Digimarc® has provided this Watermarking Guide to assist you with any

questions you might have about our watermarking technology and how to get
the most out of it in processing your images. Use this Guide as a supplement to
the User's Guide for your specific Digimarc product; you'll find more detailed
information here about the digital watermarking technology underlying all of our
products, as well as tips for watermarking images in your specific environment
and for your unique purposes.

As you become more familiar with digital watermarking, we encourage you to

send any helpful tips or pointers that you discover to [email protected]. We'll
publish selected tips in future revisions of this Guide and on our Web site.

Introduction to Digital Watermarking

This section provides an explanation of digital watermarking and the copyright

communication system it offers, and how it can be useful to you. If you are
already familiar with the basics of watermarking, you can proceed to the
following main section, "General Watermarking Guidelines," for specific
instructions and tips on watermarking more effectively.

What is digital watermarking?

A digital watermark is best described by comparing it to a traditional watermark.

Traditional watermarks are added to some types of paper to offer proof of
authenticity. They are imperceptible, except when the paper is held up to a light
for inspection. Similarly, Digimarc digital watermarks are added to still images in
a way that can be seen by a computer but is imperceptible to the human eye. A
Digimarc watermark carries a message containing information about the creator
or distributor of the image, or even about the image itself.

A watermark is used to communicate copyright information about an image in

order to reduce copyright infringement. A person opening a watermarked image
in a Digimarc-supported image-editing application receives notification through a
copyright symbol ((c)) that the image contains copyright and ownership
information. The watermark provides a link to complete contact details for the
copyright holder or image distributor, making it easy for the viewer to license the
image, license another one like it, or commission new work.

Digimarc digital watermarks are imperceptible to the human eye, yet are durable
and provide images with a persistent identity. To help hide the watermark,
Digimarc varies watermark energy within the image so that it remains
imperceptible in both flat and detailed areas. The watermark is robust, surviving
typical image edits and file format conversions, remaining with the image in
digital and printed form-and is still detectable when an image is printed and then
scanned back into a computer.

Why should I watermark my images?

There are several main reasons for using Digimarc watermarks. One is simple
pride of authorship-the same reason that artists sign their paintings. Digital
images are especially prone to loss of authorship; witness the avalanche of
images posted daily on the World Wide Web, few of which have any reference to
the photographer or illustrator. With a Digimarc watermark, you can embed
information in an image that links the viewer to complete contact information
about who you are and how to find you.

Another reason is more important commercially: a Digimarc watermark

communicates the name and rights of an image's copyright holder. With this
information, the image consumer can quickly and easily contact the image
creator or distributor to license the work or commission additional work.

Overall, digital watermarking provides creators and distributors of images three

main benefits:

 Protect your valuable images by communicating your copyright

 Track down uses of your images on the Web
 Generate incremental revenue by embedding an ad in every image

If you're an image creator, embedding watermarks in your images allows you to

communicate your copyright in every image you distribute. In addition, when
you use Digimarc watermarking, you're effectively embedding an ad in every one
of your images, encouraging viewers to contact you if they like what they see.

If you're an image distributor or if you maintain a collection of images for a

corporation or museum, you should watermark your images any time they are
released externally (licensing to a magazine, posting on the World Wide Web,
adding to a stock image collection, etc.). Watermarking gives you the security of
knowing that no matter how or where your images appear, whether digitally or
in print, they carry your notice of ownership and a simple path to contact you
through Digimarc's registry. In addition, it simplifies the process of administering
usage rights, helps to communicate your brand throughout the market, and can
even help you capture secondary revenue by bringing interested buyers to your
doorstep.

How does the Digimarc watermarking system work?

Taken together, Digimarc's digital watermarking products and services form a

complete copyright communication system that image creators and distributors
can use to communicate not only copyright, but complete contact details to the
consumers of their works. This system provides the tools and capabilities to
embed watermarks in images, to detect and read watermarks, to link to
complete contact details or web site for the image creator or distributor (for
inquiring about usage rights, licensing, etc.), and to track uses of watermarked
images on the Web.

Watermarking software

Over one million copies of Digimarc's Plug-ins watermarking software are

currently in distribution. Leading image-editing applications, from companies like
Adobe, Cerious Software, Corel, Jasc software and Micrografx include Digimarc
watermarking as a standard feature.

These Digimarc-aware applications are capable not only of embedding

watermarks, but also of detecting watermarks already embedded in images.
When an image is opened or scanned into one of these applications, the Plug-ins'
auto-detection software quickly scans the image for the presence of a
watermark. If a watermark is present, the application displays a copyright
symbol (©) in the title bar of the image window, providing an instant visual cue
that copyright and ownership information are available by reading the Digimarc
watermark. The passive detection and proactive notification are key features of
Digimarc's copyright communication system.

By simply clicking a "Web Lookup" button, the user viewing an image can link
directly to the Digimarc registry, to obtain complete contact details for the image
owner or distributor.

In addition to the Plug-ins, Digimarc offers two other applications for

watermarking images. The Digimarc Batch Embedder provides batch processing
and other advanced capabilities for watermarking large image collections.
Digimarc's Watermarking SDK provides the same watermarking capabilities in a
set of libraries that can be integrated into other software applications.

The Digimarc online registry

The Digimarc registry is an online locator service for communicating copyrights

and authorship, which brings image consumers together with image distributors
and creators by providing critical contact details. This information communicates
copyright and facilitates image commerce. Through the registry, interested
buyers can contact content owners about licensing an image, licensing one like
it, or even commissioning new work.

Whenever a Digimarc-aware application alerts the user that an image is

watermarked, the user has the option of linking directly to the registry, where a
contact profile or web site for the creator or distributor of the image is
immediately available.

MarcSpider®

MarcSpider is a patented service that reports to image owners and distributors

where their watermarked images were found on the Web. This exclusive service
enables photographers, Web content developers, stock photography libraries,
corporations and other users and creators of digital images to discover both
authorized and unauthorized uses of their works migrating across the Web.

The core of MarcSpider is a search engine that crawls through the Web looking
for watermarked images. It scans hundreds of millions of pieces of information,
locating Digimarc watermarked images and reporting back to their owners where
and when they were found.

Copyright communication system

Of course, no copyright communication system can absolutely prevent the

stealing or unauthorized use of images; but the Digimarc system offers the
considerable advantage of placing a notification of copyright ownership directly in
view, where the user cannot miss it when viewed in one of the many image
processing programs that auto-detect DM watermarks. Without a watermark, the
viewer might never have known that the image was copyrighted. The
conscientious user will note the watermark and take appropriate steps to obtain
clearance before using the image; the Digimarc Registry provides an easy
pathway for doing so. MarcSpider, as the final component in the system,
provides feedback to creators and distributors as to where their images are being
used. What information can a digital watermark communicate?

A Digimarc digital watermark contains some or all of the following information:

 Digimarc ID-A unique number identifying the creator of the image.

 Copyright Year-The year or years in which the image was copyrighted,
or
Image ID-A unique number that you assign to identify the image,
or
Transaction ID-A unique number that you assign to identify the transaction in
which the right to use the image was granted.
 Restricted Use Attribute-If "Yes" or "On," indicates that the image is not freely
distributable and that licensing terms apply.
 Do Not Copy Attribute-If "Yes" or "On," instructs the person viewing the
image that it should not be reproduced in any form without contacting the
creator or distributor.
 Adult Content Attribute-If "Yes" or "On," indicates that the image contains
subject matter that is suitable for adults only; helpful in preventing minors
from viewing unsuitable content in online applications. Through the
information found in the watermark, anyone with a Digimarc "reader"
(available in the stand-alone freeware application ReadMarc(tm) as well as in
the Plug-ins) can obtain complete contact details about an image's creator
and/or its distributor, making it simple to license the image, license another
one like it, or commission new work.

Note: Not all of Digimarc's products support all of these image attributes. Refer
to the User's Guide for your individual product for specific information.

General Watermarking Guidelines

This section provides specific instructions and tips for integrating digital
watermarking into your image processing workflow and creating the most
durable watermarks possible.

What types of images can be watermarked?

Generally, any image that can be opened by the application that is using
Digimarc technology can be watermarked. This section provides additional
information on specific aspects of images relating to watermarking.

Bitmap images vs. vector images

Digital watermarking works most predictably with bitmap images, which can be
either color or grayscale. Vector or line-art images can also be watermarked as
long as they are first converted to a raster or bitmapped format.
File formats

A Digimarc watermark can reside in any file format supported by the Digimarc-
aware host application. For example, TIFF, PICT, JPEG, GIF, PNG, PSD and BMP
images can all contain a watermark. Because a watermark is carried by the
pixels that make up an image, it survives even when the image is converted
from one file format to another.

Note: Not all of Digimarc's products support all of these file formats. Refer to the
User's Guide for your product for specific information.

Color space

The watermark is placed in the luminance channel of an image, so it is color

space independent. A watermark can be placed in RGB, CMYK, LAB or grayscale
images, and will survive when an image is converted from one color space to
another.

Note: The RGB, CMYK, LAB and grayscale color spaces are not supported by all
image-editing applications, so refer to the User's Guide for your application to
see what color spaces it supports.

In some image-editing applications, no filters are available when working with

indexed-color images (such as the GIF format); this will prevent you from using
the Plug-ins filter to embed a watermark in such an image. The solution is
simple, however: just convert the image to RGB, apply the watermark, and then
convert back to indexed color.

Image variations/randomness

Because of the way that Digimarc digital watermarks are embedded, an image
being watermarked must not be composed mostly or entirely of a single flat color
if the watermark is to be imperceptible. Rather, the image must contain some
degree of variation or randomness.

Since the watermark is intended to be information that the computer can see but
the viewer does not, Digimarc uses a patented technique called perceptual
adaptation when embedding watermarks. This is simply a technical way of saying
that when embedding a watermark, the software identifies areas of the image
that are highly detailed or very flat, and adjusts the intensity of the watermark
accordingly. In flat areas the watermark intensity is decreased, while in detailed
regions the intensity is increased. If an image is composed mostly of flat areas, it
may be difficult to conceal the watermark.

What size of image can be watermarked?

To embed a watermark and read it reliably later, the Digimarc technology

requires a minimum number of pixels with which to work. If you don't believe
that your image will be modified or compressed prior to its actual use, its size
can be as small as 100 x 100 pixels. However, if you expect that your image
might be cropped, rotated, compressed or otherwise modified after you
watermark it, Digimarc recommends a minimum size of 256 x 256 pixels. There
is no upper limit on image size for watermarking.

Note: You may find that some smaller images actually watermark successfully,
and even that some larger images do not. Such variance in successful
watermarking is a factor of the variation/randomness of the image (as discussed
in the previous section), the durability setting used when embedding the
watermark (see "Watermark durability vs. visibility" later in this Guide) and the
parameters used in applying compression, if any (see "Image compression" later
in this Guide).

Workflow overview: when to watermark images

As you prepare an image for its final use, you may take it through a number of
different transformations. You may save the image in multiple resolutions, and
you may perform a number of the edits discussed in the "Watermark Survival"
section of this Guide, such as color correction, cropping, rotating, scaling, etc.
Watermarking should be one of the very last processes you apply to your image,
except for compression. The recommended workflow sequence for watermarking,
then, is as follows:

1. Make all necessary modifications to your image until it has the desired final
appearance.
2. Embed the watermark.
3. If the final image should be compressed, compress by converting to JPEG or
GIF format.
4. If the image is intended for printed output, perform the color separation now.
5. Read the watermark and use the signal strength meter to verify that the
image contains a watermark of sufficient strength for your purposes.
6.Publish the image.

Setting watermark attributes

When you're ready to embed a watermark, you will need to set a number of
parameters for the information the watermark will contain. The following list
summarizes these attributes.

 Digimarc ID. A unique number that identifies you when you register with
Digimarc as an image creator. This unique ID is linked to the Creator's
contact profile.
 PIN. A Personal Identification Number provided to you by Digimarc for use in
validating your Digimarc ID when you submit it.
 Copyright Year. A single year or two years separated by a comma; must be
between 1922 and the current year, or
 Image or Transaction ID. A unique number that you assign to identify the
image or transaction (range is 1-16,777,215).
 Restricted Use. When selected, indicates that the image is copyrighted and
subject to restricted use.
 Do Not Copy. When selected, indicates that the image's creator and/or
distributor has specified that the image should not be copied without
permission.
 Adult Content. When selected, indicates that the image contains Adult
Content.
Setting target output

Digimarc adjusts an embedded watermark slightly depending on whether the

image will be used electronically or in printed form. You should select the
intended target output for your image--monitor, print, or web.

Watermark durability vs. visibility

Digimarc's goal is to embed watermarks that are both imperceptible to the

human eye and durable. This can be a delicate balancing act, since watermark
durability and watermark visibility are directly related. An increase in watermark
intensity, or energy, increases the durability-but it also increases the visibility of
the watermark. When a watermark becomes visible, it can appear as a slight
texture or graininess in the image.

Using the durability/visibility setting

The default watermark intensity setting found in Digimarc watermarking products

has been chosen to strike a balance between watermark robustness and visibility
in the majority of images. However, this setting is by no means "one size fits
all." Digimarc provides controls so that you can increase or decrease the
watermark intensity level yourself to achieve the desired balance between
watermark robustness and visibility in your watermarking projects. The setting
you select will depend on the intended use of the image and on the goals you've
set for your watermarks. For example, it may be quite acceptable to use a higher
watermark intensity setting with JPEG images posted on a Web site. The higher
durability helps to assure the persistence of the watermark, and the increased
visibility will often not be noticeable with medium-resolution JPEG images.

Digimarc recommends that you try various watermark intensity settings as part
of your testing process to see which setting works best for the majority of your
images. The goal is to find the balance between visibility and durability that best
serves your image and provides the persistent identity that will protect it. Use
the signal strength meter to get feedback on the watermark strength.

Note: If you use both the Plug-ins and Batch Embedder, you may notice that
while the Plug-ins offer a scale of 1-4 for watermark intensity, Batch Embedder
scale goes from 1-16. The different scale in Batch Embedder simply allows a
greater degree of precision in setting watermark intensity.

Other factors affecting watermark durability

Along with the intensity setting that you choose when embedding a watermark,
the durability of a watermark is also affected by the following factors:

 Image variations/randomness-As discussed earlier in the "Image

variations/randomness" section, the successful embedding of a watermark is
dependent on the variation and randomness present in the pixels making up
the image. For example, if you are working with an image that contains more
flat color regions than detailed areas, you may want to choose a higher
watermark intensity so that the watermark will overcome the limitations of
 the specific image. This may result in a more visible watermark, but in some
situations that is an acceptable trade-off, as mentioned above.
 Image size-See the earlier section, "What size of image can be
watermarked?"
 Compression-Saving the watermarked image in a compressed format may
affect the durability of the watermark.

See the "Watermark survival" section later in this Guide.

Using the signal strength meter

A significant aid in determining the durability of an embedded watermark is the

signal strength meter that is available in the Digimarc Plug-ins and Batch
Embedder. After you've embedded a watermark, you can immediately check its
intensity visually with the signal strength meter and decide whether or not the
watermark has a sufficient durability to survive your intended use of the image
(you can do this automatically by selecting the "Verify" option when embedding a
watermark). The signal strength meter is also available when you are reading a
watermark you've previously embedded. When you are experimenting with
different ways of preparing your images for watermarking and of putting them to
use once they've been watermarked, we encourage you to use the signal
strength meter as a "quick-read" gauge of how your various settings are
working.

Watermark Survival

As mentioned earlier in this Guide, a digital watermark is robust, surviving

typical image edits and file format conversions, remaining with the image in
digital and printed form--and is still detectable when an image is printed and
then scanned back into a computer. This section provides details on the many
factors that can affect watermark survival, and includes tips for obtaining
optimum results when embedding your watermarks.

Image compression

In general, a Digimarc digital watermark will survive image compression, but the
survival is dependent on several factors. Lossless compression, such as Compact
Pro, StuffIt(tm) and the .ZIP format, does not affect a watermark's survival at
all, because no image data is sacrificed to create the compressed version. On the
other hand, because lossy compression methods such as JPEG actually remove
some image data in order to decrease file size, they can have varying effects on
watermark survival. The following factors will influence the impact that lossy
compression has on watermark survival:

 Level of image compression-Lossy compression degrades the image to some

extent, depending upon the quality setting chosen when saving in compressed
format; most watermarks will survive as long as a moderate level of
compression is used (see below for more detail).
 Visibility/durability setting used when watermarking-The higher the intensity
setting, the better the chances the watermark will survive compression.
Again, a higher-intensity watermark provides more data to survive
compression. Since the visual quality of compressed images is often
 somewhat compromised anyway, generally a higher watermark intensity
setting yields quite acceptable results.
 Image size-The greater the number of pixels in the image, the more the
watermark can be repeated throughout it; the recommended minimum size
for an image that will be compressed is 256 x 256 pixels. The larger the
image, the better the watermark will survive compression.
 Randomness of image data-As discussed in the earlier section "Image
variations/randomness," the more randomness and/or color variation in an
image, the better; a flat color space with little gradation may not survive well,
while an image with more detail and contrast will fare better. Since the
watermark is applied more strongly within areas of high contrast or variation,
an image that contains more contrast and/or variation than others will contain
more watermark data and thus stand a better chance of surviving
compression. Note that vector images, line art or text that was converted to a
bitmap and watermarked may not survive compression well due to a relative
lack of randomness.

Unfortunately, images vary so widely in their detail, randomness and other

qualities that it is impossible to provide universal guidelines for applying
compression to all images. A watermark in one image may survive a
compression ratio of 20:1, while another may not survive anything above 10:1.
In addition, image-editing applications vary widely in how they apply
compression as well as how they present the available compression options to
the user. For example, one application may offer a compression scale of "Low,"
"Medium," "High" and "Maximum" Image Quality, while another may provide a
numerical scale of 0 through 255. As a result of these inconsistencies, it is
difficult to provide specific recommendations for which compression settings you
should use with your watermarked images; however, a setting that yields
approximately a 10:1 reduction in file size can be considered moderate, and
should ensure watermark survival in most images.

If you are using Adobe Photoshop for image editing and embedding watermarks,
Digimarc generally recommends using Level 4 (Medium Image Quality) or higher
when compressing images. If you are using an image-editing tool other than
Photoshop, you will probably want to stay within the same general range.

Note: Besides JPEG, there are new forms of lossy compression such as Wavelet
and Fractal. Because these compression formats are not yet as standardized as
JPEG, Digimarc has not done extensive testing of their impact on watermark
survival as of this writing. If you want to use either approach, we recommend
that you experiment thoroughly and be sure to test your images for watermark
readability before distributing them.

As you work with compressed images and test the readability of watermarks
embedded in them, you may want to experiment with the interaction between
the visibility/durability setting for watermarking and the level of image
compression. Depending on the contents of your image (detail, color palette,
etc.), these settings will have varying impacts on the watermark's survivability.

Resampling images

In some situations, you may wish to have multiple copies of an image at

different resolutions. For example, if the same image will be used both on the
Web and in print, you may want a 100 DPI version for Web use and a 300 DPI
version for print. To accomplish this, you resample the image, changing its
resolution in your image-editing application after using the "Save a copy..." or
similar command.

When you are working with an image that you will be resampling in multiple
resolutions, always resample before embedding a watermark. This will mean
conducting more watermarking operations than if you embedded only once prior
to resampling, but your watermarks will be much more durable using this
approach.

Watermarking images for digital output

This section identifies a number of issues that are unique to watermarking

images for digital delivery on the World Wide Web or any other online medium.

DPI settings

When watermarking an image whose sole use will be on the Internet, be sure to
resample the image to the proper DPI setting for this medium (either 72 or 100
DPI) before you embed the watermark. Correctly matching the watermark's DPI
setting to the image's final resolution will ensure the survival of your embedded
information.

GIF and indexed color formats

If you work with GIF images regularly, you are probably accustomed to reducing
the color depth of an image in order to decrease its file size and accelerate its
loading speed for Web use. When the image you're working with is watermarked,
Digimarc recommends that you use a color depth of no fewer than 216 (this is
the number of colors in the standard "Web-safe" palette that ensures equivalent
colors on all computing platforms). In some image-editing applications, no filters
are available when working with indexed-color images such as the GIF format;
this will prevent you from using the Plug-ins filter to embed a watermark in such
an image. The solution is simple, however: just convert the image to RGB, apply
the watermark, and then convert back to indexed color. You should be aware
that if you repeatedly change the image's color palette (decrease the number of
colors, or change colors in the palette itself) after the watermark has been
embedded, this may decrease the durability of the watermark, as well as the
quality of the image.

Watermarking images for print output

This section discusses several considerations involved in watermarking images

whose intended use is to be included in printed output.

Image size

For reliable watermarking of images that will be printed at 300 DPI or greater,
Digimarc recommends that your original images be at least 750 x 750 pixels in
size, or 21/2" x 21/2".
Printing devices

To ensure that watermarks survive through printing and scanning, Digimarc

recommends that watermarked images be printed on high-quality printing
devices, ranging from desktop-type printers (laser, thermal wax, dye
sublimation, etc.) to high-end web presses. Lower-quality printers, such as
consumer-level inkjet models, may produce less reliable results. For best results,
print on proof or coated stock.

DPI settings

When preparing an image that is intended for printed output, be sure to match
the image's DPI setting to the resolution of the target printer. This is usually 300
DPI, including line screen printing, because that is the effective resolution after
rosette patterning. In general, you'll achieve the best results by using the
highest DPI setting supported by the intended output device. Remember to
always resample your image to the desired resolution before embedding a
watermark.

Color space

When preparing an image that will be printed, you can embed your watermark in
the image when the image is in either the RGB or CMYK color space.

UCR/GCR settings

Typical UCR (Under Color Removal) settings of 280, 300 or 320 do not affect
watermark survival; typical GCR (Gray Component Replacement) settings also do
not affect watermark survival.

On-screen appearance

When viewing your watermarked image, avoid judging its quality by its on-
screen appearance, since the screen is a very low-resolution view (only 72 or
100 DPI). To get an idea of how the watermarked image really looks, you'll need
to print it on a high-quality printer, as discussed earlier in this section.

Unsharp Mask

The recommendations for applying the Unsharp Mask filter to an image differ
depending on whether the image is intended to be used for digital or printed
output.

When preparing an image for Internet use or other digital output, if an Unsharp
Mask filter is necessary, always apply it before embedding your watermark.
Because Unsharp Mask increases the apparent level of detail in an image, and a
watermark is embedded most intensely in the detailed areas, the Unsharp Mask
process enhances the areas where a watermark is most intense.

For printed output, embed the watermark first, and then apply an Unsharp Mask
filter of 75% or greater-this is very important in reinforcing the watermark's
durability through the printing process. In this context, the Unsharp Mask filter
actually intensifies an embedded watermark because it increases the apparent
level of detail in the image, and the watermark is already most intense in its
detailed areas.

Again, avoid judging the final output quality by the appearance of the image on
your screen. Remember that the printing process naturally blurs the image
somewhat. The Unsharp Mask counteracts this and provides enhanced
watermark survival.

Scanning watermarked images

One of the key features of Digimarc digital watermarks is that they can survive
even when an image is printed and then scanned back into digital form. This
section details approaches you can take to scanning watermarked images in
order to achieve the best results.

Scanning devices

In order to obtain acceptable results, the scanner used with watermarked images
must be a flatbed type or better-handheld or desktop sheet-fed scanners are not
precise enough to represent the watermark accurately. Scanning watermarked
images works acceptably even on low-cost flatbed scanners like the HP ScanJet
and better.

Scan settings

For the highest reliability in scanning watermarked images for use in digital form,
Digimarc recommends that they be scanned in millions of colors rather than a
smaller color palette. Also, do not use the "Sharpen" option available with some
scanners during the scan.

Correcting scanning problems

Sometimes the initial scan of an image will fail to show a watermark. In such
cases, try these options for recovering the watermark:

 Make sure that the DPI at which you are scanning matches the DPI at which
the watermark was embedded. If it does not, change your scanning settings
and re-scan.
 Rotate the image 90 degrees and re-scan; the input data will be slightly
different due to factors like stepper motor jitter, and may result in the
watermark data being read in more accurately.
 If Digimarc's reader software fails to find a watermark in a scanned image, try
blurring the image slightly and re-checking with the reader; this will
sometimes make the watermark more readable by reducing heavy moiré
patterns.

Scaling
A digital watermark will generally survive scaling, but Digimarc recommends
staying within a range of .6X to 2X to preserve the watermark.

Cropping

Because the watermark is repeated throughout the image, removing portions of

the image by cropping will generally not affect the watermark, provided that the
final image exceeds the minimum size discussed above. If the image is cropped
to less than 256 x 256 pixels, the watermark may not survive.

Rotation

Unlike other systems that embed copyright information in images, a Digimarc

watermark remains intact when the image is rotated by any number of degrees.
The rotation can be as small as 1 or 2 degrees, which can often occur when an
image is scanned, or as large as the most extreme arbitrary rotation, with no
effect on watermark survival.

Effects filters

The watermark survives most effects filters, but the general rule is that the
survival of the watermark is linked to the visual quality of the image. If an
effects filter is applied at an extreme setting such that the quality of the image is
compromised (particularly distortion-type effects such as Blur, Twirl and Morph),
then the watermark may no longer be readable from the image.

Working with layered images

When the image you want to watermark contains multiple layers (a feature
supported by many image-editing applications), watermarking it in that state
would apply the water- mark to the selected layer only, rather than to the entire
image. To avoid this problem, you must flatten the image (combine all of the
desired layers into a single image) before watermarking. To flatten the image in
Adobe Photoshop, for example, choose "Flatten Image" from the pop-up menu
on the Layers palette. In some situations, you may want to create a
watermarked version of the current image but save its layered version for later
use; most applications include a "Save As..." or "Save a Copy..." command on
the File menu for this purpose. After saving a separate copy of the file, you can
open the copy, flatten the image, and apply your watermark.

Combining digital watermarks with visible watermarks

When you are creating digital comps for an agency or prospective customer to
use on a "For Position Only" basis, you may want to apply a visible watermark in
addition to a Digimarc imperceptible watermark, to prevent their use in a final
piece. If you use such a visible watermark, be sure to apply the digital
watermark last, since in the reverse order the visible watermark could possibly
disrupt a significant number of pixels, potentially changing the image too
drastically and removing the digital watermark. Also, keep in mind that when
applying a digital watermark to a comp you can use a higher-than-usual
durability setting if you like, because some degree of image variation is
acceptable in this situation.

Working with composite images/montages

An image that is actually composed of multiple discrete images presents a

unique set of circumstances where watermarking is concerned.

If you create a montage consisting only of images that you created, you can opt
to watermark each of those images prior to adding them to the montage; or you
can wait and watermark the entire montage when it is complete. When a user
views your montage, he/she can check for watermarks by selecting individual
images within the montage using the marquee or lasso tool and choosing the
"Read Watermark" command. If you use this approach, be sure that the
individual images are no smaller than the minimum 256 x 256 pixels.

If you create a montage using images created by others, bear in mind that your
image-editing application might alert you to copyright and authorship
information stored in those images via digital watermarks. You should consider
this information carefully when evaluating whether you should use such images
in your work, or whether you should contact the copyright holder(s) before using
them. Remember that you cannot embed a watermark in an image that already
contains one. Attempting to do so will result in an error message.

Case Studies

This section presents three fictional "case studies" to help you see how you
might apply some of the techniques and concepts explained in this Guide.

Using photographs on a Web site

A photographer wants to post her own images on her Web site. In order to
present the photographs in a variety of layouts, she wants to post each image in
four different sizes: 600 x 440, 300 x 220, 150 x 110 and 75 x 55 pixels. She is
planning to post the images in JPEG-compressed format.

Issues to consider

 Resizing watermarked images can adversely affect watermark durability.

 The two smallest images are below the recommended 256 x 256 pixels for
compressed images.

Recommendations

 Resize the original image before embedding a watermark. Create the four
desired sizes from the original, and save the resized images as separate files.
 Since the Digimarc-recommended minimum size for watermarked images that
will be compressed is 256 x 256 pixels, any watermark embedded in the two
smallest sizes most likely will not survive after the images have been JPEG-
compressed. The watermark in the smallest image will certainly not survive.
Process

1. Working from the original image, resize it to the four desired sizes and save
those versions as separate files.
2. Open the two largest images for watermarking, and perhaps the 150 x 110
pixel image. Do not watermark the smallest image, since it is too small for
effective watermarking and probably has little commercial value anyway due
to its size.
3. Embed the watermarks at intensity level 2 (level 8 if using Batch Embedder).
Optionally select the "verify" option in order to check intensity using the
signal strength meter; this may be most important for the 150 x 110 pixel
image, since it is below the minimum recommended size. If you are not
satisfied with the strength of a watermark, revert to the resized,
unwatermarked version and then re-apply the watermark at a higher intensity
level.
4. Compress the watermarked images, setting compression at a level that favors
image quality over image size. If you are using Adobe Photoshop, a
compression level or 4 (medium image quality) or greater is recommended.
5. After saving the images in compressed format, close the files and then reopen
them. Choose "Read Watermark" from the Filters menu and use the signal
strength meter to see if the watermarks are strong enough-ideally they
should be in the yellow or green range. If the meter shows watermark
strength in the low yellow or red region for any image, you may want to
revert to the original image and attempt to increase the watermark's
strength, using the flow chart provided in the earlier "Watermark Survival"
section.

Watermarking digital images for print

A member of a magazine's production staff is given a digital image file and asked
to watermark it for print output. The file is a grayscale image; its size is 1" x
11/2", and its resolution is 72 DPI. The magazine in which the image will appear
is a 4-color publication.

Issues to consider

 The image's 72-DPI resolution is lower than the standard 300 DPI used for
printed materials.
 The image size of 1" x 11/2" (or 72 x 108 pixels at 72 DPI) is smaller than
the minimum recommended size for watermarking, which is 21/2" x 21/2" (or
750 x 750 pixels at the standard 300 DPI resolution).
 The printing process will likely use a 150 line screen, which is standard for
magazines. Recommendations
 Resize the image so that it is above the minimum recommended size for
watermarking.
 Resample the image to the standard 300-DPI resolution.
 Watermark the image only after these changes have been made.

Process
 1. Use the image-editing application's "Image Size" or similar command to resize
the image to 21/2" x 33/4" (preserving the proportions of the original image).
2. Resample the image to 300 DPI; the image size should remain the same,
while the resolution changes.
3. Make any desired additional modifications to the image-color correction,
cropping, filters, etc. Verify that the image is ready for production.
4. Watermark the image at intensity level 3 (level 12 if using Batch Embedder).
5. Apply the Unsharp Mask filter at 75%.
6. Choose "Read Watermark" from the Filters menu and use the signal strength
meter to see if the watermarks are strong enough-ideally they should be in
the yellow or green range. If the meter shows watermark strength in the low
yellow or red region for the image, you may want to revert to the original
image and embed the watermark at a higher intensity level before applying
the Unsharp Mask filter.
7. Output the image to film.

Watermarking JPEG-compressed images

A photographer wants to watermark his compressed images. The images are

already in JPEG format and posted on his Web site. All of the images are 330 x
280 pixels.

Issues to consider

 Watermarking an image that has already been saved in the JPEG-compressed

format can result in image quality degradation-you are effectively
compressing the image twice.
 Image size should not be an issue, as the 330 x 280 pixel size is over the 256
x 256 minimum. Recommendations
 For best results and the most durable watermark, go back and apply the
watermark to the original, uncompressed image files (TIFF format, for
example), then compress the images.
 If it is not possible to work with the original images, the JPEG-format images
can be watermarked-but this will likely result in less durable watermarks and
decreased image quality.

Process-Watermarking the original, uncompressed images

1. Watermark the original image files using at least the default watermark
intensity of 2 (level 8 if using Batch Embedder), or a higher setting such as 3
or 4. Optionally select the "verify" option in order to check intensity using the
signal strength meter. If you are not satisfied with the strength of a
watermark, revert to the original, unwatermarked version and re-apply the
watermark at a higher intensity level.
2. Compress the watermarked images, setting compression at a level that favors
image quality over image size. If you are using Adobe Photoshop, a
compression level of 4 (medium image quality) or greater is recommended.
3. After saving the images in compressed format, choose "Read Watermark"
from the Filters menu and use the signal strength meter to see if the
watermarks are strong enough-ideally they should be in the yellow or green
range. If the meter shows watermark strength in the low yellow or red region
for any image, you may want to revert to the original image and attempt to
 increase the watermark's strength, using the flow chart provided in the earlier
"Watermark Survival" section.
4. Post the images on the Web site.

Process-Watermarking JPEG-compressed images when originals are

not available

1. Watermark the JPEG files using the default intensity level of 2 (level 8 if using
Batch Embedder), or possibly 3.
2. After watermarking, save the images in JPEG format again. Remember to
favor image quality over file size when choosing a JPEG compression level.
You are effectively compressing the images twice, which may degrade image
quality at higher compression levels.
3. After saving the images in compressed format, choose "Read Watermark"
from the Filters menu and use the signal strength meter to see if the
watermarks are strong enough-ideally they should be in the yellow or green
range. If the meter shows watermark strength in the low yellow or red region
for any image, you may want to revert to the original image and attempt to
increase the watermark's strength, using the flow chart provided in the earlier
"Watermark Survival" section.
4. Once you are satisfied with both the watermark strength and the visual
quality of the images, post them on the Web site.