Comments

The documentation to the HQV HD test is far from perfect, especially with regards to the film resolution loss test that does not tell explicitly which squares must not produce artifacts for the video-processor to score the maximum amount of points.

If deinterlacing is incorrect, we see a flickering (strobing) of horizontal lines, which means the video processor fails the test. But since we are dealing with movies, not with video as Silicon Optix suggests, motion-adaptive video processing cannot be applied and, theoretically, the highest score cannot be awarded if a flickering of vertical lines is observed. Unfortunately, the latter thing is not explicitly explained in the documentation while Silicon Optix’ representatives couldn’t give us a definite answer, either. That’s why we award 25 points for the film resolution loss test, but do not award 10 points for the film resolution loss test “Stadium” due to the subjectively poor image quality.

However, the problem of a vague description of the test procedure refers not only to the film resolution loss test. Silicon Optix allows manual regulation of noise reduction for the HD noise reduction test. Nvidia caught at the opportunity and implemented very aggressive noise reduction methods into one of its driver versions, declaring a maximum possible amount of points in this test. But these aggressive methods provoke ghosting artifacts in high-contrast dynamic scenes, which makes rather questionable. Being indifferent to such optimizations and to the resulting artifacts, Silicon Optix lowers the value of its test for PC video enthusiasts.

As you see, the graphics cards vary greatly in their results, but considering the vague documentation to the HQV HD test and possible optimizations in the drivers, we can’t claim the numbers will be 100% true to life if AMD and Nvidia program their Avivo HD and PureVideo HD processors more efficiently.

In fact after we tried to use ForceWare 163.69 drivers from Nvidia, we noticed considerable improvements in “HD noised reduction” and “Jaggies” tests on the GeForce 8600 GTS, however, the GeForce 8500 GT demonstrated worse “Jaggies” quality (0 points) compared to results obtained on ForceWare 163.11. As a result, we publish the best possible scores for graphics cards we observed to date.

The GeForce 8600 and Radeon HD 2600 quite naturally boast a higher image quality than the other graphics cards when playing HD DVD, but it is impossible to choose the best out of the two basing on their scores in the HQV HD test.