A little while back, I modified a scaler for purposes of retro gaming. Simply put, a scaler is a bit of equipment whose job is to resize a picture (scaling). In my case however, I was most interested in a few side effects of using a scaler - I could separate out the picture information into RGB, and get a very clear picture, and importantly, this would allow for different outputs to be used. My objective was simple, I wanted a cost effective way to get a really, really good picture out of as many retro consoles as possible, and put that great picture on my newer TVs/displays.
One of the big reasons that people choose to keep or re-buy CRT TVs (or old studio monitors) is that new HDTVs have a certain amount of built-in lag, especially while processing analog signals like those coming from old consoles. Some HDTVs are worse than others, but to convert a signal from an analog source always takes at least some time. This time translates into a delay from when you press a button to when you see the result onscreen.
The clarity of an RGB picture is phenomenal. Just not phenomenal enough to put up with lag - if you can't land a move in Street Fighter, what's the point of seeing the picture better? Many, many older games require quick reflexes and timing. Lag can mess with timing in games to such a degree that the old Guitar Hero series added a "lag calibration" step in its option menu - intentionally allowing you to offset what counted as hitting a note! Good luck calibrating an old shoot em up - either you hit the button at the right time, or you're dead.
So when you plug in an old game system into an HDTV, what happens? HDTVs have scalers built into them that convert signals from our legacy connections to the TV, that is if you're lucky enough to own an HDTV that still supports all of the old connector types (like composite, the most common of the older analog connections). Every TV has a scaler, but they're not all created equal. Some muddy up the picture, some stretch things, some display artifacts, but all of them generate at least some lag. I set out to see whether or not the scaler I built performed about as well as the scaler built into the HDTV. If it does, then given the picture quality, I consider that a net "win". If it's worse, how much worse is it?
I started by just plugging a Sega Genesis into my HDTV (a Sony XBR4). Using Lightning Force, I cranked up the movement speed to max, and I started moving around. With the help of my lovely wife, I recorded a short (1:30) play session.
I took a video at 120fps, and slowed things down by 4x using YouTube's tools - here's the result. Bear in mind, this is a composite connection to the TV. As unbelievable as it seems, the video isn't really out of focus, that's just how bad the picture looks using the old connections blown up to a larger screen.
Slow Mo - Sega Genesis through Composite on Sony XBR4
As you can see in the video, there is some delay between pressing a directional key and the action of the ship. This isn't great, there's at least two-three frames passing between pressing the Dpad and the action seen on the screen (or releasing - that's a bit easier to see).
I wanted to know how this compared to the scaler running through the same TV. Would the lag be increased? Here were my thoughts going into the next test:
- There's some signal processing going on to make the VGA signal appear on the screen - was this going to cause any issues?
- Is the signal processing lag on the VGA input the same as the signal lag on the composite input?
- The scaler itself has to process a signal, at least minimally. Is the lag from the scaler enough to be perceived?
The idea here is to get closer to understanding whether or not this cheap scaler I built has added more delay than the benefit of the better picture was worth.
Here's my next test. This time, I've plugged the Genesis into the scaler I built, switched the HDTV to VGA input, and tried to record more or less the same type of actions as before. Again, I filmed this at 120fps, and then slowed things down by 4X.
Slow Mo - Sega Genesis through Gonbes 8220 Scaler on a Sony XBR4
Ugh, I'm not entirely sure if it's worse. But it's certainly still there. That delay - the feeling of muddy controls. But where did the fault lie?
Given that there was also delay between the directional key presses and the movement of the ship using the VGA scaler that I built, I wanted to see how much of this delay was actually due to the scaler, versus the delay built into the HDTV itself.
I pulled one of my gaming monitors off my PC. The ASUS VE248H is a fairly good PC monitor, with a very fast display (2ms gtg). It's as close as I can get to a CRT comparison (I don't have a CRT VGA monitor). I have a VGA splitter cable, so that I can display the same exact picture on the HDTV as on the monitor. I loaded a Sega CD called "The 240p test suite", and I snapped some pictures.
Here are the results. Holy crap, my HDTV is adding at least 30 milliseconds of delay (3 hundredths of a second) on the VGA port, relative to this monitor.
It was time for an additional test. While I was reasonably assured from the previous two videos that the amount of lag was comparable between the two ports (VGA vs composite), I wanted to see what a really good monitor would look like, relative to my HDTV.
The results are pretty telling, if you look at them. First off, for the first two minutes of this video, you can see that the HDTV is noticably lagging behind the monitor - check out the motion of the ship. Granted, in real-time, this isn't nearly as noticable, but it's still there, so it's still a factor in play.
Second, around the 2:15 mark of the below, I zoom in on the monitor. This gives you a better idea of the input lag produced by the scaler itself (assuming the monitor is a non-zero amount of lag, but very low). The lag produced by the scaler... isn't going to be a factor for me. Perhaps someone could calculate the lag for me, but it's looking like a frame of delay, or something similarly low.
Some of the delay seen after 2:15 might be the game itself rather than how the scaler is putting out the signal. All scalers will produce at least some lag, but when paired with a good monitor, this scaler seems to really show well. To this point, I haven't considered the scaler to be an issue, and after looking at these videos, I'm more certain of that.
Importantly though, this taught me a lesson about my HDTV. I won't use it for twitchy games. I'll use a monitor instead and the scaler instead! The lag is lower than the composite input on the HDTV.
If anything, this exercise really hammers home that the quality of the screen you play on is as important as the processor used to upscale your image, and any time you can take processing out of the equation (by gaming on a CRT for example) the better off you are.
That said, I really couldn't feel any lag whatsoever on the monitor + scaler combination, despite the slow mo video. As a result of this eye opening exercise, I decided to test the lag of other devices in my house.
Slow Mo - Gonbes 8220 Scaler on Sony XBR4 and ASUS VE248H
Below is the lag I experience when using my projector (it's about even with the lag from the Sony HDTV) through its VGA port.
Here's the lag from an LED HDTV I have in the house, again on the VGA port. Notice it's significantly improved from the other high def display devices I have. This TV has a "game" mode, which reduces the processing overhead considerably. With "game mode" turned off, the lag is similar to the other sets.
Finally, here's the lag from another computer monitor, relative to this ASUS monitor. It's much more comparible, but the ASUS still comes out just a bit ahead.
All told - I'm inclined to say that game monitors (or low persistence monitors in general) do have a provable difference in lag, especially relative to TVs. Maybe we should all be more aware of what we do our gaming on.
No comments were found for this thread.