The reason for the problem is that most search tools, including image search tools, don’t actually look at image, they look at the text around it. So unless the title of your work or the file name remain the same, there isn’t much hope for spotting a duplicate via traditional means.

However, in 2008, Tineye changed the game by introducing an image-based image search engine that would fingerprint images found online and match them against files uploaded to their service.

But, while Tineye’s matching technology is and always has been great, it’s been limited by Tineye’s rather small database. While that database has grown over 2x since I first wrote about the service (currently at just over 2 billion images), it hasn’t kept pace with the images being uploaded to the Web (Photobucket alone has 9.5 billion images).

However, last year Google entered into the fray, adding the ability to search for images by uploading or linking to another one, as with Tineye. The feature, which was initially an extension of Google’s previous Similar Image Search function, wasn’t very successful at first. However, over time, it appears Google has gone a long way to improving the tool as, in a recent spate of tests, it drastically outperformed Tineye in finding matching images consistently.

So, for artists looking to find their images on the Web, there seems to be a new king in town and it’s the same one authors have been using for years.

How to Use Google Search By Image

The easiest way to use Google Search By Image is to install either the Firefox
or Chrome extension, which simply enables you to right click an image to search for it.

However, barring that, you can simply visit Google Image Search and click the camera icon in the search bar.

That will open up the window that prompts you to either upload an image from your computer or provide the URL for one already online.

After you submit your image, Google will present a set of results. However, rather than being a “grid” like a regular Google Image Search, the results are ordred, first by exact matches and then by similar ones.

For example, I uploaded my recent image of the cover of Ocean’s Donkey Kong unlicensed port (from my recent article on video game plagiarism) and Google returned not only my article but other images containing that exact image.

In some cases, if there are multiple versions of the images but at different sizes, Google may suggest you look for alternate sizes of the image, as it did with the Limbo of the Lost cover from the same article.

If so it’s worth clicking the link to get a good breakdown of the other sizes (and places) the image appears. However, the real results, especially for images that have been widely copied, are below and you can see them by going through the various pages, as you would with a regular Google search.

But while Google Search By Image is cetainly easy to use, how well does it stack up against Tineye? The answer, is very well.

Comparison Tests

To test the two services head-to-head, I decided to have them both look for five different images used in recent articles on Plagiarism Today. These images are all either freely-available stock photos or are widely-used cover art for video games or records.

With that in mind, here’s the results of the tests:

Test 1: Angry Farm Image

Tineye Results: 1
Google Results: 558 (about)

Test 2: Generic Chart Image

Tineye Results: 4
Google Results: 555 (about)

Test 3: Facepalm Image

Tineye Results: 21
Google Results: 850 (about)

Test 4: Skull on Grave Image

Tineye Results: 1
Google Results: 3 (Not counting matching “Very Similar” results)

Note: Google’s “Very Similar” results were useless in this case as it just found other black and white photos without much regard for things that looked like the original.

Test 5: Bing Crosby White Christmas

Tineye Results: 173
Google Results: 883 (about)

Please note that the greatest limitation of this test is that I had to rely on both search engines to self-report how many matching images they had. However, I checked several pages of results with each test to make sure that the results were as accurate as possible.

Clearly though, the winner is Google, which found, in many cases, over 100x more matching images than Tineye. My suspicion is that, while Tineye’s matching algorithm is better (much fewer false positives), Google’s large database simply makes up the ground and then some, making it a much more valuable tool for image detection.

However, this doesn’t mean that Google’s perfect, there are still a few concerns and problems I have with it.

Limitations of Google

The biggest problem with Google is that, currently, there is no way to do a bulk search for a lot of images nor is there a way to do a recurring search. Though Google has an API for its Google Image Search, it doesn’t appear to work with Search by Image. Likewise, Google Alerts doesn’t allow you to create an alert for a Google Image Search at all, text or by image.

In short, there are no tools to make such searches easier and there aren’t likely to be any in the near future.

Meanwhile, Tineye has a very robust and well-established API that enables toe construction of just such tools.

Still, given how simple it is to use Google Search By Image, even with having to do the searches by hand, it’s still faster and easier than most methods, it’s still free and, in the end, just more effective, even more so than many paid-for tools.

There’s little reason not to integrate Google Search By Image into your checks and to use it at least some in your searches.

Bottom Line

When Google first launched this product, I tested it and found it to be less-than-useful. The algorithm was too flawed (based on the earlier and more limited “similar” search feature) and the number of false positives simply too high. Clearly, Google has made some great strides in the last six months and pushed this service to the point where it’s database and accuracy combine to make it the most useful image search tool available to the public, especially for free.

In the end, while I like Tineye as a company and as an offering, their database is too small and too limited to compete right now. Hopefully though, this competition will motivate both companies to improve their offerings and create a set of solutions that make things much easier on photographers and artists everywhere.

Have any suggestions for the 3 Count? Let me know via Twitter @plagiarismtoday.

1: House Committee Appears Headed Toward Approving SOPA

First off today, the House Judiciary Committee seems to be on a path of approving SOPA, sending it to the House of Representatives for a full vote. The hearing yesterday, which lasted over 7 hours, will continue today but early votes on amendments shows strong support for SOPA in the committee. The bill, which allows copyright holders to seek orders barring ISPs, payment processors and advertisers from letting users access or doing business with sites primarily aimed at copyright infringement, was the subject of nearly 50 amendments, only 10 of which were voted on yesterday. However, one of those amendments, proposed by Representative Darrell Issa, would have removed much of the controversial elements of the bill but was voted down by a 22-12 margin. The hearing is expected to continue today and may stretch past the New Year if Congress goes on break before it’s finished.

2: UMG Claims “Right to Block or Remove” YouTube Videos it Doesn’t Own

Next up today, Universal Music Group hsa responded to allegations that it improperly ordered the takedown of a viral music video posted by controversial cyberlocker site Megaupload. According to Universal, the takedown wasn’t related to the DMCA, but rather, a private contract that UMG has with YouTube to remove videos that infringes any of their rights, not just copyright. This, according to UMG, seems to stem from a partnership the labels signed with YouTube as part of creating the offical VEVO channels. The music video featured various musicians offering support for the site, even as the record labels are condemning Megaupload as a haven for piracy. The video, however, is back on YouTube pending further information from UMG.

3: UK Government Outlines Plans to Reform Copyright Law

Finally today, the UK government has issued its proposal for reforming the nation’s copyright laws and the recommendations follow closely many of the suggestions from the controversial Hargreaves report earlier this year. Those changes include the legalizing of format shafting copyrighted works (IE: ripping a CD to your MP3 player), expanding an copyright exception for researchers to include data mining, expanding “fair dealing” to include satire and, most controversially, making orphan works (works without known copyright holders) available for reuse. The proposal now enters a 14-week public comment period, ending in March of next year.

Suggestions

That’s it for the three count today. We will be back tomorrow with three more copyright links. If you have a link that you want to suggest a link for the column or have any proposals to make it better. Feel free to leave a comment or send me an email. I hope to hear from you.

Want the Full Story?

Tune in every Wednesday evening at 5 PM ET for the live recording of the Copyright 2.0 Show or wait and get the edited version Friday right here on Plagiarism Today.

The 3 Count Logo was created by Justin Goff and is licensed under a Creative Commons Attribution License.

When it comes to detecting the use of your images on the Web, you have two choices: Watermarking and Fingerprinting

Fingerprinting is taking an image and running an algorithm over it and then detecting similar images to it. Watermarking is taking an image that hasn’t been released, adding something to it (either visible or invisible) and then detecting that watermark as the image is transformed.

Fingerprinting is the technology used by PicScout and Tineye while watermarking is used by Digimarc.

But while both systems can be effective and come with their own advantages/limitations, the solutions for professional photographers tend to be very expensive. Though ImageRights has added in an inexpensive option for artists wanting to use fingerprinting, AdvancedPhotoTools has a similar system for those who prefer to watermark their images.

Entitled SignMyImage, it is an affordable way for photographers and artists to add watermarks to their image and track their use on the Web. However, it doesn’t mean anything. So I’ve put the product to the test and decided to find out how practical the system really is.

The Basics

SignMyImage works similarly to any other image watermarking system. You first download either the Photoshop plugin or the standalone application. For the purpose of this test, I used the latest version of the Windows standalone app on my Windows 7 system.

Before you sign the image, you should first register the application. It will ask for your username, email and a signing code that you wish to use on your images. The code can be up to 10 characters and include letters, capitals, numbers and dashes.

Once done with that, you open up the image you want to sign, in this case I am using a web-formatted image I took while visiting Newcastle, UK for a conference a few years back.

You then click to sign the image and it selects your signature for you (Note: This is just a test signature for this review)

You then apply the signature and, after a few seconds, you get a notice about where the signature has been applied on the image and a visual description of the region. Depending on the size and contents of the image, it may be added to multiple places.

If you have not purchased the app, the program will also add a visible watermark promoting the program in the lower right-hand corner of the image. Purchasing the app will also unlock the batch mode, which will allow you to sign multiple images at once.

You can then post the image online. If you have purchased the app or otherwise have a acquired a license to use SignMyImage’s spider, it will automatically begin scanning for your signature, alerting you to any signed images that it finds.

SignMyImage costs 14.99 Euros for either the plugin or the standalone app or 19.99 Euros. That is about $20 or $26 respectively. Either come with a year of free access to the search spider, which costs 9.99 Euro annually after that, or about $13. Since the spider is simply looking for your signature, you can protect an unlimited number of works with a single account.

The Good

The price is the clearest selling point for SignMyImage. For just $20 for the first year and $13 for each after that, you can protect all of your images including monitoring. Professional photographer routinely pay hundreds or thousands of dollars for similar protection of even small libraries.

For that matter, the technology seems to work fair well. My testing with the app showed that it worked reasonably. I tested both the Windows and Mac version and, though the Mac version moved painfully slow on my older machine, especially when trying to add or detect a watermark, the Windows version moved very quick and light on its feet.

The watermarking itself also seems to work well. Though it isn’t as advanced as Digimarc’s, it does seem to do a decent job. APT provides a sample page with images at various sizes, qualities and with variable amounts of cropping to show how durable the watermark is. All in all, it seems to survive most cropping (remembering it only has to have one signature area intact), compression and resizing within reason.

Also, the German photo site Heise performed a good comparison of SignMyImage against competing products, including Digimarc, and found that their watermark had good durability, though it could not survive rotation.

Finally, SignMyImage claims that their watermark causes less distortion than Digimarc’s though I have been unable to tell the difference myself. However, I am not an image buff by any stretch.

The Bad

The weakest link in SignMyImage is its spider. To date, the spider has scanned just shy of 170 million images. While that may seem like a great deal, remember that Tineye currently scans over 1.6 billion as a free service and without watermarking. However, even Tineye only has a fraction of the Web as Photobucket alone has over 7 billion.

While Tineye has a respectable and growing fraction of images indexed, SignMyImage has only scanned about 1/10th of that.

This is somewhat mitigated by the fact that you can set up SignMyImage’s spider to do custom scanning, focusing on URLs that you think are important. However, this doesn’t help if you don’t have at least some idea where the spider should look.

In short, don’t count on SignMyImage to find all or even a majority of the uses of your work. Though it has found some 1,300 matches to date most, likely, were on the artist’s own site according to APT’s owner. Filip Krolupper.

Also, the signature itself is fairly limited. You are given only ten characters with which to sign the image and that information does not connect to any publicly-available database. If you scan an image signed by someone else, you simply do not know who created it. Compare this to PicScout’s very public ImageExchange platform.

In short, SignMyImage is not going to detect all the copies of your work nor is it going to resolve the orphan works issue. That may make it a solution you don’t wish to invest in.

Bottom Line

Though the watermarking technology itself appears to be good and the price is almost unbeatable, until SignMyImage improves the spider its usefulness as an image detection system will be very limited.

Unfortunately, since the signature doesn’t include any additional information about the creator, it doesn’t solve the orphan works problem either, making it a solution looking for a problem. That’s not to say that SignMyImage is bad, it is only inches away from greatness, but those inches are very important.

The bigger problem with SignMyImage, however, is that it, along with all other watermarking solutions, does little to help artists who already have images on the Web. Though they can re-upload their images and make sure all future copying is tracked, existing copies won’t be and that is a problem that requires fingerprinting.

To that end, hybrid solutions are usually best, but usually impractical given the costs and time.

So while I like SignMyImage as a solution, I can’t heartily recommend it at this time though it is one to watch and see. If they can begin improving their scanning system, they may have the perfect solution for small-to-midrange photographers and artists. or at least a compelling added layer of protection.

In November 2008 I wrote an article that highlighted how Flickr and Facebook both strip out EXIF data in images that are uploaded. A comment to that post recently got me excited about the possibility Flickr had changed its ways and initial testing indicated it might have.

However, sadly, a year and a half later, nothing has changed. I’ve redone the tests with no luck. Flickr and Facebook are both stripping out EXIF data to uploaded images. The only exception is that I can now confirm Flickr is preserving the data on original images, just not on the automatically generated resized versions.

Below is a quick analysis of the problem and why it is a serious issue that every photographer should be worried about.

How Flickr Does It

If you upload images to Flickr, the site automatically generates up to five different versions including square, thumbnail, small, medium and large. If you are a non-pro account user, these are the only images sizes available as they can not download originals.

The original files preserve the EXIF data but the other sizes, the ones generated by Flickr. Do not. For example, compare the EXIF data stored in the two sizes of this image, the original being on the left. (Note: Original image by mikebaird and licensed under a Creative Commons License)

Since that image was from 2008, I redid the test using this image, which was uploaded this month and found the same results. (Note: Original image by dicktay2000 and licensed under a Creative Commons License)

In short, Flickr still has the same problem it did two years ago and is stripping out the metadata on images when it resizes them. Unfortunately, it resizes all images uploaded and, in the case of unpaid users, there is no way to access the originals. Furthermore, since the original images are usually too large for Web use, it is the smaller ones that get passed around the most.

Facebook Too

I also repeated the test with Facebook, using an image I took during a recent geocaching run, and compared an original from my phone to the version I downloaded from Facebook. The original is on the left.

Unfortunately, with Facebook, there is no way to download the original photo and, as such, the EXIF data is never seen publicly. All you can do is either add a visual watermark or hope that, if the image is passed around on the Web, that attribution remains intact.

Why This is Bad

With the recent near-miss in the Digital Economy Bill in Britain over orphan works, it is clear that orphan works legislation will be a part of the future of copyright.

Though there were differences between the UK bill and the two that were tried in the U.S. without success, the crux is the same. If the owner or the rightsholder of a copyrighted work can not be found or identified, the work can be used so long as the person using it performed a reasonable search.

This issue most directly impacts photographers and visual artists as their work is more difficult to search for and often has no attribution affixed to it. As such, image sharing sites, such as Flickr, should be doing everything they can to help identify the authors of the works they host, including preserving the EXIF data.

The failure to do so, especially for Flickr, which caters so strongly to professional photographers, is very dangerous and may create serious problems down the road for the users who trusted the service.

Bottom Line

While I grant that this is an invisible problem in that EXIF data is not immediately seen by those who view or use the site, it is an important one to fix. Hopefully both of these sites will work on a fix to preserve the metadata of images uploaded to them to prevent creating orphan works.

After all, the problem of orphaned works is not one that’s going to go away and the only way it can be resolved with current works is through cooperation of all involved and that includes photo sharing sites.

In the meantime though, just be aware that, when you upload your images to Flickr or Facebook, your metadata may not travel with them.

If you want to check your blog for plagiarism, there’s Fairshare (previous coverage). if you want to check your static text for plagiarism, there’s Copyscape (previous coverage) and Plagium (previous coverage). If you want to check your images for plagiarism, there’s Tineye (previous coverage).

But what about your Twitter stream? Up until now there has been precious little.

However, Robby Grossman has developed a new service entitled Don’t Plagiarize Us that it hopes will fill that gap.

The service promises to check your Twitter stream (or anyone else’s) for duplicate tweets and report back on what it finds. However, after putting it through a few paces, I found the results to be interesting, but not altogether useful, especially considering the nature of Twitter and how it works.

Still, it is likely worth a try, especially considering it only takes a second and is completely free.

How it Works

The idea behind Don’t Plagiarize Us (DPU) is pretty basic. You put in your Twitter username, or whatever username you wish to check, and it scans the most recent tweets, usually about ten, and it finds tweets that are similar but aren’t marked as Retweets or replies.

In short, if the tweet is the same or close and it doesn’t contain attribution, the service marks it as suspicious.

One interesting thing DPU does is mark cases that it considers possible “indirect” or “fuzzy” plagiarism. These are cases where the tweets are similar, but not exact, to the source. The idea is to catch people who maybe copied a tweet, but altered a few words or rearranged it in some way.

With that in mind, I gave DPU a few tests and my results are below.

My Tests

For my first test of the service, I decided to run my Twitter account through it. It searched back through over 3000 tweets on the first click and only found three instances of suspected plagiarism, all of them of the “fuzzy” kind.

However, none of the incidents seemed to really be relevant.

For example, it matched my Tweet:

Long day. Time for some Wendy’s.

With this one by @thequestion77:

No baseball practice today. Which is good, cause I’m beat & I have a long day tomorrow. Time for Wendy’s & then some serious lying around

I seriously doubt either was plagiarized and it is almost certain that the similarities are purely coincidental. It was a pattern that fit with the other “fuzzy” matches that DPU handed me.

I then tried the account of my friend and Copyright 2.0 Show co-host Patrick O’Keefe, AKA:@ifroggy. However, this time, DPU only went back ten tweets and failed to find anything. Even after clicking the “Go Back Farther” button a few times, getting back 37 tweets, still nothing showed up.

I then tried a few Twitter celebrities including Ashton Kutcher (@aplusk) and Diddy (@iamdiddy).

Kutcher’s account produced a lot of matches though almost all were “fuzzy” ones that had little significance. There are also false negatives, exact quotes listed as fuzzy ones, and several cases where tweets were in the list twice.

Diddy’s account was less interesting, having found only one duplicate tweet, after going back about just 20 tweets. This one was an exact match that appears to be from a Twitter bot that took Diddy’s tweet and reused it without attribution. This is, most likely, the only true plagiarism I found.

Finally, in the name of fairness, I also ran the creator’s account through DPU. After going back almost 1800 tweets the system found a slew of similar tweets though, once again, the majority were simply “fuzzy” matches that were almost certainly not plagiarisms and the exact matches were far too short to be useful, less than six words.

Of all the copies of his tweets, only one was interesting, but it seems to be a retweet that was missing the attribution, not an attempt to plagiarize as it was just a promo for a livestream.

Thoughts on the Service

After all of this testing, the impression I walked away with was that it was a neat idea for a service, but one that was extremely flawed.

First, the matching did not seem to work very well. As I mentioned above, the fuzzy matches were almost completely useless and even the verbatim matching was often too short to mean much. A service like this would clearly have to focus more on quality, not quantity of matches given the shorter length of tweets and the sheer volume of them.

Also, the service itself seemed unreliable. When I initially tested it I received error messages but, at the encouraging of the creator, I gave it a second try and did indeed find it working, for about 3/4 of my tests. Far too often I was greeted with a notice like this one:

The errors were fairly random and I could often just go back and try my query to fix it, but they were still frequent enough to be extremely annoying.

However, the bigger problem is with Twitter itself. The nature of the service, with tweets under 140 characters, most of which are mundane in nature, and billions of Tweets being passed around, it is almost inevitable that there will be similar tweets without any malicious intent.

Though the service does a decent job ignoring retweets and replies, two elements that could have greatly messed with the results, the fact is that there is still so much coincidental matching that there is little hope for separating the actual plagiarisms from the coincidences and, even if you can, there is little one can do as, in most cases, tweets likely aren’t protected via copyright.

Still, this isn’t to say you shouldn’t give this service a try. It is a fairly cool idea and it is is a great way to see who is tweeting about similar things to you. It might not help you catch a plagiarist, but you might find a few cases where a retweet went sour or people were talking about the same topic as you.

It could, if nothing else, be a new way to meet people on Twitter.

Bottom Line

The system is a neat idea but I don’t think it has much practical use now. With some tweaking it could become more useful for Twitter plagiarism detection though it is unclear what one would do about that in the majority of cases.

The site does mention that they are working on “real time monitoring” of Twitter plagiarism, a feature that could be very useful for those whose tweets are often copied without attribution, but I suspect that the usefulness of that service will be limited to only certain kinds of Twitter users.

All in all though, definitely give this service a try. Even if it isn’t the most useful, it’s free and can still provide some interesting information.

1: Anti-piracy Laws Passed: ‘Fears of Wave of Censorship’ Raised

First off today, the Digital Economy Bill passed the House of Commons in the UK last night, paving the way for it to be signed into law as early as today. The controversial bill will allow, with a court order, the government to demand ISPs block certain Web sites and may pave the way for a “three strikes” system that could cause alleged file sharers to lose access to the Web. Gone from the last version of the bill is the orphan works provision, which would have allowed use of copyrighted works for which the owner could not be found.

2: YouTube ‘Star Wars’ Sensation Explains Copyright Issues

Next up today, Mike Stoklasa, the creator of the wildly popular YouTube series making fun of and reviewing the first two “Star Wars” prequels briefly had his most recent series, a 90-minute review of “Attack of the Clones”, briefly removed from the site due to a copyright infringement claim, supposedly by Cartoon Network. However, the videos were restored but questions were raised about whether his long-format reviews were a fair use or not, an area that is gray enough to discourage Stoklasa from filing a counter-notice. The videos, however, have been restored.

3: Hot news: The Next Bad Thing

Finally today, commentator James Boyle has some thoughts on “Hot News”, a legal precedent that extends well past copyright and into protecting facts that were collected at expense and might give competitors an unfair advantage. Some news organizations have pushed for such protection, even relying on an early 1900s court case, but, according to Boyle, that could backfire by not adding any significant protection and prohibiting news organizations from actually covering the news. An article definitely worth reading.

Suggestions

That’s it for the three count today. We will be back tomorrow with three more copyright links. If you have a link that you want to suggest a link for the column or have any proposals to make it better. Feel free to leave a comment or send me an email. I hope to hear from you.

Want the Full Story?

Tune in every Saturday morning for the live recording of the Copyright 2.0 Show or wait and get the edited version Monday morning right here on Plagiarism Today.

1: UK Gov Nationalises Orphans and Bans Non-Consensual Photography in Public

First off today, some last-minute changes to the Digital Economy Bill in the UK is drawing the ire of visual artists. A largely overlooked portion of the bill deals with the licensing of orphan works, works for which the copyright holder is unknown, and opens the door to their use so long a licensing fee is paid to an unnamed licensing body. Photographers and visual artists are deeply upset about this but it seems that the bill will now go through unchallenged.

2: NYT reporter Kouwe resigns amid allegations of plagiarism

Next up today, New York Times reporter and blogger Zachery Kouwe has resigned after the editor of the Wall Street Journal sent a letter to the paper expressing concern over some passages in his work. A further sweep of his writing found several more suspect passages and Kouwe resigned at a meeting within the newspaper. Kouwe has said that the plagiarism was unintentional.

3: Aamir Khan Resigns from Copyright Act Panel

Finally today, in India, Bollywood star Aamir Khan has resigned from a government-assembled panel to discuss changes to the country’s copyright act. This comes after, according to media reporters, Khan said that a popular song of his needed his star power to become popular and that, by inference, writers should receive lower royalties. This angered many authors who accused Khan of trying to hurt them.

Suggestions

That’s it for the three count today. We will be back tomorrow with three more copyright links. If you have a link that you want to suggest a link for the column or have any proposals to make it better. Feel free to leave a comment or send me an email. I hope to hear from you.

Want the Full Story?

Tune in every Saturday morning for the live recording of the Copyright 2.0 Show or wait and get the edited version Monday morning right here on Plagiarism Today.

photo credit: Joi

Joi Ito, best known in copyright circles as the CEO of Creative Commons, has joined Picscout’s advisory board.

Ito, who is also the CEO of Neoteny, an Internet investment firm, and on the advisory board for Digital Garage as well as many other boards, is working with Picscout on their new Image IRC and Imagexchange tools, previously covered here.

The Image IRC is a database of fingerprinted images with their metadata (including licensing information) that is accessible via an API. ImageExchange is the end-user component for identifying images within that database and licensing them, currently used in the form of a Firefox extension (other updates, including an Internet Explorer extension, are reportedly in the works).

Ito brings with him a significant amount of clout to the project and a great deal of expertise. Though his role is vague at this time, there is little doubt that it is a major coup for Piscout as it tries to position itself as the go-to source for image detection and licensing assistance.

Ito’s Possible Role

As I said in my original article, two of the greatest problems I saw with the Image IRC/ImageExchange combination was the lack oif broad Creative Commons (and other licensing) support as well as the hurdle of reaching out to every day users. Ito has a great deal of experience with both of these areas and may be able to provide advice and support it both better integrating CC licensing with Picscout’s tools and educating the broader public about their purpose.

Ito’s involvement could be a great asset to the service as it works to both gain traction among visual artists and users. However, his impact may be felt further down the road as Picscout starts to integrate more and more CC-licensed images into the Image IRC and display CC-licensed images alongside stock ones.

It is worth noting that this does not indicate any formal partnership between Creative Commons and Picscout, as mentioned above, Ito works with many different organizations, nor will Ito be stepping down from his Creative Commons position, but there is little doubt that his knowledge and expertise can be of a lot of use in these areas.

Expanding Reach

In addition to bringing Ito onto their team, Picscout has been expanding the reach of their ImageExchange service, adding Dreamstime stock image library (over 7 million images) as well as Masterfile, Life, Alfo, Mauritius, Blend Images and Glow Images, creating a mix of microstock, royalty-free and rights-managed images.

ImageExchange has also integrated CC-licensed images through Flickr, though only non-commercial images at this time.

This is clearly a sign of momentum for Picscout and may be a sign of the role the Image IRC/ImageExchange solution will play.

Bottom Line

Though the news is good for Picscout, obviously it is still very early in the beta process and far too early to tell how successful its new products will be. Though these are clearly big steps in the right direction, it’s sill just speculation as to what will come of it all.

In the meantime though, I’m putting ImageExchange through some paces on my Windows computer and should have a report back to you later this week about its effectiveness and usability.

Until then though, this is clearly a company that visual artists of all stripes need to keep an eye on moving forward, especially considering both the existing orphan works problem and looking down the road to a potential bill on the subject.

The image detection field is changing rapidly. For years I’ve been predicting the growth in image searching and it appears that is finally coming to be. Not only has PicScout has begun public testing on its newest offerings and other announcements are in the works.

As such, over the next few weeks and months, it seems almost certain that I’ll be posting articles about new technologies and advancements in this field as it continues to heat up and prices begin to drop. This means that many photographers, including smaller ones, will be confronted with the question about how to best protect their images and track their usage.

Three years ago, this was almost unthinkable. Limited options and high prices made such tracking attainable only to large companies but now we’re preparing to enter a very different age for visual artists when it comes to following their work on the Web.

With that in mind, I wanted to take a moment to briefly revisit the two schools when it comes to image detection, watermarking and fingerprinting, discuss their pluses and look at which solution may be right for you.

Hopefully, by the end of this, you’ll be better prepared to decide on your own image protection scheme and know which technology to watch for.

The Two Schools

The challenge in finding copies of images on the Web has always been, to put it bluntly, that they aren’t text. Search engines, including image search engines, look for text as it is what is easiest to search for. Computers don’t understand what is inside an image (the reason CAPTCHAs are reasonably effective) and they have a hard time comparing one image to another. Scale that problem to the billions of images on the Web, and this issue becomes clear.

Solving the problem hasn’t been easy nor has it been perfect. However, there have been two different approaches that have been tried, both with great success but different ideal uses.

1. Watermarking: Though most think of image watermarking as the placing of a visual mark over an image, usually a logo or name, Watermarking in this sense involves placing either an invisible or nearly-invisible watermark over an image that can only be detected by a machine. This method is used by Digimarc and C-Registry.us (Note: The latter uses both for various purposes).
2. Fingerprinting: Fingerprinting is an automated process that takes an existing image file, hashes it and converts it to a unique fingerprint. That string is, in turn, compared to other fingerprints for potential matches. This method is favored by PicScout and Tineye.

As you can see, the two systems are fundamentally different in terms of how they work and what they can do, these differences have an impact both in terms of how they are used and the situations they are best suited for.

Watermarks

The advantages of using a digital watermark are many-fold. First, they provide much more reliable detection. Where a very similar image by a different artist can trigger a fingerprint match, only copies of your actual image will trigger a watermark match. This also greatly helps with providing evidence for a potential court case since it can be shown exactly which image was copied from.

Watermarks can, typically, also survive great modification by users. Often images with watermarks can be detected even after cropping, rotation and distortion that would have caused a fingerprinting system to miss it (though fingerprints are getting much smarter about this as well). Some need only a few pixels to remain intact.

Also, additional information about the creator can be hidden in the watermark or stored in a database. This can help greatly with the worries surrounding the orphan works bills, which may allow certain uses of works if the author can not be reasonably identified.

In short, it works like image metadata in that it is invisible and is buried within the image, but it is much harder to remove and travels with the image even after it is modified and resaved.

However, watermarks can’t do anything to protect an image that’s already “in the wild”. Though you can re-release an image with a watermark, any copies of the non-watermarked image will not be detected. Also, watermarking systems tend to be expensive to use and, since one company usually controls the watermarks and the system for detecting them, they’re vulnerable to being rendered useless if the company closes.

Fingerprinting

The biggest asset of fingerprinting is that it works on any image, including one already on the Web. There’s no need to watermark the image or do any preparation work at all before publishing it, the technique works from the image itself.

Fingerprinting has become more and more accurate over the years, now competing with watermarks in that area, and is largely cheaper, including a free, effective offering from Tineye.

Also, since there are many fingerprinting systems in place and all work roughly the same way, its likely that you’ll be able to continue to match your images this way for many, many years to come. One company closing does not destroy your copyright protection system.

That being said, Fingerprinting does not, by itself, address the orphan works issue nor provide any proof of ownership, which is why PicScout recently released the Image IRC, nor is it always 100% reliable. Similar images, such as photos of similar subjects, have been known to trip fingerprinting matches, though the problem is nowhere near as severe was it was.

So Which to Use?

When deciding which to use, the easiest answer is to just say “both”. Since you can fingerprint a watermarked image, there’s nothing, save perhaps cost, that will prevent you from using both systems. Many already view watermarks as more of a copyright verification tool than an image detection one, making the two a natural combination.

However, duplicating efforts doesn’t make much sense. Due to refinements, both systems can do an extremely good job at finding matches, what is more important than the method is the individual system and how well it works. If it can find good matches and work as advertised, it’s probably a keeper, regardless of the system.

That being said though, you can’t even consider using a watermarking system if your images area already on the Web unprotected. It is only used, ideally, for new images as much of the usefulness is lost with unprotected copies being passed around. However, the added protections provided by watermarks make them worthwhile in many cases, especially when there is tight control over distribution and the images haven’t been leaked.

In short, if you have images that you are waiting to put on the Web or routinely put up new ones (and can justify the expense both in money and time), watermarking is likely a good solution due to its added benefits. For those who have their works online already or don’t wish to invest the time/money required, fingerprint matching, via Tineye, is free.

However, even that distinction is being blurred by PicScout’s Image IRC, which is a database of fingerprinted images, which provides something of a “best of both worlds” solution. The problem though is that it, right now, is targeted solely at large stock agencies. We’ll have to see later how approachable it is by smaller artists once it is finally released.

Bottom Line

In the end, any robust image protection system will likely use at least some of both. If your images are valuable, trusting any one solution by itself to handle detection and orphan works prevention is risky. Multiple tools are, most likely, the way to go when practical.

But the most important thing, as mentioned above, is not that the system works, not the technology behind it. If it works well, it doesn’t matter if it uses fingerprints or watermarks. That is where the real challenge lies for these companies.

So, with that in mind, I’m looking forward to, over the next few months, talking more about developments in this field and how they might impact visual artists of all kinds and all backgrounds.

Imagine if you could open up your browser, search for an image using Google and instantly know who produced every image that came up and how you could license it for reuse. Likewise, as an artist, imagine that you could index your works into a database and everyone would know who the creator was and be able to license it from you, no matter where they saw it on the Web.

That’s the future that Picscout is trying to bring about with a pair of new initiatives , Image IRC and Image Exchange.

But while this brave new world likely sounds very enticing to visual artists, the framework of it has just been announced and it has many hurdles to overcome before it becomes a part of most artist’s daily lives.

Still, it is an interesting idea with a great deal of promise.

How it Works

The system consists of two parts which are both required for it to function:

1. Image IRC: This component is essentially a very large database of Images that PicScout has collected information on. The images are fingerprinted to make them easily identified and are then paired up with their metadata, which includes information about who owns the image and where it can be licensed.
2. ImageExchange: This is the component that is end user facing, allowing Web surfers to automatically locate and license images contained in the Image IRC. One example is a planned Firefox extension that will be released that allows users to automatically find out if images they see on the Web are licensed via the Image IRC.

The result is that, if a user is surfing the Web with the ImageExchange addon running, they will see an “i” over images that are in the Image IRC database.

Users can then click on one of the “i”s and will get a popup like the one below with more information about the image.

From there, they will be given the ability to correctly license the image, including the ability to click through and purchase a license if needed.

Currently the Image IRC includes “tens of millions” of images, including some 7 million CC-licensed images from Flickr. PicScout is also opening up to agencies, professional photographers and photo-sharing sites that can provide bulk image delivery.

PicScout hopes to open up the platform to smaller copyright holders and sites in the near future.

Why It’s Important

What is interesting about these new tools is that image matching and fingerprinting has typically been used as a one-way street. Creators have used the technology, often PicScout’s, to locate copies of their own works. The idea is to take the process and put it in reverse.

Rather than scanning one image to find its copies on the Web, one now automatically scans the images they see to find which are available for use. It is the same technology but, rather than starting from a known source and locating unknown matches, you’re starting with unknown sources trying to locate known matches.

This has the potential to be a boon for those who regularly license images. Rather than going through stock photo libraries to find the right image, they can have their browser (or a third-party site using the API) passively scan images as they surf the Web, allowing them to use the broader Web and better search tools.

These tools could also, foreseeably, have an impact on the orphan works issue, or works with unknown authors. By creating a database of easily-identified images, it decreases the likelihood an image’s creator might become unknown. Though PicScout is careful to not guarantee its service will be part of a “reasonably diligent search”, which was the language of the previous orphan works bill when determining whether a work was an orphan or not, it is easy to see how it could be useful in such searches.

However, this service does have a lot of obstacles in its way and there may be at least one problem that it can not solve with technology alone.

Problems and Concerns

Obviously, since the service has not been unveiled publicly, there isn’t much that I can say about it. However, there are several areas of concern that need to be addressed before this system can be called a success.

1. Database Size: Though the database should swell in size as testing begins, tens of millions of images is not an impressive sample. Tineye has over a billion images and is growing very rapidy and, even then, it doesn’t escape criticism from me for not catching all infringements. While it is true that the Image IRC database doesn’t have to be as large as Tineye’s as it is only indexing originals of images, not their copies, if the database is too small, it will not be useful for performing image searches and that will make it useless for everyone.
2. Licensing Limitations: All of the images in the Image IRC must be made available for licensing either as a commercial work or under one of a few CC licenses (according to a representative from PicScout, the image must be licensed for non-commercial use at this time). There’s no way to use the Image IRC to say that you do not want your images used. This limits the usefulness to photographers who have no interest in image licensing.
3. The Wrong Audience: While this will be seen as a great product for those who routinely purchase photos online, it remains to be seen how well it will be adopted by those who misuse images, either through ignorance or malice. Those who don’t know or don’t care if they are violating copyright are unlikely to download a Firefox plugin to start buying images they see on the Web. If that’s the case, the actual impact on getting users to license images could be small.

However, the gravest concern I have is that this tool could create even more confusion when it comes to licensing. Given the current level of copyright understanding on the Web, it is conceivable some could use this tool and think think not that they can license the images with the “i”, but that those are the ones they have to pay for and the others are free.

In short, without proper education about how the tool is to be used and without proper groundwork on copyright education, it is conceivable that this product could actually backfire, especially against those not in the Image IRC system.

That being said, these are issues that can and should be addressed as the product development rolls along. Right now we have a very brief glimpse at a promising new tool that could, if it is able to grow to a large enough size, could revolutionize the way people buy stock images and the way people find works for their various projects.

Bottom Line

There is no doubt that PicScout’s matching technology is among the best in the industry and its ability to use fingerprints of images rather than watermarks (visible or invisible) to track works enable it to track a much wider range of material. There is also no doubt that this application of their technology can have a lot of benefits to copyright holders of all kinds.

The questions that remain unanswered is how will this system be developed and how will it be used. If PicScout is judicious and wise with how they build this service, it could become an incredibly powerful tool for users and a major coup for visual artists, who have struggled with being found since the dawn of the Web.

If it is developed unwisely, it will likely remain a niche product, used primarily by those already in the business, that does little to improve the overall copyright climate. At worst, it could even introduce additional confusion and headaches.

Without a doubt, this will be an offering to follow over the next year or so. No matter what, it seems destined to make an impact.

Full Sample Image

Click for full size.