From Paper Watermarks in Bank Notes to Digital Watermarks in Photos

I have no new tutorial for you this time, but I think a little history lesson about watermarks can also be very interesting. To get my bachelors degree in computer science back in 2006, I wrote a thesis about digital watermarks. In this blog post, I would like to sum up the most important facts that I have investigated in my thesis and provide an overview about digital watermarks. And yes, this time we will not only talk about watermarking photos.

Watermarks in history

Everybody knows watermarks in bank notes. The first paper watermarks were initially spotted in the end of the 13th century in Bologna, Italy. In these days people did not add watermarks to of send hidden messages, copyright protection or copy protection. Paper manufacturers embedded symbols and messages into paper to sign their product. These identification marks are also often used to date historic documents.

Basically, there are multiple definitions for the word “watermark” that are used by authors, historians and scientists. Mostly, a watermark is defined as not immediately visible data within different kinds of media. Usually, this data cannot be separated from the carrier. Watermarks are therefore part of the science of steganography.

Nowadays, visual watermarks are mostly used to protect photos with logos or texts. Their hearable counterparts are audible watermarks like noise or spoken words. Some people are claiming that noticeable watermarks are ruining a picture or a song, therefore the distribution of additional hidden informations with watermarks is getting more and more important these days. The big problem with hidden informations is that they can be removed from the medium easily when transforming digital data. Invisible photo watermarks for example can get lost if the image format or size gets changed for example. On the other hand, watermarking photos with visible logo or text watermarks can be removed by cropping the watermarked parts of the image. So the more robust and stable a digital watermark is, the more effective your image, music, video or document gets protected from content thefts.

The rise of digital watermarks

Digital watermarks are not an invention of the internet era. You are probably surprised to hear that experiments with digital watermarks have started back in the 1950s. These efforts were not really honored by the industry then.

The first application of watermarks was used for signing audio signals by Emil Hembrooke of the Muzac Corporation in 1954. It was a patent entitled “Identification of sound and like signals” that allowed the positive identification of the origin of music. So it was the first electronic watermark to prevent piracy.

But, it should take a little longer that digital watermarking gets popular. In the mid-1990s when the Internet conquers the world, the entertainment industry sits up and takes notice. The Internet connects the whole world digitally and is a new medium for distributing data and services. In the end of the 1990s, the situation escalates when the file sharing service Napster gets fought by the music industry after distributing music files for free. Of course, the music industry defeats Napster and after the shut down of the service, the huge media groups are working on copy protections and digital rights management solutions. Now digital watermarks are on everyone’s lips because unprotected published content can be copied easily digitally.

Characteristics of digital watermarks

The characteristics of a digital watermarks are dependent on the type of medium and how a person can consume this medium. The problems when designing a watermark are often the robustness and noticeability of the watermark. These two characteristics play off against each other because one behavior effects the strength of the other.

As mentioned before, robustness is a big factor when considering to watermark photos or audio files. Watermarks are robust when they survive within their carrier medium. Take a photo watermarked by bulkWaterMark or other batch watermark software for example: The watermark is only robust as long as it is visible on the picture. By removing the watermarked parts of the picture, the watermark gets also removed. If the watermark is positioned on the most interesting part of the photo, the watermark will probably never get cropped from the picture.

And so the dilemma begins: The most robust visual or audible watermarks are destroying the image or audio data – they are noticeable. So the more robust, the more noticeable. Invisible audio and video watermarks are often realized with proprietary file formats, but there was never a popular and inexpensive solution for image protection. While invisible watermarks can also be embedded into photos by adding fingerprint data to the image, robustness is very difficult to achieve since the watermarked image must be a standard image file format. iTunes has proved that people are ok with proprietary audio and video files, but this approach does not work for photos and images. The aforementioned resize and compression of image data can eliminate an invisible watermark easily. Most professional photographers today tend to use discreet visual watermarks for photos, while scaling down the pictures additionally. In the times of HD displays, nobody wants to steal a pixelated and heavily compressed JPEG image with a watermark.

Conclusion

So noticeable watermarking is still the most used type of watermarks for protecting photos and audio files. Photo watermarks can be used not only for image protection, they are also very popular for transporting advertising messages. The days of invisible DRM audio watermarks are probably over, but audible watermarks are still used for demo clips on content marketplaces like Envato.

I hope you liked the excerpt of my thesis! Stay tuned for our next blog post – it’s a tutorial again! 😉

Cheers

Phil