Performance
While we appreciated the Mitsubishi's deep black-levels and the ability to tweak some aspects of its color, we really missed full color temperature controls and found some strange artifacts caused by its geometry correction system.
Our first step, as always, was to adjust the set for optimal performance in our completely dark home theater. We turned the WD-65734's light output way down (this set can get blindingly bright) to a comfortable 40 ftl, which also improved black-level performance. Mitsubishi only provides two color-temperature presets, and the best, called Low, was still pretty inaccurate, skewing toward red and especially green. Unfortunately the company doesn't provide user-menu gain and/or cut controls for grayscale adjustment, so you can't do anything to improve color temperature outside of hiring a professional.
During setup we also used the PerfectColor and PerfectTint to improve the color fidelity of the WD-65734. The set's default color decoding was pretty inaccurate, but with the PerfectColor, it improved quite a bit, and after adjustment, decoding was very good. Primary color accuracy was likewise subpar in the default settings (see the Geek Box below), and PerfectTint again provided a significant improvement, allowing us to get both primary and secondary colors much closer to the DTV standard (although we couldn't do much with yellow). See our complete user-menu settings, or check out the Tips & Tricks section above.
After our tweaking session, we set up the Mitsubishi next to a few other like-size HDTVs we had on hand, including the 56-inch Samsung HL-T5687S LED-based DLP, the 61-inch Panasonic PT-61LCZ70, and the 58-inch Panasonic TH-58PZ700U plasma. We watched The Marine on Blu-ray at 1080p resolution, courtesy of our Samsung BD-P1200.
In dark scenes, the Mitsubishi showed off its ability to produce a realistic shade of black. When the Marine gets captured by the hicks in the swamp, the shadows in their shed and dark spots in his hair were nice and deep, although not quite as inky as on the Panasonic plasma. Shadow detail was also very good, and gradations from darkness into lighter shadows appeared natural and not too abrupt.
Colors on the Mitsubishi, even after we'd adjusted the "perfect" controls, were still a mixed bag. Primary and secondary color accuracy were quite good, with the exception of yellow. In the first scene of the Marine's homecoming, the yellow cab appeared more orange than we'd like to see, and the same can be said for the arch-villain's ubiquitous tie. When the bad guys take the Marine's wife into the woods, the green of the Mitsubishi's trees and bushes looked lifelike and lush,--much better than either of the Panasonic's, which were both too yellow in the green areas--but not quite as good as the Samsung's.
Skin tones, however, weren't as natural. During the couple's reunion scene, their skin did look a bit too ruddy, especially the red on the back of his neck. In better-lit scenes, such as when the pale-faced, brunette villainess hitches a ride from the truck driver, we noticed a slight greenish tinge to her skin when compared to the more accurate Panasonic plasma, although the effect wasn't quite as noticeable as the redness during lighter scenes. We felt the plasma had the edge in the color department overall, owing to its superior black levels (which increased apparent saturation) and color temperature.
Details in the film, from the wife's blonde hair to the texture of the fabric on the car seat to the dashboard vinyl, looked quite sharp on the big screen, but when we looked at test patterns we noticed something unusual. The 720p, 1080i and 1080p patterns from our Sencore VP403 signal generator, as well as the PC-based every-other-line full-screen patterns from DisplayMate, showed curved irregular artifacts on the right and left thirds of the screen, where there should be only vertical lines (the middle third of the screen was unaffected and showed full resolution). The same curves were visible in the horizontal lines of vertical resolution patterns. The artifacts were similar to the effect produced by engaging keystone correction on a front projector. Indeed, when we asked Mitsubishi about what we saw, the company replied that the WD-65734 utilized an electronic geometry correction system, similar to that of front projectors, to combat geometric distortions.
Frankly, we'd prefer some minor geometric distortions to the artifacts we saw, and while they weren't visible in The Marine, we did notice them elsewhere. For example, in Chapter 9 of Aeon Flux, the wires hanging from the ceiling showed a broken pattern similar to the curves on the resolution chart, a pattern that was not visible on other TVs in the room. We're sure other evidence of this issue will appear on similarly difficult material, although it certainly won't be noticeable the majority of the time. The WD-65734 passed the test for 1080i deinterlacing of video-based content, although like most HDTVs we tested, it failed the one for film-based material.
Although Mitsubishi claims its geometry correction system leads to fewer distortions than with other rear-projection designs, both the Samsung and the Panasonic RPTVs delivered straighter lines with fewer distortions. The WD-65734's horizontal lines curved upward (the smiley-face effect) toward the top of the screen especially, and its vertical lines bowed inward in the middle and outward toward the top and bottom of the screen (the pincushion effect). Like most such distortions, they were basically unnoticeable during normal program material, but onscreen graphics with straight lines, such as the menus and program guide from our DirecTV HR20 receiver or the bars to both sides of 4:3 content, appeared less than straight. As always, individual samples may have more or less distortion than the one we reviewed, and a professional calibrator should be able to adjust them using Mitsubishi's system.
Uniformity across the WD-65734's screen was solid for a rear-projection HDTV. As usual, the middle of the screen appeared brighter than the edges, but there was no discoloration visible to the sides and edges, as we saw with the Samsung and the Panasonic RPTVs, and the brightness difference wasn't obvious during normal program material. We did notice the Mitsubishi's stationary screen grain, which appeared as tiny dots in bright areas and was a bit more noticeable than either of the other RPTV sets. Speaking of uniformity, we also didn't notice any rainbows during The Marine, even in the fastest-moving, highest-contrast areas, and only saw them when viewing white-on-black text of the credits and other high-contrast material. We're quite sensitive to the rainbow effect, but on this DLP we found it much less bothersome than on many--although the LED-based Samsung was still the antirainbow champ.
With PC-based sources, the Mitsubishi again delivered every line of 1,920x1,080 resolution when connected via its HDMI input to the DVI output of a computer, although those curves were still in evidence. We didn't see evidence of the curves in text, although it did look slightly softer than the other RPTVs in the room. As with all rear-projection sets we've tested, the WD-65734 showed overscan along the edges of the screen, which in the case of a computer desktop caused elements like the taskbar along the bottom and icons on the left side to disappear. Many video cards have overscan compensation to correct this issue, although they won't deliver the full resolution after correction.
With standard-def sources, tested via 480i component-video using the HQV test disc on DVD, the Mitsubishi turned in a below-average performance. It delivered every line of horizontal and vertical resolution from the disc, and the stone bridge and grass from the detail scene looked sharp. The set didn't do a very good job of smoothing out the edges of moving diagonal lines, however, so we saw quite a few jagged edges on the stripes of the waving American flag. We were also surprised to note that the WD-65734 failed the test for 2:3 pulldown detection; when Film mode was engaged, the set eliminated moire in the grandstands briefly, then the lines came back. Noise reduction, on the other hand, was very good, and each of the three settings progressively eliminated more and more of the snowy motes of noise in sunsets and sky shots.
| TEST | RESULT | SCORE |
| Before color temp (20/80) | 5821/7193K | Average |
| After color temp | N/A | |
| Before grayscale variation | +/- 441K | Good |
| After grayscale variation | N/A | |
| Color of red (x/y) | 0.673/0.311 | Poor |
| Color of green | 0.302/0.654 | Average |
| Color of blue | 0.147/0.074 | Average |
| Overscan | 4% | Average |
| Black-level retention | All patterns stable | Good |
| Defeatable edge enhancement | Y | Good |
| 480i 2:3 pull-down, 24 fps | Y | Good |
| 1080i video resolution | Pass | Good |
| 1080i film resolution | Fail | Poor |
| Mitsubishi WD-65734 | Picture settings | ||
| Default | Calibrated | Power Save | |
| Picture on (watts) | 240.26 | 213.94 | 212.47 |
| Picture on (watts/sq. inch) | 0.13 | 0.12 | 0.12 |
| Standby (watts) | 12.23 | 2.98 | 2.98 |
| Cost per year | $80.39 | $66.78 | $66.33 |
| Score (considering size) | Good | ||
| Score (overall) | Average | ||
What You'll Pay
- Set Price Alert