Image-Quality: HQV’s High-Definition Video Benchmark
Remember the drill before watching HD video: ensure that “pull-down detection” is enabled in the Radeon’s Catalyst drivers and that “inverse telecine” is activated in the GeForce drivers. Even though these options tend to be enabled by default, it's good practice.
HD Noise-Reduction Test: Out of 25 Points
This test shows scenes that are plagued with noise artifacts. Good noise reduction mitigates most of these artifacts and makes the image appear more natural and much less grainy. The trick is to make the noise reduction work without losing detail.
| Graphics Processor | Score |
|---|---|
| Integrated Radeon HD 3200 | 25 |
| Integrated Radeon HD 4200 | 25 |
| Integrated GeForce 8200 | 25 |
| Integrated GeForce 9400 | 25 |
| Integrated Intel G45 | 25 |
Things are off to a very good start, with the Radeon and GeForce integrated chipsets achieving a full score. While Intel's G45 seems to be working, the quality doesn't seem to be in the same league as the Radeon and Nvidia solutions, but we still gave it a full score. Out of all three, the Radeons appear to have the highest-quality noise reduction, followed by Nvidia and then Intel. This is, of course, a somewhat subjective test, and since the Radeon and Nvidia chipsets met the criteria, they both get full points.
It is interesting to note that noise reduction was not enabled by default on the GeForce boards. We found that a setting of 70% would give the best results, but personal taste will, of course, be a factor.
Most disturbing, however, was how the GeForce 8200 board tended to hiccup when noise reduction was used. So, while it achieved an ideal score in this synthetic test, it's not really viable in a real-world situation.
Video-Resolution Loss Test: Out of 20 Points
This test simply shows a pattern of lines and color bars from an interlaced source. If the DVD player can show the smallest lines without flickering, it is successfully de-interlacing the image.
| Graphics Processor | Score |
|---|---|
| Integrated Radeon HD 3200 | 20 |
| Integrated Radeon HD 4200 | 20 |
| Integrated GeForce 8200 | 20 |
| Integrated GeForce 9400 | 20 |
| Integrated Intel G45 | 20 |
All of the integrated chipsets handle video resolution loss with no flickering and are performing well in this test. While some of the chipsets would take a half a second or so to compensate for the video resolution loss, they all did the job adequately and scored the full 20 points. Note that the GeForce 8200 failed this test when using drivers newer than 182.5, such as the current 190.38 driver, so the points are awarded for the older driver.
HD Video-Reconstruction Test: Out of 20 Points
This test simply shows a few rotating lines in a circle and is also from an interlaced video source. As the angle changes, interlacing makes the lines appear to become stepped. Successful de-interlacing will remove this stepping.
| Graphics Processor | Score |
|---|---|
| Integrated Radeon HD 3200 | 0 |
| Integrated Radeon HD 4200 | 0 |
| Integrated GeForce 8200 | 0 |
| Integrated GeForce 9400 | 20 |
| Integrated Intel G45 | 10 |
There are some very interesting developments here, indeed.
First, the GeForce 9300/9400 was the only IGP able to achieve a full score in this test. However, it failed the test with the 190.38 driver and would only pass once we reverted to the 182.5 driver.
The Intel G45 managed to perform de-interlacing, but it didn't quite receive the full 20 points for this test because the smoothed edges would display some artifacts. Since it was clear that the G45 was performing some enhancements here, we gave it half points.
As far as the GeForce 8200 and Radeon solutions go, they performed no diagonal filtering whatsoever. This appears to be a feature they leave to their discrete card lineups, but since the Intel G45 can handle jaggy reconstruction, it's hard to believe the GeForce 8200 and Radeon 4200 don't have the power to handle it. Perhaps the feature was omitted to drive users toward add-in boards.