One major thing separates the HC7900 from the less expensive BenQ W1070 and Epson 3020: light output. The HC7900 is quite a bit dimmer than either one, making it less versatile in terms of screen material and size as well as tolerance for ambient light. In a completely dark room the HC7900 is a solid performer, with standout video processing, deep-enough black levels for the price, and accurate color in bright scenes.
|Comparison models (details)|
|Epson PowerLite Home Cinema 3020||3LCD projector|
|JVC DLA-X35||D-ILA projector|
Black and white level: If you use price as a proxy for black level, according to our measurements the Mitsubishi is slightly worse than you'd expect: about even with the Epson 3020 and a bit better than the BenQ. Watching the darkest parts (like the beginning of chapter 12) of "Harry Potter and the Deathly Hallows: Part 2," the three appeared very close to one another, and quite a bit worse than either the 5020UB or the JVC. The image of the army massing outside Hogwart's had significantly more pop and depth on the more expensive models, while the Mitsubishi, BenQ, and Epson 3020 were roughly equally washed-out in comparison.
The HC7900's shadow detail was good but in the very darkest scenes, like the shot of Hogwarts before the assault (46:23) and the Room of Requirement (57:40), it crushed detail a bit compared with the BenQ, although not quite as much as the Epson 3030.
As usual the iris mode did improve black levels visibly and measurably, so I ended up leaving it on. I chose the Auto 1 setting because it seemed to offer the best combination of black-level performance and gamma. Unlike the iris on the 3020 it was completely silent, and indeed the HC7900 was the quietest among the three lower-end projectors overall. I looked for overt iris effects -- relatively abrupt and unnatural changes in luminance -- but didn't notice any.
The Mitsubishi is the dimmest projector we tested, and that's its main weakness. After calibration it failed to achieve our target 16 fL on a 120-inch diagonal high-gain screen, topping out at just 13.4 fL. There isn't a huge visible difference from 16 fL, but it did contribute to there being a bit less pop and impact in mixed bright and dark scenes.
More issues arise if you want a larger screen, one with a more neutral gain, or if you want to watch in a brighter environment than pitch darkness. In its brightest default picture mode the HC7900 measured just 17.71 fL, which works out to just 582 lumens. That's a bit less than half as bright as the next-dimmest projector we tested, the JVC, and about a third as bright as the BenQ and Epson 3020.
Color accuracy: Despite some less-than-perfect results (listed in the Geek Box below), the overall color of the Mitsubishi was very good in bright areas. The face of the mourning Mrs. O'Brien in the opening of "Tree of Life" (7:01) looked quite close to how it looked on our reference JVC, although I did notice a slightly paler, greenish cast to some areas, and the colors weren't quite as punchy -- a difference that has more to do with the differences in black level than in actual accuracy.
The CMS did fall short of perfecting magenta and especially blue, but it was otherwise very good. The large "Blue error" listed in the Geek Box is less of a big deal than the number might indicate, mostly because such errors are less visible than in other colors. Yes, I noticed a bit less saturation in blue areas like the walls of the bedroom (7:10) and the jellyfish in the ocean (29:35), but it wasn't drastic.
The Mitsubishi's biggest color problem is one it shares with the other DLP projector, the BenQ: greenish blacks and near-blacks. Between the two the Mitsubishi seemed somewhat worse in this regard than the BenQ, because its green tint seemed to creep more into bright areas, whereas the BenQ's seemed to remain less visible. If I had to choose I'd pick the bluish near-black tint of the Epson 3020 over either one.
Video processing: Compared with the other like-priced projectors in our test, this category was a strength for the HC7900. Mitsubishi's adjustable Frame Rate Conversion system is one of the most versatile I've seen. The company mentions a 96Hz driver in its manual, and the setting consists of multiple options that offer improved motion resolution, smoothing, or both.
In True Film mode, designed for 24-frame sources like most scripted TV shows and 1080p/24 Blu-rays, the five different settings had progressively smoother effects, and I really appreciated the fineness of the gradation from one to another. The first three modes produced visible judder that I preferred to the more buttery look of modes 4 and 5, and modes 1 and 2 in particular looked very close to Off. I'd probably still choose Off for myself for films, but among all the smoothing modes I've seen, Mitsubishi's True Film is among the most palatable at its lowest settings.
There's also a True Video mode for higher frame rates, with its own five fine levels of gradation. These combinations delivered varying degrees of motion resolution (blur reduction) although, unsurprisingly, the less smooth options measured the worst in our. True Film settings 1 and 2 (my favorites) were barely better than Off, and not until 4 and 5 (my least favorites) did the projector reach its maximum motion resolution of just above 600 lines. As usual I'll trade a bit of blurring (which I always have a tough time discerning) for true film cadence and minimal smoothing, but it's great that Mitusubishi offers so many options.
As for non-24-frame sources like sports, I'd probably choose True Video level 1, which achieved nearly full motion resolution and minimal artifacts. As I stepped up toward level 5, incidences of artifacts, for example breakup in moving objects on my test pattern, became more common.
I also noticed about the same amount of "rainbow effect" on the Mitsubishi and the BenQ. It's an artifact unique to single-chip DLP projectors, and appears as occasional, brief flashes of color as the eye perceives the color wheel's imperfect construction of light. It manifested primarily when I looked away from the screen and then back, and then mainly in highlights like the shine of the urn from "Hallows" at 59:28, or the white Pause symbol of my PS3's display in the letterbox bar. All in all it was mild, but if it bothers you I recommend getting an LCD or LCoS projector instead.
Bright lighting: As you can imagine with such a low level of light output, the Mitsubishi didn't hold up well against the competition -- particularly the two Epsons and the BenQ -- with the lights on. Its image looked significantly more washed out than any of the others on its brightest default setting. In the dark is always the best way to watch a projector, but the Mitsubishi is even less resistant than most to ambient lighting.
3D: The Mitsubishi's 3D performance was pretty good. The highlight was basically the complete lack of crosstalk during "Hugo," my favorite 3D torture test. The GK Films logo at the beginning, the floating hand as Hugo reaches for the mouse (5:01), the tuning pegs on the guitar (7:49), and the face of the dog (9:25) were equally clean on both units, without even the merest hint of the double image. The downside, again, was its relatively dim light output. Comparing their default 3D picture modes the Epson 3020 and JVC were more than twice as bright as the Mitsubishi in 3D.
The BenQ and Mitsubishi were equally dim, on the other hand, which is a surprise given the BenQ's vast light output advantage in 2D. I'm not sure why, but I'm guessing it's an issue with the BenQ's default 3D mode, and no easy fix was obvious.
For what it's worth I also found the XpanD "Made for Mitsubishi"/X103 glasses relatively uncomfortable, with too tight a grip on the back of my head. The also-compatible X104s are a much better fit for me.
|Geek box: Test||Result||Score|
|Black luminance (0%)||0.0069||Good|
|Avg. gamma (10-100%)||2.14||Good|
|Avg. grayscale error (10-100%)||1.0032||Good|
|Near-black error (5%)||0.606||Good|
|Dark gray error (20%)||0.995||Good|
|Bright gray error (70%)||0.863||Good|
|Avg. color error||4.213||Average|
|1080p/24 Cadence (IAL)||Pass||Good|
|1080i Deinterlacing (film)||Pass||Good|
|Motion resolution (max)||600||Average|
|Motion resolution (dejudder off)||300||Poor|
|Input lag (Game mode)||48.1||Average|
Read more about how we test TVs.