit the site, Vetta will reappear in the search results.

Instead, or as part of your workflow, click the “Save” button, underneath “Save these settings”.  Then, all your settings will be applied when you do a regular search from the search box in the upper left of any page.

Don’t forget it’s there though, lest you are wondering why you can’t find images with people, because you checked off the people filter yesterday, or all your returns are horizontal images.

Have a great weekend!  Portfolio update coming Monday…

Seasonal Reminder

With our favorite fall holidays coming up, I just wanted to remind everyone about the Seasonal Search page at iStockphoto.

It offers quite a few main holiday images, and each links to a subset of searches for that holiday.

Keep in mind, though, that these are merely searches of the main collection, and not edited lightboxes or anything like that.  So the above link for “Witches and Vampires” is a regular search on:

(witches or vampires) and halloween

This is ok for most cases, but in some it gets a little tricky, with the iStockphoto controlled vocabulary and all.  So, I’d suggest using this as a starting point, and then drilling down by adding your own keywords to “search within” or to create your own seasonal search.

Don’t forget though, that you can also search through member created and edited lightboxes for specific themes as well to aid you in your buying.

Only 127 days until Christmas!

Don’t Pay For Plain Pixels!

“Don’t Pay For Plain Pixels”, otherwise known as “Buying smart”.   I’m talking about trying to keep your spending smart by not buying more image than you need, in particular when it comes to a white background.

Very popular on microstock sites, including iStockphoto, is the “isolated”, “isolated on white” or “cut out” image, such as the one below:

I’ve put a 2 pixel border around the image so you can appreciate the white area.  This image was originally shot on a Canon 5D, at the full pixel resolution of 2912 x 4368.  This is a smaller version, but the same aspect ratio.  By the way, you can find this image on iStockphoto here.  As you can see, I’ve filled the frame pretty fully with the model.  I’ve not added any more area than the camera originally captured, nor have I cropped any pixels from it.  I think this image provides plenty of “subject” real estate for use – ie. there are a lot of useful pixels in the image.

Imagine, however, I decided to shoot this horizontally, putting the model on frame left.  This might appear useful to a designer, with lots of room for copy or text on the right, but that is something easily added (I’ll show you how, below).

As you can see, if I uploaded the original, uncropped, straight from the camera version of this image to iStockphoto, while it may look nicely composed in the search thumbnail, over half of the image is nothing but white pixels, something easily added by a designer.  Essentially, you are paying for pixels that have no value to you.

Now, I’m not saying that subjects should be cropped within 3 pixels of the edge, but I think we can see where some cropping is necessary to ensure buyer value.  In fact, iStockphoto does reject some images with too much white space.  However, some do make it into the collection with too much white space in frame (IMO).

However, your design may call for white space in a certain area, relative to the subject.  It’s easy to add white space to your image using Photoshop (or any other program – the idea is the same).  For example, you could use the free online graphic tool “Phoenix” by Aviary.  It has the canvas function mentioned below, although the process is slightly different.

1. First step, open your image.
2. Under “Image”, go to “Canvas Size”.
3. “Canvas Size” will show you the current dimensions of your image.  For the clearest (to me, anyways) way to increase space, change the second pulldown for width and height to “pixels”.  You can now tell Photoshop the new image width in pixels.  Let’s say “400″.  But before we hit “OK”, we need to click on the leftmost, middle box in the bottom graphic.  This tells Photoshop, in the new, larger canvas, where to stick the old image.  We want it smack to the left, so we click the left middle box.  Any of the left boxes would work, since we aren’t changing the height.  Lastly, change the “Canvas Extension Color” to white, so the added pixels will be white.
4. Hooray!  We now have our new canvas, with plenty of copy space to the right of the model.

The point of this post is that you should make sure that you get the best value for the credits you are spending.  Given the choice between two similar images, see which one gives you the most useful pixels to work with.

Where to Look?

Just a few keyword terms (from the controlled vocabulary) for iStockphoto.com buyers who are trying to find photos with people looking in a certain direction.  The terms are pretty self explanatory, although some need some disambiguation choice to get the right option.  For example, “looking up” can apply to what the camera is doing, as well as what the person is doing.

• looking at camera
• sideways glance
• looking up (looking)
• looking down (looking)
• looking away
• looking over
• over the shoulder (looking over the shoulder)
• face to face

I’ll use the simple term “looking” as a synonym for “examining”, like if someone is checking out a sign, or looking through binoculars.

Have a good weekend!

Best Match 2.0 Awakens

Over at iStockphoto.com, ever since the Best Match sort algorthythm presented itself, we (the contributors) have been discussing the secret ingredients that make up the equation.  How does an image get presented on the front page of a search return?  Nary a week goes by without a debate on what is “best” when it comes to sorting results.

Now, the answer has arisen, and it’s name is “relevancy”.  Best Match 2.0 was announced back in December, but it has finally come to life with this announcement from Kelly Thompson yesterday:

After a long time in development and several months of internal testing, we’re happy to present our new default search sort: Best Match 2.0.

The algorithm takes many factors into account. One of these, keyword relevance, rates the importance of keywords in the files. Files can have up to 50 keywords, but normally only a few of those keywords describe the primary subject or concept.

We’re not quite sure of the magical mixture that determines a keyword’s relevancy for an image, but it does seem to work very nicely.  Even better, this attribute is adjustable.  Open up your advanced search options at the top of the page.  On the right, you will see the “relevancy” slider, which is only enabled if you have Best Match chosen as your search.

If the slider is moved all the way to the left, you will get a broad sort of the files returned from a search, where the relevancy factor is low.  In the following, I’ve searched on cheese, with a low relevancy factor of ‘0′.

Sure, all those things have cheese in them, and ‘cheese’ is a valid keyword, but is it the most “relevant”?  No, but low relevancy may be good for brainstorming about things with cheese.

Want to see some cheese?  Crank up that relevancy factor to ‘100′.

You actually get pictures of cheeses!  These images have been determined (somehow) to have a high relevancy factor when it comes to the keyword “cheese”.

Here’s something else neat.  You can get an idea of what keywords for an image are actually proving relevant.  For example, go take a look at my Red Carpet images or my Film Strip images.  Click on one of the files and if you look at the keyword listing, the keywords are listed in the order of most relevant to least.

The introduction of Best Match 2.0 should save you time searching, and we all know time is money.  Don’t forget to sort by age as well, to be sure to catch some fresh new stuff occasionally.

View More Images

In a sort of “Where things are” post yesterday, on iStockphoto.com, COO Kelly Thompson mentioned something interesting for buyers:

We now offer the option of displaying up to 200 image results per page.

Where do you find this option to change your search return results?  Under the “advanced search” link at the top right of every page.  There is a pull down which allows you to set your return results amount.

Makes for speedier searching.  Great!

“More Like This” in full effect

On iStockphoto.com, when you look at the detail page for any image, on the middle, left side, there is a link that says “More Like This”.

In the past, this link tended not to be very useful.  At one time, it took words from the title and ran a basic search on those words.  Most recently, it picked the first three keywords a contributor had entered and ran a search on those.  Neither of these is/was particularly useful, because they didn’t necessarily represent what I’ll call the “representative keywords” for the image.  ie., what several terms could you use that immediately describe this image.

For the image above, you could say the representative keywords/terms, would be “pot of gold, st. patrick’s day, rainbow”.   However, when entering terms, I might have entered “green, nobody, horizontal” first.  That wouldn’t give you a useful “More like this” search.

Yesterday, that changed.  Using the new “relevancy” keyword factor iStockphoto has been gathering, “More like this” now returns a search based on what could be guessed to be the most relevant (or representative) keywords for that image.   Once the above image starts gathering data, it is likely that “More like this” will return a search based on the three terms above, and a fourth – maybe “green’ or something.

Anyways, the point is that when you do click “More like this”, because you loved mine so much, but need a slightly different angle, or pot, you will get a targeted search based on the keywords that have brought other buyers to this image.  So, you’d be happy to find a search full of rainbows and pots of gold, without having to go back and re-type your search all over again.

Let me know how it works for you.

The Other Big News For Buyers

The iStockphoto.com news about collections and pricing was very interesting the other day, but the best part was yet to come.

In a front page blog post, CEO Bruce Livingstone described a new element to the Best Match Sort.   I talked about searching and sorting way back here – you may want to read that to review the system.  “Best Match” is the default way that the search engine will return an image set from a keyword search.  Basically, you search on “Christmas Tree”, and anything that matches that keyword is returned as a result.  Now, imagine each image has a number from 1-100 assigned to it, and that number is assigned by mixing together how new the file is, how many views it’s had, the size of the file, and a bunch of other stuff.  So, the BM takes the result from the search and orders it, so images that have a number like 1 are close to the beginning (front page) through to images with a rank of 100 (last page).

Now, I don’t know exactly how the sort works, but that’s how I imagine it.  Problem is, sometimes when they’re adding and multiplying things together to get the rank, things may not work like they intended.  For instance, my poor isolated tree below came online, and was doing well, selling 10 times in 5 days.  Suddenly, it dropped off the face of the earth, appearing on something like page (100 images per) 35 of 36 when using Best Match.  Hopefully the intention of the Best Match sort isn’t to punish images that are doing well, but with so many factors, something got messed up somewhere.

The other problem with Best Match, is that there is no notion of relevancy, whether the image is actually an appropriate example of the keyword searched for.  Every image with the keyword is a valid return – the sort is based on a mishmash of data.

All that is about to change, as described here:

A long time ago we developed an algorithm to rank keywords on each file. Since then, we’ve been tracking data for every single file on iStock. Guess what? It works. Starting in mid-January we will start applying these results to BM searches. For clients it will mean more accurate, more meaningful and more relevant results. For artists it means a massive shift. The results for everyone, are going to be very, very different.

So how does this work?  In my imagination, when you search on “Christmas Tree”, you are presented with a selection of results, the way it has always been.  The system remembers your search term, and you go on, paging through a few (dozen) pages looking for the right image.  You come upon my image above, and decide to buy it.  This is essentially your vote that my Christmas tree image is a good example of the term “Christmas Tree”.  That keyword then gets a point added to it, raising the ranking of “Christmas Tree” on that image, but leaving the rest, like “isolated” alone.  The next time someone searches on “Christmas Tree”, my image moves further towards the front of the search, being a proven appropriate match for the term by your purchase.

What is the benefit?  In the end, I see it saving the buyer’s time.  There are a lot of images that validly have the keyword “Christmas Tree”, like this one:

However, if someone is solely searching on “Christmas Tree”, is this the file they want to show up, even if it has high downloads, is new, etc… ?  Probably not.  So, this shifts back in the search for “Christmas Tree”, because it’s ranking on that term is low.  It probably got bought on “senior woman Christmas”, so those would have higher rankings.

I think this is a neat development.  We’ve been asking for this for a while.  It’s sort of a self regulating spam filter.  People can spam “business” onto whatever image they want, but if its ranking never improves with that term, it sliiiiiiides to the back.

So Best Match really does have an element of “Best” now, as determined by you, the buyer.  This is supposed to phase in slowly over the next couple of weeks.  Keep an eye out.

Searching for Numbers

Just a quick tip that may catch you up occasionally.

On iStockphoto.com, if you enter the following into the search box:

2009

you will get a return of the image, numbered 2009.  If what you actually wanted was the year 2009 as a keyword, you need to put it in quotes:

“2009″ (my portfolio)

This does not apply to a multiple phrase search:

2009 render (my portfolio)

Same goes for any other number you may have cause to search on.  You could also use this tip as a reminder that you can just type an image number in the search box, and away you go.

Use More Than One Word!

iStockphoto.com has over 3.5 million images.  That’s a lot of content to search through when you’re looking for the right something.

Unfortunately, the search engine is not psychic.  It does a simple word match and sorted return.  See this article for more details.  The main point being that when you type in “apple”, you may be thinking:

“What I want is an apple on a plate on a white background, and that seems like the obvious thing the ‘best match’ sort should return at the top…”

What the search and sort thinks when you search on apple is:

“Find anything keyworded ‘apple’ and return them all in the sort of their choice”.

The “best match” sorting engine does not incorporate any relevancy factor, or keyword weighting system.  It just knows that somewhere in that image, there’s an apple, as keyworded by the contributor.

All is not lost, however.  iStockphoto allows the contributor up to 50 keywords.  As noted by Ethan Myerson, a keywording admin, there are three types of keywords:

• Literal: the actual nouns, adjectives and verbs of the image. Examples include “Truck”, “Walking”, “Blue”, “Heron”
• Conceptual: the emotive and conceptual aspects of the image. Examples include “Ominous”, “Hopeful”, “Business”
• Compositional: the terms that describe the composition or creation of the image. Examples include “Isolated On White”, “Horizontal”, “Studio Shot”, “Aerial View”

So, you should use combinations of these to find what you want.  There is no limit, as far as I know, to the number of keywords you can use to search.  So, for that isolated apple:

“isolated on white” apple nobody

If you want to get really fancy, here’s a search I made for someone who was just looking for an isolated pineapple.  I kept refining the search until I had just pineapples:

isolated pineapple fruit not juice not woman not salad not female not male not drink not raisins not grapes not orange

So, my point is that trying to find content with a one word search, while it might be good for brainstorming, isn’t really very useful for finding what you want when you know what you want.  Use multiple keywords!  It will save time.  If you need help working up a search, contact me, or post in the Request Forum.