Where is the watermark?
Digital watermarking is often seen as a light but visible mark within an image, and can include a logo or URL to show copyright ownership. It is there for everyone to see (and remove). Invisible watermarks take this concept a step further and make the watermark less obvious and harder to remove. Digimarc is an invisible watermarking filter for Photoshop and is automatically included as a default. It is supported by a third party, Digimarc, and requires a subscription to take full advantage of it’s features. Its purpose is to embed an invisible watermark into an image to protect against copyright infringement and track the use of the image on the web. Digimarc works on two fronts. It embeds an invisible (strong) watermark into an image by encoding information (Author name and creation date) into a noise tolerant image’s least significant bits. That information is also stored in a database and can be cross-referenced against other copies of the same image all over the web. If an unauthorized use is discovered, the original author can prove who they are beyond any reasonable doubt and call out the infringing party.
I had a problem with copyright infringement of my own and thought Digimarc might be a perfect solution, but discovered that Digimarc’s service would not access the virtual world server where my texture images were stored. This roadblock inspired me to come up with my own type of invisible watermark using Photoshop to regain control over both the stored data, storage method, and file format. It is also important to note that the Digimarc filter can still be used to embed a second watermark into the same image if needed for distribution on the web.
Here are the steps I use for invisible digital watermarking in Photoshop:
- Give the original image a New Channel for encoding information. This converts it to a carrier file.
- Generate a second black and white image of the same dimensions and place multiple copies of a QR code in it. This is the information file which can carry contact information, a URL, or any message.
- Run a Gaussian Blur on the QR code to soften the edges so it will blend well with the carrier file image. Test this with a code scanner, then Copy and Paste it into the New Channel in the carrier file.
- Select part of the carrier image using the newly created Alpha Channel, Copy, then Paste it as a New Layer. This produces a second layer with only the coded portion of the image. It should be completely invisible at this point.
- To embed the code as imperceptibly as possible, run a Hue Saturation adjustment (altering the image hue by no more than five steps) and add 3% noise into the image with a noise filter. More alterations can be added if needed, but the general idea is to cover the coded layer with noise that uniformly alters the pixels slightly through all colors and shades.
- If the effect is too harsh the Opacity of the layer can be reduced to make the effects more subtle.
- Copy Merged, Paste the coded image into a New File, Flatten it and Save it to any popular file format (JPEG, PNG, Targa)
- Calculate the Difference between the original master file and the encoded file, Copy Merged, Paste into a New Layer. Run an Equalize adjustment on that layer to reveal the code again.
- To make the code readable under higher compressions, copy and compress the master file, then repeat steps 7-8.
This steganographic method may not be as sophisticated as Digimarc’s, but works for my application needs and gives a stealthy tracking mechanism for policing my work on a virtual world server hosting lossless JPEG 2000 file formats. It also survives moderate resizing, cropping, and compression. Best of all, it can be done entirely within Photoshop’s tool set without need of a third party add-on or a subscription.
Coded Cat Image
Raw extracted QR code
Extracted and enhanced QR code (scans well)
Step-By-Step Stage 1: The color comp
This comprehensive color mock-up lays out the placement and style of major elements for the cover design. The art directors notes are laid out in the margins as suggestions to follow. As the design progresses you will see adjustments made to some of the typographical elements at the bottom and top right of the design as well as some background elements. A decision to focus mainly on a customized type font for the word “Cardinal” carried off a movie poster look and feel in a cleaner fashion for a magazine sized media guide layout.
Step-By-Step Stage 2: Digital test shots
Here are two examples from several low resolution test shots produced with various poses. Ball position, stance, ambient lighting, cast shadow, facial highlights, and backdrop got adjusted at this stage to produce the best possible results in preparation for a final photo shoot.
Step-By-Step Stage 3: Final photo set
The final photo set used in the design incorporated the best qualities of all critera. Each photo had its background removed with a third-party filtering tool called Mask Pro, which greatly speeds up a time-consuming hand editing process if done with standard tools.
Step-By-Step Stage 4: Additional photo elements
Smaller photo elements provide background objects for better depth in the composition. Exploded basketballs, pulled from a previous “Towers of Power” project done for the same client, and a rusted backboard, give better depth for the cascading elements added later in the design.
Step-By-Step Stage 5: Players positioned in the foreground
This group of layers build from several hue and saturation, levels, color balance, and shadow layers linked to alpha channels. All the layers except the original player masks remain invisible in this slide to show contrast with the next slide which reveals the difference these adjustments have when turned on against a white background.
Step-By-Step Stage 6: Shadow and glow
All color, level, saturation, and glow layers are on in this slide. See how they interact differently against a cardinal red background in the next slide.
Step-By-Step Stage 7: Background & cast shadows
The cardinal red background and cast shadow layers are now all visible. The adjustment layers turned on in the previous slide are now interacting with all layers placed beneath them. The glow and cast shadows will interact naturally with sub-layered elements and change accordingly as more textural elements get introduced into the design.
Step-By-Step Stage 8: Asphalt & shadow color
The asphalt texture and shadow color begin to build up solid foreground. The asphalt, rendered from a seamless texture, repeats once across the spine. The color layer in the cast shadow has been re-applied as a color overlay from the original photo to interact with the subtle colors in the asphalt, giving it a slightly purplish hue.
Step-By-Step Stage 9: Background elements
More background elements introduce further expansion of the image depth. All of the background elements have a severe level adjustment that suggests they are partially obscured by a thick haze. Notice the mist-like texture building up at the ground level.
Step-By-Step Stage 10: Final background elements
Several layers have now been applied affecting everything from the shadow in the foreground to the mist and glowing elements in the background. New building elements have also been applied to the background. These were not in the original color comp because they were a client request midway through the design. Digital illustration makes requests like this easier to work into a design than they would otherwise with a more traditional method.
Step-By-Step Stage 11: The cascade
A familiar cascade of code, rendered against the cardinal background, adds a final touch of depth to the composition. In this case the text is not code, but the school’s name and season details. Each spell out details vertically and repeat hundreds of times across the background.
Step-By-Step Stage 12: Final copy
The last group of layers added, and the only group not flattened for delivery, was the copy. The “Matrix” font did not look good with the Cardinal title. Instead, this font was custom designed in Illustrator by converting a normal style font to outlines for the redesign of various type elements. This allowed for font designed specific to the word “Cardinal”.
Step By Step Stage 13: Final design & printed guide
All the layers and design elements (featuring the three most senior players) come together to produce a visually grabbing cover image designed to inspire potential college basketball recruits. The deliverables for this project included two high-resolution Photoshop files (one containing a separate copy layer for the printer to adjust) and an Illustrator file of the copy on the back cover for last-minute schedule changes, or for the option to produce a text knockout.