Get insider access
Preferred store
Your browser is not supported or outdated so some features of the site might not be available.
To try to better understand how long a TV should last, we're running 100 TVs through an accelerated longevity test for the next two years. We've just posted our 1-year video update with our latest findings on temporary image retention, burn-in, and more!

Test Bench 1.3


See previous 1.2 changelog.

See next 1.4 changelog.


Picture Quality

  • Overhauled the Viewing Angle test:
    • Data is now converted to the CIE LCh (ab) color space, which is more perceptually accurate
    • Divided color scoring into Color Washout (chroma loss) and Color Shift (hue shift)
    • Added a Gamma Shift test by measuring how gray shades shift with viewing angle
  • Improved the Local Dimming video by filming it at an angle, so the blooming is more visible
  • Pre-Calibration test changes:
    • Changed the scope of what settings can be changed during the Pre-Calibration test; now all settings that don't vary from unit to unit (such as picture mode, color temperature, gamma) can be changed
    • Added information fields for the color temperature and gamma setting we tested at
    • Remeasured several TVs that were affected by this change, such as TVs with adjustable gamma settings
  • Post Calibration, added some information fields:
    • How fine the white balance adjustment is (20 point, 2 point, etc)
    • Whether color calibration is present
    • Whether the TV has an auto-calibration function that can be used with an external measurement device
  • Remeasured SDR brightness with a PC as the source, rather than our Blu-ray player
    • We want to use a PC as the source from now on, so we can automate testing
    • We found that our Blu-ray player was sending a brighter-than-white signal all this time, causing TVs to measure brighter from the Blu-ray player than from the PC, so we had to remeasure all 2018 TVs using a PC
  • Remeasured HDR brightness for a few TVs with a PC as the source (LG B8, LG C8, and LG E8, Samsung The Frame 2018, TCL S405, Sony X750F, Toshiba Fire TV 2018)
  • Removed the Picture Mode field from the Post Calibration box, as it's always the same as the Pre-Calibration picture mode
  • Update 04/19/2019: With test bench 1.3, the peak brightness for both HDR and SDR is derived from two minutes of measurement data rather than five. This change was mistakenly omitted from the changelog.


  • Fixed a bug with the BFI scoring, so 60 Hz now scores a perfect 10


  • Added some Input Lag tests:
    • 1440p @ 60Hz
    • 1440p @ 120Hz
    • 1440p with Variable Refresh Rate
    • 8k @ 60Hz
    • 8k with Variable Refresh Rate
    • Auto Low Latency Mode (ALLM)
  • Added some Supported Resolutions tests:
    • 1440p @ 120Hz
    • 8k @ 60Hz
  • Changed the Supported Resolutions scoring, from equal weighting to weighting based on how commonly used that resolution is
  • Added an "Audio Passthrough" box, and migrated five tests there
  • Added some new tests to Audio Passthrough:
    • eARC support
    • Dolby Atmos via TrueHD via eARC
    • DTS:X via DTS-HD MA via eARC
  • Added "HDR10+" to Inputs Specifications
  • Merged the Rear Inputs and Side Inputs photo boxes into one Input Photos box
  • HDMI 2.1:
    • Added an "HDMI 2.1" field that is based on whether the manufacturer advertises the TV supports it, as we currently have no way to test it
    • Removed "HDMI 2.1 Full Bandwidth"; TVs this year will start to support higher bandwidth but we have no way to test it (no sources support it yet)
  • Removed some tests, to free up time for the new ones. If we cut a test that you care about, create a discussion below.
    • Input Lag: 1080p @ 60Hz + HDR
    • Input Lag: 4k @ 60Hz @ 4:4:4 + 8 bit HDR
    • Supported Resolutions: 4k @ 30Hz @ 4:4:4
    • Input Specifications: MHL

Smart Features

  • Removed HDR in Amazon Video, to streamline testing


  • Added a "Market Context" section for future reviews, explaining where the TV fits in the overall TV market, how it compares to other TVs

41 TVs Updated

We have retested popular models. The test results for the following models have been converted to the new testing methodology. However, the text might be inconsistent with the new results.