Reply To: 1920 lines horizontal sampling/output for all inputs + aspect ratio controls

[marqs]

1. It is now possible in all modes with upsample2x, but not as straightforward as you noted. If you want exactly 1920 pixel active horizontal area, you should set samplerate to around 2046 and H.active to 1920 (or half of those if mode uses upsample2x). Since typically 75% of an analog scanline contains picture, your effective picture width would then be 0.75*2046 = 1534, which comes close to ideal 4:3 aspect when strected to full 16:9 screen. However, the picture will not be centered with those settings (and you have little room for backporch and synclen adjustments), so the centering logic from L5 modes would need to be adapted for other modes as well. As for manual adjustments not yielding same results as high res sampling would, I’m not sure what you mean. With upsample2x horizontal values are just duplicated internally in respective modes so there should be no difference. I cannot either fully agree on supersampling producing best results in all cases, but naturally it’s good to have that option available.

2. I’m afraid such flexibility is not really feasible with OSSC HW. An output with active width of 1920 requires a fixed sampling rate, so you’d always get that ~1534 pixels of picture from the digitizer. That could be naturally scaled to whatever portion of the said 1920-wide frame depending on source aspect, but I’m a bit doubtful whether a linear (or NN) scaling as a postprocessing step would look very good.