Digital Watermarking

In view of the problems present and perceived regarding copying and image theft, we have been looking at a service called Digital Watermarking that is supported by all the major graphics editing software such as Adobe, Paint Shop Pro etc etc, and have subscribed for a year to the Digimarc Pro service to see how good it is.

It is currently on special offer at $US 79.00 for up to 5000 images without tracking, or another $120 to enable tracking. There is a hobbyist version at $49.00 for up to 100 images.

From today, we will be enabling the digital watermark in all our images on the three websites that we run, and will update the service to include tracking later on.

formatting link
Will let everyone know how we get on.

Peter

-- Peter & Rita Forbes snipped-for-privacy@easynet.co.uk Engine pages for preservation info:

formatting link

Reply to
Peter A Forbes
Loading thread data ...

"Peter A Forbes" wrote in message news: snipped-for-privacy@4ax.com...

It is a pretty sad state of affairs when you have to incur extra costs to protect your work from those who will steal anything which is not nailed down. While I have come across a few of my images on web sites being used without authorisation, the major issue I have encountered is that of hot-linking, (theft of bandwidth). Around October of last year it became such a problem that I was faced with either the choice of withdrawing the sites or taking action that would stop the problem without the need for me to spend all my time chasing those causing the problems. As I was not keen on learning new tricks, the first option was tempting, but having spent many an hour working on the site all the effort would have been rendered pointless. Anyway to cut a long story short I decided to bite the bullet and spent a few days learning about .htaccess and how to set it up so that anyone hot-linking to one of my images would be discouraged from using the practise. I am pleased to say that the method would appear to have been entirely successful, if there is any such abuse still going on it is so insignificant that I no longer see the evidence in the statistics. This link:

formatting link
illustrates what happens when someone hotlinks to a Belgian Aviation Patch on one of my sites. When most people click on the link they will see a warning image, in certain cases where the users security settings are high they will see the "stolen image" an 'Ace Guard' patch. It seems that lower security settings are required in Forums where most of the hot-linking was going on, and that the protection method has been successful.

I have checked out some of the images on your site, but so far I have not found any evidence of digimarc protection. Perhaps I have been looking in the wrong places, I would be interested to see how effective the method is and would be grateful if you can post a link to a page where the graphics have been watermarked. While I was looking at your site I noticed that the image in the top left of this page:

formatting link
is not appearing. Looking at the above URL another issue comes to mind, I don't know if it is still the case, but a mix of upper and lower case in a URL could cause problems in certain browsers, or perhaps it was servers.

Reply to
Richard H Huelin

Richard, Does that mean posting interesting URLs on NG is a no no?

Reply to
Nick Highfield

As it happens, I know something of such things... (I worked on the BBC's wildlife image site, the Robert Harding picture library and

formatting link
I also spent much of the last two years building media vaults to store such things, handling services like watermarking internally and automatically)

There are two sorts of watermarking; visible and invisible. Visible puts an obvious trademark on the image, invisible is hidden in the innards of the digital bits (but any copy of Photoshop can read it).

It's trivial to crop off a visible watermark if it's small, and a large one is irritating for genuine users. They work best for diagonal or triangular objects on complex backgrounds - so a small one can be placed in the "boring" corner, but is still hard to remove. However this needs manual placement (or clever software) so it makes the watermarking process more labour intensive.

By some clever geekery, invisible watermarking is hard to remove. Even if you edit the image, it's still recoverable. It's most robust for colour photos, however for b&w line drawings or scanned text documents then it _is_ removable with not much trouble.

Visible watermarking can be done by any image editing program, but the extra sophistication of better (i.e. Photoshop's task scripting) or a commercial product can reduce the workload of doing it. Invisible watermarking basically means Digimarc, which requires money.

My experience has generally taken the same approach to protecting content, an approach which was developed on BBCWild.

Images exist in three forms; high quality from the scanner, "comping" images and "retail".

High quality may be in TIFF (huge and ugly) or something like Kodak PhotoCD. These are the "crown jewels" and they don't leave the building.

Comping images are high enough quality to be useful as advertising and for some purposes (comping is a process in magazine design, of laying up a dummy page). They're of low enough quality that it doesn't matter who has them, or what they do with them. Typically they're of a size appropriate to the audience, they're visibily watermarked (which can be ugly) and the image compression makes them of poor quality. For something like an image library serving magazine publishers, a comping image might actually be quite large (we've all seen the cheaper magazines who've actually gone to print with such things, pixels jaggies and all). If your final audience is just the web, then you have to "uglify" the comping images by heavy watermarking to stop them being used instead of the real product.

Retail images are of high quality and high resolution. Usually they're not visibly watermarked (depending on the market) but they are invisibly watermarked. If they're really valuable, you can generate a new watermark for each set of images, incorporating the buyer's account ID or an invoice ID. This makes each image individually traceable; not only can you prove that it was pirated from you, but you can prove which customer first bought it and was then either careless or themselves plagiaristic. Of course, this multiplies the per-resource watermarking costs. Not too bad if you sell relatively few copies of each image from a wide selection, but it is noticeably more expensive if you're mainly selling a handful of popular images.

On the whole, my experience with watemarking is less than ideal. It works, and it's a good technical product. However it doesn't do much to _stop_ piracy, all it does is make piracy traceable afterwards. If you're trying to stamp it out, it's equally important to explain things to the pirates beforehand, and make sure they understand what you've done and how it might affect them.

1) It has a discouraging effect. This depends on your opposition, and how seriously they take the threat of 4). It's especially useful at the high-end, where you can trace a pirated image to the original purchaser - so long as they're aware of this beforehand. 2) It makes piracy less useful. Particularly for visible watermarking, it might make the pirated product less attractive. 3) It can aid in tracking your resources. This is more use if it's a big market (maybe chart music) and you're interested in the source of individual items in a large stream that's largely legitimate. In your case though, you can probably recognise your own content quite easily, just by subject. 4) It can be very useful for demonstrating piracy in court (particularly invisible). However this all depends on you finding the pirates in action, and being prepared to take things further.

Hope some of this is useful to you.

-- Smert' spamionam

Reply to
Andy Dingley

Only got the system ID this morning, Richard, so don't expect the 1500 or so images to reappear overnight all marked up :-))

The capital letter in the url is not a problem, it's mainly email addresses that suffer from upper case.

The missing image in Carb6.htm is a result of the reorganisation last spring, we still find a few odds and sods that are displaced or never got changed over from the big single directory.

It's there now, and I also noticed that the "alt" header for each of those icons was wrong so I corrected the 50 or so of those as well :-))

Peter

-- Peter & Rita Forbes snipped-for-privacy@easynet.co.uk Engine pages for preservation info:

formatting link

Reply to
Peter A Forbes

Yup, Digimarc is the product and it is invisible but can be ID's by Paint Shop pro and the other registered products that are listed on their site.

I have today put a new index.htm page on the site with a note to the effect that all new stuff and eventually everything else will be digitally watermarked so there's no excuse, particularly as, and I was much surprised by this, most of the 'theft' seems to be from guys already in the engine hobby in some form or other.

I don't know how PTFE and Patrick Knight get on with their own image stocks, but it is a real pain to come across something on someone else's site that is quite obviously your own, be it text or a picture.

I spent last weekend and New Year going through the directories of Oldengine.org so weed out a few of the non-inhabited directories, and also to check that the sites that were there were actually usuable. Lo and behold, in one site I ran into my text on a diesel matter, quite badly taken with no acknowledgement and the same spelling error as in the original!

Peter

-- Peter & Rita Forbes snipped-for-privacy@easynet.co.uk Engine pages for preservation info:

formatting link

Reply to
Peter A Forbes

In message , Peter A Forbes writes

The problem is mainly with servers. Microsoft powered servers dont seem to have a problem if you make the anchor a mixture of upper and lower case and forget when you name the file called up by that anchor and don't use the same mix exactly.

Linux or Unix based servers are VERY case sensitive and will cause the image to fail under exactly the same set of parameters.

Reply to
George Hendry

Time to run Apache as a server and look up the use of .htaccess to avoid hotlinking (any number of web guides on how to do it)

-- Smert' spamionam

Reply to
Andy Dingley

I can't say that I've ever found any of my pictures where I wouldn't expect to, but then I haven't looked very hard... I'm not hugely bothered about the stuff I put up on Webshots, though as my albums have had over 20,000 hits, and over 1000 downloads then someone, somewhere, must be looking at them.

I am careful about who gets high resolution scans of literature and pictures, and my slide library is not on the web at all, though I am always happy to help out genuine enquiries. I don't have a web page with chunks of text on it, so this again isn't a problem for me. It is a shame hat people will steal work that is not their own, but I suppose its just human nature and was ever thus.

Back down to the shed now. Sadly the big end is done for on the 8hp Atomic, and I'm having to take the whole thing apart again to extract the crankshaft (Takes two people to lift it plus the main bearing housing...) and effect repairs. I may be bringing it along to the 1000 engine in June, it cauesd a lot of interesta couple of years ago,

Regards

Philip T-E

Reply to
ClaraNET

Regia has a big, information-rich website

formatting link
and we've always encouraged downloading for academic use, but have a (hopefully) scary policy on copyright theft of our images *if used for gain.* As far as we are aware, it's worked OK for the last ten years or so.

We occasionally find some of our deathless prose turning up on other sites, but a polite "I say old boy ........." has always had the desired effect so far.

As for my Webshots albums (163,775 hits and 788 downloads since May 2002), I keep the resolution fairly low, but in any case it has done me/us much more good than harm, getting filmwork and public events, school talks and sales to publishing houses.

Regards,

Kim Siddorn

Reply to
J K Siddorn

It's server issue - Unix platorms are case-sensitive, whilst Windows platforms will tell you that they are case insensitive, when they really aren't completely - some functions will preserve the case of a filename whilst others won't, and different webservers on a Windows platform may react differently.

Always match the case of a URL to the corresponding disk file - sometimes you can get away with this on Windows but why be sloppy and take the risk?

Another gotcha: people publishing web pages on a Unix platform will typically name their files with a '.html' extension because, after all, they are html files and not htm files. Windows users tend to use a total mixture, which is areally a hang-up from the MSDOS days when filenames were restricted to 3 letter extensions (hence the htm extension). Of course there's no such restriction in Windows these days, so for portability purposes always using .html makes sense and having a standard makes coding effort easier.

Spaces in filenames case no end of problems. Use underscores instead. A lot of server *and* browser software doesn't correctly handle the encoding and decoding of URLs with spaces in it, not to mention the headache it can create when using any scripting utilities at the OS level both client and server-side (say, backing up content on the webserver, or using a client-side utility to mirror a website)

cheers

Jules

Reply to
Jules

yes, I've been there too. Poses some quite interesting technical challenges!

It's also trivial in nearly all cases to remove such a waterwark without cropping, through the use of retouching tools. No, it's not a

1 minute job like cropping, but it can be done even on complex images with not much work. If you have valuable data then it isn't worth the risk and digital watermarking would be much more suitable.
[comments snipped]

Agreed with all of those!

One gotcha - and I don't know if it's relevant here - is that I've found scanning data in black and white significantly reduces the ability for OCR software to do a good job on the output should you want to ever do that at a later date; I've taken to scanning all my historical data as

8-bit greyscale as it greatly speeds up any OCR process in terms of the time needed to check and correct any mistakes. Storage is cheap enough these days that it isn't a problem.

cheers

Jules

Reply to
Jules

Was discussing this very issue with Tim in the States last evening, and he basically agrees with that, and uses grey scale for most of his scanning regardless of whether he is OCR'ing or not afterwards.

I bought a copy of the old Mullard Audio Valve amplifiers on ebay a while back, and Tim wanted a copy, so I shipped it direct to him from Oz and he is scanning/cleaning storing it back onto CD for me. I could have done it here but he has more spare time than me :-))

Peter

-- Peter & Rita Forbes snipped-for-privacy@easynet.co.uk Engine pages for preservation info:

formatting link

Reply to
Peter A Forbes

As I understand, it hotlinking is when one posts a web address and it goes all blue and underliney, so that when clicked upon it takes the clicker directly to the picture or whatever at the end of the link. This is why I asked if posting such things on NG was a no no.

I suspect that those in the know thought the question was flippant, but I am genuinely curious!

Reply to
Nick Highfield

Should have added that I assume the "bandwidth theft" bit comes in when the posted link leads to material that the poster does not own.

Reply to
Nick Highfield

I am successfully using .htaccess and it has reduced the problem to such an extent that I no longer consider it an issue. Before doing so hotlinks accounted for five of the referrers among the top thirty results for my site. The main point that I was trying to get across is that I find it sad that those who provide web sites as a free service are having to, either jump through hoops or incur extra expenses because of the greed or stupidity of a few.

Some theft such as hot-linking is easy to detect, those who simply copy images or the written word are much more difficult to trace. The recent incident where copyright material stolen from the Internal Fire site was a real eye opener as to just how brazen some spivs can be. Another less than desirable result which can be time consuming, is that giving site users an easy means of email contact will eventually result in a constant barrage of spam.

Reply to
Richard H Huelin

For my benefit, at least, could someone post a Janet and John explanation of:

-what hot-linking is

-why it is evil

-how you detect it thanks Roland snipped-for-privacy@petternut.co.uk

Reply to
Roland and Celia Craven

That'll be good as I only have the vaguest idea myself. I think I follow most of it, but now I'll see if I was right!

Regards,

Kim Siddorn

>
Reply to
J K Siddorn

AFAIK no one objects to links being posted on newsgroups and mailing lists, or even websites where the link is to the site. What bandwidth theft is is where a website uses an image on another website without permission, by using code in the page to fetch the image from the original website and display it as if it was part of the new website, rather than using a copy on the same webserver.

Obviously this would be a problem if you had a picture on your personal webspace which was referenced by a high traffic commercial website; every hit on that site would generate traffic on yours and you would soon be hearing from your ISP.

Reply to
Niall

Nope, I had that wrong! Thanks for the cogent explanation - now I get it!

Regards,

Kim Siddorn

>
Reply to
J K Siddorn

PolyTech Forum website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.