Initial testing was done with a Sencore VP-403 high-definition video generator. My first investigation was into the slightly soft picture. Bandwidth on this TV is good at 720p with the one-pixel-on / one-pixel-off section of the Multiburst pattern (37.5 Mhz) reproduced distinctly with only mild banding. Surprisingly, the HDMI inputs had very little advantage over the component inputs. The demanding Focus test pattern also revealed little or no difference in fine detail between the two input types or between 720p and 1080i scan rates, though the pattern always looked a bit soft and was never reproduced to perfection (something 1080p sets usually can do). Turning DNIe on did increase the apparent resolution of the Focus test pattern yet with minimal edge enhancement artifacts. Focus uniformity across the screen was excellent, and far, far better than any CRT based RPTV ever was. The ATSC digital tuner slightly outperformed even the HDMI input in measured resolution.

Light output was a high 107fL in Dynamic mode; slightly less (89fL) in Movie mode. Black level measured 0.045fL. Movie mode peak contrast ratio was 1978:1. ANSI contrast ratio with a PhotoResearch contact photometer was 184:1, which isn't bad for an RPTV.

Overscan was good at 2%, geometry was very good, and centering was fairly good. It was not possible to overdrive the set into white crush even at maximum settings of the contrast control in any mode (including with the ATSC tuner), nor did brightest whites discolor at high settings. The Brightness and Contrast controls did interact more than some sets. The sharpness control isn't overly aggressive and can be run as high as 40 (DNIe off), and 20 (DNIe on), without undue edge enhancement artifacts.

Color temperature as delivered measured a bluish (compared to neutral gray) 9000 Kelvins in the Warm 2 setting, and even more bluish (10,300K) in Warm 1. There are also three more settings with even more bluish excess. Excessive blue in the black and white picture does make the final color picture look brighter, but also takes a huge toll on color accuracy. Unfortunately, the human eye can become accustomed to a bluish tint, making neutral gray look terribly reddish by comparison. Avoid these cool color temperature settings from the beginning and the warmer, more accurate ones won't trick your eyes. Vivid defaults to Cool2, which is why these sets look so bright on the showroom floor.

Movie mode was factory calibrated a bit different from the others, with Warm 2 measuring about 7500K—just slightly more bluish than the D65 (6500K) industry standard. Since Movie mode was closest and can be set up separately from the others, that's the one I chose for grayscale calibration. After calibration, the HL-S missed grayscale perfection only due to a region around 40 IRE that wanted to have a slight excess of green. Preventing a dreadful green error (very visible to the eye) required leaving just a touch more blue than ideal (hard to notice). Green had to be allowed to drop a bit above 50 IRE so it wouldn't be excessive at 40 IRE. Overall, while some other DLP sets have calibrated to perfection, the HL-S still came fairly close.

Some sets, especially certain plasmas but even some projectors, measure quite differently depending upon how much of the screen is lit. This, of course, would mean that grayscale is dynamically changing with the picture and that an accurate calibration would be impossible. The Samsung was notably free of this flaw. Color temperature was constant whether viewing a small window pattern, a staircase pattern, or a full screen (raster).

With some sets, horizontal and vertical staircase patterns show abrupt narrow band aberrations in grayscale tracking (I'm referring to changes in the "color" of "gray" as bright white fades to gray then black.). In fact, earlier Samsung HLR set's I've seen have shown an unwanted reddish band just above black that couldn't be calibrated out and visibly colored the picture with a reddish tint in warmer color temperature settings. The HL-S had no such problem, and to make sure, I checked the grayscale up from black in tiny 1 IRE increments, yet still found none. This is exceptional performance that clearly helps this set achieve a good picture.

The HDMI and component inputs' grayscale (all scan rates) measured nearly identically, with the ATSC tuner slightly better than either. Grayscale consistency between various inputs and scan rates is fairly unusual and highly commendable. Even some high-end projectors can't match the Samsung's consistency in this area.

Color primaries in modes other than Movie were typical of modern sets—fairly accurate reds and blues but greatly oversaturated greens. Movie mode differed mainly in the shade and saturation of green. Considerably less green saturation (down close to HDTV standards) made for convincingly natural grass and foliage.

Scaler/deinterlacing tests using my reference Pioneer Elite DV-79AVi DVD player and the Silicon Optix HQV Benchmark test disk showed the Samsung as a real winner with 480i sources. The Jaggies tests looked as good as I've seen (much better than running the Pioneer player at 480p) and 3:2 pulldown (when turned on) was lightning-quick to lock in. The Detail test at 480i still looked a bit soft. Translated into everyday recommendations, this set should beat the pants off your cable or satellite box when it comes to upconverting 480i programming. Set your box to "Native" if possible, which would allow the set, not the box, to do the scaling, and don't hesitate to use the component inputs over the HDMI inputs for these sources (which won't accept 480i anyway). They'll often look better.