3 ms·

50ms is pretty high, even by LCD standards. I have one of those MiSTer Laggy measuring things, and when I have my cheap Vizio TV in "Game Mode" the latency is a

by tombert 6mo ago

50ms is pretty high, even by LCD standards. I have one of those MiSTer Laggy measuring things, and when I have my cheap Vizio TV in "Game Mode" the latency is around 24ms, a little lower on the top of the screen and a little higher on the bottom, but still considerably lower than 50ms. Moreover, I think that OLEDs can get less than 10ms nowadays (though I do not have one to test at this moment). Since most retro games ran around 60fps, so about 17ms, we're talking about 1.5 frames of latency for the LCD, and about half a frame of latency for an OLED. With something like the MiSTer, you can also enable high speed USB polling, which I believe is roughly 1000hz. My understanding is that it doesn't work with all controllers, but it has worked with all the controllers I have tried it with. The composite video artifacts are definitely noticeable though; I noticed the weirdness of the waterfalls in Sonic when I was playing it recently. It doesn't bother me that much but I could see why it bothers other people.

jamiek88 6mo ago
Yeah with mister laggy measuring and my lg g1 oled (six years old now so it may have got better) in game mode latency is 8ms.
- pipes 6mo ago
  Same here, Samsung s95b QD oled, mister laggy tested it, as far as I can remember it's about 8ms. Also snac adapters by pass usb entirely and are pretty much zero lag as far as I understand. Retro arch has run ahead latency reduction etc, I'd like to see some comparisons of that Vs mister. I could do it myself but I've never got round to it. I've noticed that fiddling with latency reduction in retro arch really works, but it is a lot of fiddling.
  - tombert 6mo ago
    I did the preemptive frames thing with Retroarch with Sonic the Hedgehog 3 a couple years ago, and I certainly convinced myself that I could tell a huge difference...and then I kept taking hits and dying just as much as I was without doing anything. It's entirely possible that someone who is better at video games can tell a huge difference (e.g. speedrunners and the like), but I'm afraid that I'm not good enough at most games to be able to realistically tell much of a difference. I might still fiddle with it a bit; someone told me that it helps a lot with Mike Tyson's Punch Out, which is a game I have never beaten with an emulator.
    - jamiek88 6mo ago
      Interesting. I bought sonic origins as a palate cleanser the other day and I really feel like I can feel the latency. Sonic 1 was the only game me and my brother had for our mega drive so we know/knew everything there is to know!! Our speed runs were crazy. I don’t know if feeling the latency is just my age though, although I’m a semi pro SIM racer still and competitive in my late forties it’s a different kind of twitch reaction.
      - tombert 6mo ago
        It's certainly worth trying the preemptive frames in RetroArch to see if you like it. Pretty low risk experiment.
chongli 6mo ago
Since most retro games ran around 60fps, so about 17ms, That’s an oversimplification. Many retro game consoles don’t use a frame buffer. Instead they render the game state to the screen on the fly, one scanline at a time, and they’re able to process input mid-screen because they read the controller input many times faster than 60Hz (on the order of 2kHz). In practice, this means input lag is way below even 1ms. Lightgun games, for example, rely on very precise timing of the control input vs the CRT raster and simply do not work without a CRT.
- mrob 6mo ago
  >Lightgun games, for example, rely on very precise timing of the control input vs the CRT raster and simply do not work without a CRT. Perhaps the most famous light gun game of all time (Duck Hunt on the NES), does not rely on especially precise timing. It draws one white rectangle per frame over each duck when you pull the trigger and checks if the Zapper can see it. LCD latency will probably still break this, but it's not like the later Super Scope for the SNES that actually does track the precise raster position. I expect it would be possible to patch the timing in software to make it work for a specific model of LCD. But even if you did this, the Zapper also includes a bandpass filter at the CRT horizontal retrace rate (about 15kHz) to better reject other light sources, so you'd need to mod it to bypass that, or mod the LCD to strobe the backlight at the right frequency.
  - chongli 6mo ago
    It draws one white rectangle per frame over each duck when you pull the trigger and checks if the Zapper can see it Almost, but not quite. First it blanks the entire screen to solid black and uses that to calibrate the black level of the gun, then it draws a white rectangle over one duck on one frame, then a white rectangle over the other duck on the next frame. The NES could use this information to determine where the gun was pointing by firing an interrupt at the exact moment when the zapper’s photodiode reached a threshold brightness level above black, and then only register a hit if that occurred while the game was drawing the white rectangle. I think in reality the game didn’t care that much about the timing, only that a rising edge occurred after the fully black frame but before the return to a normal colour frame. Either way, an LCD doesn’t work because it can’t transition full black to full white within a one frame window. It sometimes works in the 2 duck mode, but it usually records a hit on the wrong duck. In any case, it requires black to white latency less than 16ms

by tombert 6mo ago

jamiek88 6mo ago

Yeah with mister laggy measuring and my lg g1 oled (six years old now so it may have got better) in game mode latency is 8ms.

pipes 6mo ago
Same here, Samsung s95b QD oled, mister laggy tested it, as far as I can remember it's about 8ms. Also snac adapters by pass usb entirely and are pretty much zero lag as far as I understand. Retro arch has run ahead latency reduction etc, I'd like to see some comparisons of that Vs mister. I could do it myself but I've never got round to it. I've noticed that fiddling with latency reduction in retro arch really works, but it is a lot of fiddling.
- tombert 6mo ago
  I did the preemptive frames thing with Retroarch with Sonic the Hedgehog 3 a couple years ago, and I certainly convinced myself that I could tell a huge difference...and then I kept taking hits and dying just as much as I was without doing anything. It's entirely possible that someone who is better at video games can tell a huge difference (e.g. speedrunners and the like), but I'm afraid that I'm not good enough at most games to be able to realistically tell much of a difference. I might still fiddle with it a bit; someone told me that it helps a lot with Mike Tyson's Punch Out, which is a game I have never beaten with an emulator.
  - jamiek88 6mo ago
    Interesting. I bought sonic origins as a palate cleanser the other day and I really feel like I can feel the latency. Sonic 1 was the only game me and my brother had for our mega drive so we know/knew everything there is to know!! Our speed runs were crazy. I don’t know if feeling the latency is just my age though, although I’m a semi pro SIM racer still and competitive in my late forties it’s a different kind of twitch reaction.
    - tombert 6mo ago
      It's certainly worth trying the preemptive frames in RetroArch to see if you like it. Pretty low risk experiment.

chongli 6mo ago

Since most retro games ran around 60fps, so about 17ms, That’s an oversimplification. Many retro game consoles don’t use a frame buffer. Instead they render the game state to the screen on the fly, one scanline at a time, and they’re able to process input mid-screen because they read the controller input many times faster than 60Hz (on the order of 2kHz). In practice, this means input lag is way below even 1ms. Lightgun games, for example, rely on very precise timing of the control input vs the CRT raster and simply do not work without a CRT.

mrob 6mo ago
>Lightgun games, for example, rely on very precise timing of the control input vs the CRT raster and simply do not work without a CRT. Perhaps the most famous light gun game of all time (Duck Hunt on the NES), does not rely on especially precise timing. It draws one white rectangle per frame over each duck when you pull the trigger and checks if the Zapper can see it. LCD latency will probably still break this, but it's not like the later Super Scope for the SNES that actually does track the precise raster position. I expect it would be possible to patch the timing in software to make it work for a specific model of LCD. But even if you did this, the Zapper also includes a bandpass filter at the CRT horizontal retrace rate (about 15kHz) to better reject other light sources, so you'd need to mod it to bypass that, or mod the LCD to strobe the backlight at the right frequency.
- chongli 6mo ago
  It draws one white rectangle per frame over each duck when you pull the trigger and checks if the Zapper can see it Almost, but not quite. First it blanks the entire screen to solid black and uses that to calibrate the black level of the gun, then it draws a white rectangle over one duck on one frame, then a white rectangle over the other duck on the next frame. The NES could use this information to determine where the gun was pointing by firing an interrupt at the exact moment when the zapper’s photodiode reached a threshold brightness level above black, and then only register a hit if that occurred while the game was drawing the white rectangle. I think in reality the game didn’t care that much about the timing, only that a rising edge occurred after the fully black frame but before the return to a normal colour frame. Either way, an LCD doesn’t work because it can’t transition full black to full white within a one frame window. It sometimes works in the 2 duck mode, but it usually records a hit on the wrong duck. In any case, it requires black to white latency less than 16ms