I'm hoping for a little help with an experiment.
In a nutshell, what I'm trying to do is determine whether there are any exceptions to what I've determined about the dithering Samsung TVs implement. I have identified two types, and have put together test images which make it easy to identify the presence of either without having to closely scrutinize pixels. But the truth is that I only own a single Samsung TV myself, and even though I've spotted the same dithering on Samsung TVs in brick & mortar stores, there's nothing to guarantee that every 4K Samsung TV made exhibits the same dithering patterns.
The first dithering type
seems to be permanently on at all times, non-defeatably. It is a pattern where every other horizontal line is visibly somewhat darker, giving a look not dissimilar to CRT scanlines in e.g. a retro video game. It seems to affect blues and oranges the most. This is the test image for identifying the "scanline" dithering.
The second dithering type
is a crosshatch pattern
—every other pixel, on both x and y axes, exhibits a pattern of alternating brightness and/or hue. This pattern can be disabled in theory, but the reality is there's a 99% chance you're staring at it right now. To explain: On my Samsung model (Q70R), in order to force the pattern to go away and not return, I must: 1) Label the HDMI port as "PC". 2) Use only 29, 30, 59 or 60 Hz refresh rates—no movie-compliant 24 Hz allowed. 3) When first turning the TV on, briefly swap the refresh rate to 59 Hz and back to 60 Hz. The latter step is required because my TV will "forget" to leave the crosshatch dithering off whenever it is freshly turned on. This is the test image for identifying the crosshatch dithering.
The file format of the image is .MNG, which is animated .PNG, and the only viewer I know of which will view it correctly is IrfanView
. The reason it must be animated is that the crosshatch dithering is something the Samsung TV will occasionally only turn on under certain circumstances (after which it won't turn it back off), and my test image has been developed to trigger those circumstances, thus ensuring no anomalous behavior from the TV.
As noted in the images themselves, both images must be viewed at "100% zoom", which is to say not
fullscreen but instead in a window with no resizing. IrfanView
will do this by default. It might also be helpful to make certain you're using 4:4:4 color, which can be tested with this image by JoelArt
, as found in this post
What I anticipate is that everyone will see evidence of dithering from both test images. If crosshatch dithering is not in evidence, it might be worthwhile to try turning the TV off for a minute, and then turning it back on to see if crosshatch dithering has returned, as it will have in the case of my Q70R. As previously noted, it should be possible to make the crosshatch dithering go away by swapping refresh rates (60->59->60). But I do not believe the "scanline" dithering can be defeated.
One thing I hope people will confirm for themselves is that the crosshatch dithering will always return at 24 Hz, which is the particular issue I have with my own TV—if I want to watch movies at their intended framerate, I have to stare at crosshatch dithering and lose some detail. I generally just watch movies at 60Hz because of this.
Why is dithering a problem? Aside from the perhaps obvious point that the TV is proving unable to accurately reproduce the video it is fed, resulting in somewhat inaccurate colors, the perhaps more important issue is that dithering of these types lowers the effective resolution of the TV
. Not just the color resolution—the entire screen resolution. You are not getting 4K. The actual effective resolution is not easy to quantify, and is in fact dynamic, because the dithering itself tends to depend on what colors are being displayed, but for example take a look at this image from FlatPanelsHD:
The crosshatch dithering is readily visible in the Q90R specimen. When sitting from a typical viewing distance, the crosshatch pattern likely diminishes, but what doesn't diminish is the loss of fine detail. This picture of an eye is meant to have a certain amount of detail, which the Sony AF9 reproduces faithfully, but which the Q90R effectively crushes due to its dithering. It is no longer 4K.
Rtings discusses dithering briefly in their What Is The Resolution
article. They accurately note that crosshatch dithering on the Q90R can be disabled in "PC" mode. What they fail to note, however, is that one still needs to jump through hoops to ensure it's gone, and
the user still
cannot watch movies at their native 24 Hz because the crosshatch dithering will simply come right back. They also fail to account for the "scanline" dithering which is ever-present.