• Software
  • Intel iGPU Driver Patch Notes - Dithering

Intel HD Graphics Driver 20.19.15.4835 64-bit - Driver from 2017 (patch notes below in quotation)

This may be old news. And also some say Intel dithering is not harmful (just spatial). But I posted it anyway because it's interesting. I'm using 10.18.15.4279 driver from 2015 on @Sunspark recommendation and finding it good (this was previous to this change).

  • Issue seen because dithering was disabled by default on external display. Fix was added by introducing a new RegKey to enable dithering.

    Good job finding that patch note. I hadn't seen it before.

    I think the OS and other components also affect which driver is ok.. for example, with my current setup on my HD 6000, I am actually finding 20.19.15.4624 (automatically pushed) in W10 19042 more comfortable than 4279 on W7. I tried 4279 in W10 and it was not as good (possibly not very compatible due to the age).

    degen

    I've heard that on Linux, at least, spatiotemporal dithering is disabled and only spatial dithering is used. You can see the Intel Graphics mailing list post here about eyestrain, quoted below:

    I'm really confused about what could be causing this. Resolutions don't matter; you always get the eyestrain. My first thought was temporal dithering (or spatiotemporal dithering), but my reading of the intel driver source is that the kernel drivers have disabled ST dithering in favor of spatial dithering.

    I wrote to the gentlemen who made that post, it took him about 3 months to get used to the display, but after that, he was fine. He did not get any diagnosis of a BVD nor pursued vision therapy nor prism glasses.


    But, I wrote to another gentlemen who replied on that Intel eyestrain thread. He wound up going to an orthoptist and he had to get eye exercises which resolved the issues. Just another report of vision therapy working! He adds:

    ...my conclusion is that there was indeed something different with that machine. However, the muscles of my eyes should have been able to change positions in order to be able to adapt myself to that difference. Therefore, the problem was myself and not the machine.

    I should add that the ditherig program is open source and available here.

    a month later

    Also the problem with relying on singular commits like these is the initialization "pipeline" of GPU drivers is very complex, and in my experience there can be duplicated or separate functionality for the differences between a eDP (laptop display) and for dedicated GPU outputs, etc.

    dev