Video Quality Tested: GeForce Vs. Radeon In HQV 2.0

Test Class 1: Video Conversion

This video conversion test class focuses on the video processor’s ability to take different varieties of video and play them back as smoothly as possible. Ideally, jagged edges and artifacts will be eliminated, while proper color reproduction should remain intact.

Chapter 1: Video Resolution tests

The first chapter of this class tests the hardware’s ability to display static and moving objects in the same interlaced scene with as little stepping (also referred to as “jaggies”) as possible. To achieve a good score, the video processor has to merge or interpolate between fields of video to create smooth edges where appropriate.

Test A: Dial

A straight line is rotated in the center of a circle. Interpolation becomes increasingly difficult at smaller angles, so an ideal score of five is achieved if the line appears smooth under 10 degrees from the horizon. Four points are awarded if jaggies are seen between 10 and 20 degrees, and two points if the jaggies disappear from 20 to 30 degrees. No points are given if the jaggies appear above 30 degrees.

This test can be difficult to judge, even though it should be entirely subjective. This is because the dial moves quickly and it’s hard to be 100% sure where the artifacting begins (if it starts close to 10 degrees).

It’s important to note that some image quality enhancements can increase the result of some tests while lowering others. The dial test is an example of this. When the edge-enhancement feature is enabled, the jaggies can increase a little bit. This is why the GeForce 210 has a higher score than more expensive GeForce cards that have this feature enabled.

Test B: Dial with Static Pattern

This test appears similar to the previous one, with the same line rotating within the circle. The focus here is different, however, and is on the grid. If the video processor breaks or warps the grid while the line moves through it, points will be lost. Note that a halo effect around the line is acceptable as long as the grid isn’t being warped or broken. A perfect score with no grid artifacts is five points. Some degradation of the grid or slight artifacts lowers it to two points, and a flickering or obviously degraded grid decreases the score to zero.

While this test is somewhat subjective (as far as the amount of degradation is concerned), all of the graphics cards we’re testing appear to leave the grid intact. There’s a small halo effect, but that doesn’t prevent all of the cards from getting full marks.

Test C: Gray Bars

This test determines if the video processor’s de-interlacing capability performs consistently at all brightness levels. The focus here is not to see which bars have jaggies and which do not, but to see if all four shades of bars have the same de-interlacing without artifacting. Five points are awarded if all bars appear similar, three points if the darkest bars have artifacts, two points if the second darkest bars have artifacts, one point if the second-brightest bars have artifacts, and no points if all bars have artifacts.

The scoring criteria are objective, but the test relies on subjective observation. We can’t detect any differences between the levels of brightness on any test card, so we give all cards the full five points in this test.

Test D: Violin

This test simply appears to be some footage of a violin player. The violin strings are at low angles to the horizon and move around as the instrument is played. The full five points are awarded if the moving strings show no motion artifacts. The score is lowered to three points for small artifacts during motion transitions and zero points are awarded if there are notable artifacts during most of the video sequence.

Whether or not there are any artifacts should be fairly objective, but the amount of artifacting is subjective. All of the cards we tested achieved five points, according to our observations.