Follow these guidelines to get the best results when using the Digimarc Discover™ Online Services Portal.
This section provides you with some simple watermarking guidelines. As you gain experience, you will undoubtedly refine these recommendations for your use.
- A busy image hides a watermark well, you will be able to easily apply a strong watermark signal without impacting your image quality.
- Areas of bright highlights or dark shadows do not accept watermarks well. In highlights, where color coverage approaches 0%, there are too few pixels to modify to add the signal. In shadows, where color coverage approaches 100%, the ability to modify pixels is essentially halved because of ink saturation. You may need to increase the strength setting using the slider to improve watermark detectability. You can also strengthen the watermark in highlights, but this usually negatively impacts visibility.
- For the same reason that highlights do not accept watermarks well, pure white areas — those with 0% color coverage — cannot be watermarked at all. If you must watermark a pure white area, consider adding a light color tint into larger white areas. It will need to be dark enough for your printer to reproduce for sufficient color coverage to embed a detectable watermark. Unfortunately, the watermark may become more visible with the added detectability.
- Identify areas for different watermark strengths by roughly the same criteria you use for unsharp masking in image editing software. For example, faces will probably require a lower watermarking strength; areas with lots of detail can take a watermark of higher strength.
Examples of images well-suited to digital watermarking
Below are several images that are good candidates for watermarking due to these characteristics: All three images have textures that will effectively absorb a watermark. In the beach photo, the sand, the water on the beach, and the clouds offer adequate texture. The dog, the trees, and the lawn provide texture in the dog photo. While there is an expanse of red in the photo of the girl, it's not a flat tint. The bedspread contains plenty of texture for embedding the watermark. The colors in all three images are predominantly in the midtones, with only some small areas of very light and very dark colors.
Image selection guidelines
| Better | Characteristic | Maybe Problematic |
| Higher contrast | Contrast — High contrast provides more visual variety to make a watermark less visible. | Lower contrast |
| More mid-tones, lots of colors | Tonal Range — Areas of high or low ink density provide less room to add or remove ink as needed to embed a watermark. | More highlights and shadows, single color cast |
| More texture | Texture vs. Flat Tints — Areas of texture provide more visual variety to make watermarks less visible. Flat tints accentuate the effects of a watermark on visibility. | More flat tints |
| Less white space | White space — White space lacks the ink required for modification to embed a watermark. | More white space |
| Less text or vector graphics | Text or vector graphics — Text and vector graphics should be avoided. Straight edges and lines reveal the effects of watermarking, and human perception is too attuned to text and graphics not to notice these effects. | More text or vector graphics |
| Less color saturation | Color Saturation — Too much saturation in a single channel reduces the room to add ink as needed to embed a watermark. | More color saturation |
Color spaces supported
- CMYK — The watermark is optimized for printing commercially in color images. Commercial color printing typically involves using Cyan, Magenta, Yellow and Black inks, often referred to as CMYK or 4C (color) process. CMYK images prepared for commercial printing, especially SWOP US Web standards, should watermark well if they meet the image selection guidelines.
- RGB — Red, Green, and Blue light waves are used by displays to project images and by digital cameras to capture images. RGB has a much greater range of colors than CMYK, which may or may not be available when printed. The process of converting an RGB image into a printed image varies by each user's set-up. Due to this variation we cannot calibrate the embedder to provide consistently optimal results with RGB images.
- Grayscale — A single channel image has only black and white data available to be digitally watermarked. Often watermarks in grayscale images may be more noticeable. This is due to less available image data for the watermark to be embedded into.