Regardless of the method used,
once a test is performed,
the result needs to be evaluated so that a decision can be made if the test failed or not.
The evaluation is basically always based on comparison—by comparing received results with a stencil (expected result) one can decide whether the output is correct which in most cases means identical to what is expected.
Sometimes when the test conditions influence the result, only some characteristics of the original and the newly obtained test result have to match.
With GUI enabled applications the test can use regular methods for comparing results but it can also use ways that are specific to applications that produce some kind of visual output.