While Digital Watermarking has received much attention in recent years, it is still a relatively young technology. There are few accepted tools/metrics that can be used to evaluate the suitability of a watermarking technique for a specific application. This lack of a universally adopted set of metrics/methods has motivated us to develop a web-based digital watermark evaluation system called the Watermark Evaluation Testbed or WET. There have been more improvements over the first version of WET. We implemented batch mode with a queue that allows for user submitted jobs. In addition to StirMark 3.1 as an attack module, we added attack modules based on StirMark 4.0. For a new image fidelity measure, we evaluate conditional entropy as an image fidelity measure for different watermarking algorithms and different attacks. Also, we show the results of curve fitting the Receiver Operating Characteristic (ROC) analysis data using the Parzen window density estimation. The curve fits the data cl...