Results: Video Processing And 3D Crosstalk
This is a new area in our benchmark suite, so I’ll explain the results as I go. We use a series of pass/fail tests to determine the ability of an HDTV to process different kinds of video signals. Most of the time, you want your source components handling this because they're more capable. If you own an Oppo Blu-ray player, for example, it will exceed the capabilities of pretty much any display. Set your player to output 1080p video, and the display does no video processing whatsoever. An example of the reverse would be a cable or satellite receiver, which is usually poor for scaling and deinterlacing.
The 65L9300U is a little different because of its higher resolution. There are some Blu-ray players and A/V receivers that can upconvert 1080p signals to 4K. Or, you can have the TV tackle it. We don’t have a 4K source available, so our testing uses 1080p or lower.
The first tests consist of a group of video clips from the Spears & Munsil HD Benchmark Blu-ray Edition, which is available to anyone online for about thirty bucks. Here’s a quick rundown of what's covered:
2:2 pulldown: This is the cadence most commonly found in content shot on video cameras (at concerts and sporting events, for example). The original image is interlaced, two fields per frame, and the display must integrate them into a single progressive frame.
3:2 pulldown: The cadence most often used to convert 24p film to 60i video, its order is two fields of the first frame and then three fields of the next, in alternating sequence. If the display doesn’t integrate the extra field properly, there is a very obvious artifact that shows in our test clip and results in a failure.
Accepts 24p: Film content on Blu-ray is encoded at 24 frames per second, and all current players can output the signal at that rate. Most displays can accept this signal and process it to a refresh rate that’s a multiple of 24 using repeated frames.
- 2:2 pulldown: Fail
- 3:2 pulldown: Fail
- Accepts 24p: Pass
The failures on the 2:2 and 3:2 tests show that Toshiba’s CEVO 4K video processing engine is not as good at de-interlacing as it is with scaling. Where would you find interlaced content? It’s most common in high-def broadcasts, which are 1080i. A notable exception is Fox, which sends its signal out at 720p. What you see in practice is extra-jagged steps on diagonal lines as the 65L9300U fails to lock on to the proper cadence converting interlaced video to progressive. For film-based content on Blu-ray, the set processes 24p correctly only if you set ClearScan to Standard. Then it simply repeats each frame 10 times to match the native 240 Hz refresh rate.
The second group of tests covers an HDTV’s ability to show signals below black and above white. Unlike PC signals, which range from 0 to 255, a video signal truncates that to 16-235. The areas above and below those values are considered head and toe room, and are not used in correctly-encoded content. It is desirable, however, for a display to at least be able to show the levels between 0-34 and 236-255. It makes calibration easier, and occasionally content does stray outside the limits.
The Chroma Burst pattern shows a series of single-pixel lines, in color, to determine if a display actually achieves its maximum native resolution. Most HDTVs return different results for RGB signals than for component (YPbPr) video. 4:2:2 is the minimum bit depth output from a source; 4:4:4 is more common. Some players can output RGB, which usually eliminates a conversion step in the display. Our test shows which signal mode provides the best resolution performance.
|Above White||Below Black||Chroma Burst|
Many HDTVs only show their full resolution when fed an RGB signal. Toshiba's 65L9300U, however, performs equally well with both RGB and YPbPr signals. Since most Blu-ray players only output YPbPr, this display has a distinct advantage. The passing results on the above-white and below-black patterns mean that the 65L9300U maintains the proper black level threshold for both types of signals.
Since our Accupel signal generator supports 3D and contains a pattern for measuring crosstalk, we tested that as well. The pattern consists of nested squares; one for the left eye and one for the right. To measure crosstalk, the left-to-right timing is adjusted until the squares blend together. In our test, we checked each primary color plus gray and observed less than one-percent crosstalk for all patterns. This is well beneath the threshold of visibility. Unless there are errors in specific content, you won’t see any crosstalk or ghosting in 3D presentations on the 65L9300U.