snax

Forum Replies Created

Viewing 3 posts - 1 through 3 (of 3 total)
  • Author
    Posts
  • in reply to: OSSC v1.xx series beta firmware #58393
    snax
    Participant

      Some more experimentation:

      • I didn’t realize Line6x uses pixel repetition.  Since this halves the effective horizontal resolution, it’s better to use Line5x if you want to use shadow masks.
      • At Line5x, A-Grille and TV masks look fine.  PVM is OK by itself, but it introduces beat patterns when combined with scanlines (since the scanline pattern repeats every five lines and the mask pattern every four)
      • I’ve noticed evidence of crosstalk in the green signal path (presumably between the FPGA and the HDMI transmitter?) when using shadow masks in Line5x.  There is pixel static/snow in areas with non-zero green components.  The static is not present on the input; turning off shadow masks fixes it, as does using Line4x.  Presumably the combination of the high pixel clock of Line5x and the frequent pixel value transitions introduced by shadow masks are causing a latent issue to manifest.
      in reply to: OSSC v1.xx series beta firmware #58376
      snax
      Participant

        Some random observations since modifying and upgrading my OSSC to 1.07:

        • Line6x works on my TV, but my USB3HDCAP chokes on it.  I suspect the resolution is just too high for it.
        • Video mode changes in Line6x tend to cause sound to cut out (e.g. FMVs in Mega Man X4 on PS1).  Toggling the audio downsampling setting restores sound.  Turning on full TX setup seems to prevent this issue.  Line5x and lower don’t seem affected.
        • I generally use custom sampling settings with 224 active lines to eliminate the vertical overscan/unsafe area entirely rather than just masking it off with a border.  When doing this in Line6x 256×240 optimized mode, my TV interprets the output as 16:9.  Setting the 256×240 aspect to 1:1 cancels this out and results in a roughly 4:3 image.
        • Shadow masks are cool, but seem a bit too aggressive given the available resolution.  Even in Line6x, the A-Grill setting has extremely chunky “phosphors” that cause significant color fringing on white text.  It also introduces an ugly shimmering effect on scrolling backgrounds.  The TV and PVM settings are less aggressive and look good from a distance when combined with scanlines.
        in reply to: OSSC v1.xx series beta firmware #58366
        snax
        Participant

          @marqs, would it be possible to allow the reverse LPF to take negative values and thereby function as a forward LPF?  This would be helpful for blurring dithering patterns in Genesis/Mega Drive games.  The RetroTINK-5X allows this with its equivalent “pre-emphasis” filter taking negative values.

        Viewing 3 posts - 1 through 3 (of 3 total)