How about this?
Disucssion:
Let me ask you a question. You know what “fully” and “every” mean. But a software tool can not see what “fully” and “every” mean unless you tell the software how to detect the difference between what is acceptable for you and what is not. How can you tell the tool? And you would surely have an important requirement: you do not want to write hundreds lines of test script that check all components in your web page.
“Visual Testing” approach proposes you a solution to this question.
- take screenshot of a web page TWICE. You create a pair of 2 images. “Exepected” and “Actual”.
- compare the Expected image and the Actual image to find out how much they are different visually each other.
- If you find 0% of difference, then the Actual web page is “fully” acceptable.
- If you find certain difference greater than 0.0, then the Actual page is possibly unacceptable.
- The test does not tell much detail about the cause of image differences, but the test notifies you of which URL is questionable and where in the page you should investigate.
- If you have many URLs to take care (say 1000 pages), then it is nice to know that just a few URLs among them are marked questionable and others are OK. The tool improves the productivity of testers.