4K sources testing
While the XBR-X900B did a superb with 4K, none of its picture quality advantages had to do with resolution. They stemmed primarily from its superior contrast and black levels, as well as from other picture quality advantages that apply regardless of whether the source is 4K or otherwise.
Methodology: Compared to last year there are a few more 4K sources publicly available, namely-- "House of Cards" Season 2 and now "Breaking Bad" -- and additional content for Sony owners who pony up for one of the aforementioned video players. I also had access to more test content for this test than last year, including a 4K signal generator (the DVDO AVLab TPG), two hard drives full of 4K demo content supplied by Sony and another TV maker who asked to remain nameless (these drives are used by dealers and unavailable for sale to the public), and a wide variety of individual files from various sources that I played from a PC. To send the 4K signal to each TV, I used an Atlona AT-HDDA-4 distribution amplifier.
In case you're wondering, I did test the XBR-X900B with one of Sony's 4K video players, the FMP-X1 (below). Unfortunately I couldn't get the player to work consistently enough to get any serious testing done -- too often the TV wouldn't "see" the player. Hopefully I'll get the chance to revisit Sony's 4K catalog when the second-generation FMP-X10 arrives later this summer.
In the interest of time I did not perform a full 4K calibration on any of the TVs I compared for the 4K portion of this test. Instead, I simply chose the most accurate picture setting (determined during each TV's full 1080p calibration; see below) and adjusted basic settings, namely light output, contrast, and black level. For comparison purposes, I arranged all four 4K sets so they were as close to the same seating distance as I could make them: between 77 and 87 inches (roughly 6.5 to 7.5 feet, which is quite close for 65-inch TVs). To make this arrangement work, one TV (the LG in this case) had to be the odd man out, with a distance of about 115 inches.
4K TVs with 4K content: My first goal was to see if any of the TVs' rendition of the best 4K material I had on-hand looked substantially different. The answer is yes, but only because of differences that had nothing to do with 4K resolution. Instead, the major differences were in contrast (black level), color, uniformity, and picture quality areas detailed in the 1080p tests below.
Here's where I reassure you the 4K demo content looked exceedingly good. The demo boxes from both TV makers contained spectacular footage, some of it at 60fps, of breathtaking scenery, cityscapes, natural and man-made wonders, baroque interiors, a concert, golf and soccer, and of course tech demos extolling the virtues of said manufacturer. Details were sumptuous across the board, and the camera work and subject matter was designed to show off the extra resolution. Unfortunately, since I didn't have a 1080p version of the same material to compare, I couldn't tell how much that extra resolution made a difference.
Differences in detail from one TV to the next were nonexistent to my eye. Color was the biggest difference between the TVs, but since I didn't calibrate for 4K, I don't know which TVs were more "correct" or accurate. Differences in black level were similar to what I describe below in the 1080p section.
The same went for the demo files I tried, including a particularly nice piece of demo material: "The Ultra Definition showcase" from Florian Friedrich, available at UHDcontent.eu. Its shots of a beachside town, animals, and mountains were equal in quality to any of the manufacturer-supplied footage I saw.
4K vs. 1080p content from test files and Netflix: I didn't have access to the RED players I used last year to compare the same content at 4K and at 1080p resolution, but I did have some video files in both 4K and 1080p, as well as Netflix's feeds. For comparison purposes I swapped out the Samsung curved TV for ato represent 1080p and kept the seating distances the same as described above.
The highest-quality nonproprietary source I had on-hand in both 1080p and 4K was some demo material of a piece of duck and lobster being served to an eager female diner. Beautifully detailed, if weird, I set it up to play simultaneously via two separate PCs on both the 4K TVs and the 1080p Panasonic S64. I could not see any major differences in detail. Yes, looking very closely at the gold embroidery of the tablecloth I detected the slightest bit of extra resolution and clarity on the 4K sets, and the legs of the lobster and the flesh of the duck appeared oh-so-subtly sharper. But those differences were anything but "major."
Moving back to a seating distance of about 8.5 feet, even those vanishingly small differences vanished. I looked at a couple of other pieces of footage I have in both resolutions (some courtesy of David Mackenzie and some purchased from Stockfootage.com), and the story was the same. The only area where I consistently noticed a marked improvement in detail was in small graphical elements, for example onscreen text that described the virtues of the cameras used to capture the footage.
I tried Netflix 4K as well, streaming a couple of episodes of "Breaking Bad" and "House of Cards." The latter looked most impressive, but both appeared slightly better-looking than the Super HD (1080p) stream. I compared the Sony at 4K to the other sets (including the S64) at Super HD from my PS3's Netflix app, and the slight increase in sharpness and detail was apparent in some areas, like faces and fine fabric textures. It was quite subtle, however, and I bet most viewers would have a tough time telling the difference, even at the theatrically close distance from which I was sitting.
1080p sources testing
The majority of our testing was done with mere high-definition resolution sources, and the Sony was outstanding in just about every regard. Deep black levels; accurate, well-saturated color; and plenty of video processing options made it the best of the 4K sets I had on hand. Its 3D picture was its main weakness, but that's inconsequential in the scheme of things. And as you might expect from a TV with speakers like this, it sounded great.
Click the image at the right to see the picture settings used in the review and to read more about how this TV's picture controls worked during calibration.
Black level: The X900B was one of the best performers in the stellar lineup I assembled for this review, beating every TV in this area aside from the HX950 and the ZT60 -- both of which are no longer available for sale. In some cases it actually surpassed the ZT60, in particular during extremely dark scenes.
In my favorite dark torture test from "Harry Potter and the Deathly Hallows, Part 2" the X900B's letterbox bars remained darker than any of the others aside from the HX950's during the initial hilltop shot (46:00), providing inky contrast to the rest of the darkness.
The LG and Panasonic AX800U were worse (lighter), the Vizio M was in the middle, and the Samsung came closest, delivering a very respectably deep black itself -- albeit not nearly as deep as the X900B. This ranking order remained intact in other very dark scenes from the film, for example the bombardment at the start of Chapter 14.
In mixed albeit still mostly dark scenes, like the Room of Requirement (57:40 and 59:36), the ZT60 took its place as the best in the room, with superior pop and contrast, not only in the letterbox bars but also the depths of the shadows. Even the HX950 couldn't match it, although it and the X900B again came closest. The X900B also did a very good job maintaining the brightness of brighter objects against dark backgrounds, for example the face of Voldemort overlooking Hogwarts (46:20). That said, a few of the others, most notably the Samsung, did an even better job -- although the X900B's superior black levels still allowed it to outdo the Samsung in overall contrast.
Shadow detail on the X900B was also superb, outdoing the HX950 and especially the Samsung, which suffered from some crushing. Only the ZT60 maintained superior, more natural-looking shadows than the X900B. The M, meanwhile, held its own well but showed somewhat too-light details compared to the reference ZT.
Like all of the local dimming sets the Sony showed some blooming -- where the backlight illuminates dark areas adjacent to light ones -- but it was very well controlled. My Oppo Blu-ray player's pause icon in the upper right, for example, caused blooming in the nearby letterbox bar, but on the X900B it was relatively dim. The effect was brighter and more noticeable on the HX950 and the Samsung in particular, as well as the LG (especially when its backlight flashed).
That said, this test did reveal a limitation of the X900B's edge-lit dimming compared to the full-array HX950 and Vizio -- in the latter two cases, the lower letterbox bar remained dark and blooming-free, while on the X900B and the other edge-lit sets, it also showed stray illumination.
Color accuracy: As you can probably guess from the charts and Geek Box below, the Sony had no problems whatsoever in this area. Colors showed all of the accuracy in program material I expected from its superb measurements, from the depths of its shadows to brighter, more colorful scenes like young Snape's reverie (Chapter 19). Skin tones, the green grass, and the blue sky all appeared accurate. I turned to "Samsara" for even better colors and wasn't disappointed. From the brilliant golden temples and lush jungles of Chapter 3, to the resplendent primaries of the Buddha in Chapter 4, colors on the X900B were superb.
Compared to the other sets, it showed a slight advantage in some areas, particularly red, which on the sunlit monks' robes for example (8:39) appeared a bit more full and lush than on the Samsung and Vizio, albeit about the same as the ZT60 and HX950. In other words while colors looked accurate and almost perfectly saturated, they didn't seem head-and-shoulders above the best TVs I've tested.
Video processing: Sony has increased the number of smoothing/Soap Opera Effect modes this year, but unlike Samsung and some other TV makers, it still has yet to offer one that combines full motion resolution with proper 1080p/24 cadence. For viewers who demand the proper look of film with 1080p/24 sources, the MotionFlow settings of True Cinema and Off (between which I couldn't tell any difference) are the only ones that qualify on the X900B. They also the ones with the lowest motion resolution; just 300 lines.
The Impulse setting, said to incorporate both black frame insertion and blinking, looked the best in our motion resolution test, delivering the full 1,200 lines extremely cleanly. It truncated light output by a whopping 75 percent, but unlike Impulse on some previous Sony sets, the flicker it introduced was subtle to nonexistent to my eye.
Two other modes, Clear and Clear Plus, also hit the 1,200-line threshold but weren't quite as clean as Impulse on the test pattern. Neither dims the image as much as Impulse -- although neither gets as bright as the others. All three introduced a minor but still easily discernible amount of smoothing. Finally there's Standard and Smooth, which introduce significant and massive smoothing respectively, preserve maximum light output, and registered 600 lines of motion resolution.
Like the W850B, the X900B failed our 1080i de-interlacing test, despite setting CineMotion to Auto. Previous sets, like Sony's own W900A and R520A from last year, pass. Failure isn't a huge deal, but attentive watchers may notice some artifacts, for example moving lines or moire patterns, at times with film-based material delivered to the TV in 1080i, for example via a cable TV connection.
Using the highly-detailed mandala scenes from "Samsara" Chapter 4, I played around with the Reality creation settings. I ended up turning it off completely. The Resolution slider seemed to simply exaggerate details, hardening edges and bringing out the tufts in the robes and the strands in the monks' hair to a larger degree than the film captured, and adding weird dots to the sand of the mandala at extreme levels. Some viewers might like the effect, but I didn't. Noise filtering had no effect I could discern, which is a good thing because this Blu-ray is very low noise in the first place.
I didn't bother with Sony's "Mastered in 4K" setting, mainly because it applies only to a handful of Blu-ray titles.
4K upconversion: To test how well the Sony converted 1080p material to its 4K pixel array I wanted a lineup of same-size TVs viewed from the same theatrically close seating distance I used for 4K testing. I removed the smaller HX950 and ZT60 from the lineup and swapped in the 65-inchto use as my 1080p reference. I also arranged them in the same theatrically close seating distance I used for 4K testing, described above.
I chose the best-looking Blu-ray I know, "Samsara," and stared hard at highly-detailed scenes to try to discern any difference in resolution, sharpness, perceived depth of field, and so on between the 4K Sony and the 1080p S64. If there was any, I didn't see it. The material from the monks' mandala to the destroyed buildings (Chapter 6) to the Baroque interiors (Chapter 7) looked equally detailed on all of the sets.
Switching to a lower-quality source, my Verizon Fios TV connection at 1080i, was the same story. I checked out an episode of "Mad Men" and a baseball game, and none of the 4K sets provided any increase in detail or resolution. As with the Blu-ray, none of them looked any less-detailed either.
I also asked a couple of CNET staffers to check out the same scenes from "Samsara," and while one agreed with me he couldn't see any difference, the other was able to pick out the 1080p S64 relatively quickly. When I asked him why, he said he noticed the very subtle presence of pixel structure in some flat fields, for example blue skies, and that tipped him off. He didn't pick out actual pixels from that distance, but did describe a vague sense of the picture elements nonetheless.
When we moved the couch back to about an 8.5 distance he said he didn't notice the difference as much, but at that point we agreed confirmation bias was too much of an issue to fully trust what he saw. In any case, he stressed that the difference he saw was minuscule, likely impossible to pick out when not looking at a side-by-side comparison using select parts of extremely high-quality material, and much less important than some other aspects of picture-quality differences he noticed, like black level and color.
Uniformity: While not quite as uniform in flat field test patterns across its screen as the Vizio or HX950, let alone the plasma, the Sony was still very good in this area. In flat fields there were very slight variations, typically slightly brighter along the lower and upper edges with a slightly darker middle (the opposite of the Samsung). None of these variations were visible in program material however, and my review sample was devoid of large uniformity issues. To be fair, so were the other sets with the exception of the LG.
The X900B maintained black-level fidelity from off-angle very well, beating every set aside from the X900B and the plasma in this area. The Samsung and Vizio in particular washed out more quickly. With bright material it showed no particular advantage.
Bright lighting: The glossy screen of the X900B showed brighter reflections than the the ZT60 or the HX960, but outdid the Samsung, LG and Vizio in this area. Yes, the matte Vizio scattered bright reflections better, but for most reflections, such as the white shirt of a viewer or a light wood coffee table, the X900B dimmed them better. The Sony also did a very good job maintaining black under the lights, matching the Samsung and HX950, and outdoing the others in this area, including the ZT60 by a nose.
Sound quality: [Evaluation by Ty Pendlebury] Unless you're looking at a television from Bang and Olufsen, most built-in speakers are woefully underpowered and good only for watching the news. The Sony XBR-65X900B is a different audio animal entirely, with better sound than any TV we've ever tested, and better even than one of Sony's own sound bar systems.
I tried the X900B with and without the company's optional wireless subwoofer, and found that for an $299 the extra money is definitely worth it, especially since you've already spent so much on the TV itself.
I started my tests with music and was very pleasantly surprised by the performance, even without the sub connected. For the first time since we started testing TV audio in 2013, Nick Cave was actually enjoyable to listen to on a television. With its dedicated tweeters, mdrange woofer, and transmission-line-style bass, the Sony is able to capture music in a way I haven't heard from a flat-screen before. The result was very even with no particular part of the spectrum poking out: relatively supple, deep bass, excellently nuanced vocals and plenty of sparkle where it's required.
Despite sounding a little thinner without the sub, the TV's sound blew the $499 Sony HT-CT770 soundbar away for music. The bass of that system's subwoofer sounded bloated and boomy, and the vocals sounded very constrained, lacking the wide stereo imaging of the TV.
Adding the Sony TV's optional subwoofer delivered a lot of low-end punch, and while it was initially too loud, I found that setting the crossover to 70Hz and limiting volume to -3 gave the most natural results for music (Settings>Sound>Wireless Subwoofer).
Unsurprisingly, Cinema mode in its default settings was more convincing with "Mission Impossible 3" resulting in clear dialogue, wide stereo effects, and booming explosions. Adding the subwoofer again provided extra oomph, and if you watch a lot of movies, this combination is well recommended.
If you buy this display, you can reasonably get away with not having to get an additional a sound system. Only if you're looking to spend well over $600 on an external system would you get better results than what the X900B/sub combination is capable of.
3D: The X900B's 3D performance wasn't as bad as I saw on the W850B, but it was still the worst in my lineup, and simply unacceptable for a TV of this level. Crosstalk was rampant. Watching "Hugo," the ghostly double image of crosstalk was more evident than any of the other sets, including the ZT60, in difficult areas like the statue looming over Hugo in the graveyard (14:56), the pillars below the clock (1:52) and its numbers as the ghostly outlines converge, and the sleeve of Méliès right before Hugo approaches to swipe the mouse. And yes, Hugo's hand in that scene also showed worse crosstalk than the others.
As expected, the LG was the best 3D performer in the group, with virtually no crosstalk and the added brightness of passive 3D. Among the active sets the Samsung was second-best, followed closely by the Panasonic AX800U -- all of which far outclasses the Sony X900B. Yes, the Sony's deep blacks and solid color were assets in 3D, and the glasses comfortable enough, but its crosstalk was a deal-breaker.
It's also worth mentioning again that theuses a passive 3D system, so I expect it to offer superb 3D picture quality.
|Black luminance (0%)||0||Good|
|Avg. gamma (10-100%)||2.35||Good|
|Avg. grayscale error (10-100%)||1.148||Good|
|Dark gray error (20%)||0.27||Good|
|Bright gray error (70%)||0.626||Good|
|Avg. color error||1.908||Good|
|1080p/24 Cadence (IAL)||Pass||Good|
|1080i De-interlacing (film)||Fail||Poor|
|Motion resolution (max)||1200||Good|
|Motion resolution (dejudder off)||300||Poor|
|Input lag (Game mode)||40.9||Average|