e8root wrote: ↑Sun Feb 22, 2026 10:18 pm
For VGA CRTs? CRTs with native 480p?
IMHO for (S)VGA CRT monitors it still better use modes that have nice CRT filters to get 15KHz look but for square look or scanline look pure line doublers make perfect sense for CRTs and will technically have least latency.
Technically, it's less lag. But, we're talking about a ridiculously small amount of time, here.
If we're being super picky, a CRT will always have it's own characteristics and native scanlines. We can try to add a filter on top of that, but it never quite looks the same as a native 15kHz CRT.
Of course, I can add blank lines and filters with an XRGB-1, XRGB-2, OSSC, or one of the new community scalers and get lag that's too small to detect using a Time Sleuth. I could use a line doubler, but I don't have to. I can get filters and blank lines from the other machines with essentially no lag.
The DAC is where people run into problems. The real appeal of the OSSC Pro with a CRT is the DAC expansion card. External DACs are a pain the arse.
I see the advantage of the OSSC Pro for feeding a CRT, but it's because of the DAC card, not the size of the frame buffer or upscaling algorithm.
e8root wrote:
BTW. <1ms means the implementation is optimal.
Yes. That's one of the main reasons why these community gaming machines exist, no? For optimal performance.
Video gear manufacturers for home, event, and commercial installations made it crystal clear (over the years) that they care very little about latency and gamers.
Customer service has been bad as well. I've heard horror stories about dealing with Extron on the forum (although I had better luck). I wouldn't say I've had negative experiences overall, but I can tell you that gear manufacturers rarely take gaming seriously.
That's how these community machines came to be...
e8root wrote:
For SD to HD (or reverse) you get almost 1ms but only on either top or bottom and on the other end lag goes to almost zero.
I don't folllow. The information arrives on the wire from the source machine, it is processed by the video processor, and it continues up the wire to the display. What's changing? Are you talking about blanking?
I understand that frame rate can affect lag. For instance, the latency can be reduced if our output frame rate is higher than the input frame rate, because we can get information up the wire to the display fast. For instance, we could use 240Hz output to handle a fully frame buffered 60Hz source. If we have a 60Hz game console source, it takes ~17ms to transmit from the source machine to the video processor and buffer it all. The signal is buffered and processed in ~17ms in this example. The final step is to send the completed frame to the display. If we transmitted at 60Hz, it takes about 17ms to send the frame. At 240Hz, the wire is faster and we send the new frame in ~4ms. (We can repeat the frame afterwards or add BFI.) Regardless, the new frame appears on screen after ~4ms. That means: the video processing penalty for using a full frame buffer with a 60Hz source (for something like deinterlacing or tate rotation) will be ~17ms at the top of your screen and ~21ms at the bottom of your screen: plus any additional lag your display may add (if any at all). Display lag will be extra if you don't have essentially direct scanout. (A 60Hz output would send the information to the display slower and you get ~17ms at the top of the screen and ~34ms at the bottom.)
Native scanout of a 60Hz source connected directly would be ~17ms at the bottom of the screen on a CRT, so the full frame buffer processing adds ~4ms of lag with 240Hz output, because we got ~21ms at the bottom.
I know about that scenerio. I don't follow what you're saying, though. If it's all within a millisecond of variance, I don't care very much (anyhow). That's not enough time to matter for me.