Lag testing in 240p test suite with a light gun?

[DirkSwizzler]

In my quest to calculate input lag on my setup I've gone through several phases.
1. 240p test suite manual test. "Hey look I only get 1.5 frames" (this would later be proven as 6-7)
2. 240p test suite timer + NES advantage + slow motion camera.
*A bit more accurate. But still a PITA and only works with my NES because of existing advantage led light.
3. Leo Bodnar HDMI lag tester.
*Easier to use, but only diagnoses the already HDMI portions of the setup and only at 1080p.
*It also didn't work on my main TV for some reason I never nailed down.
4. A cheapo 8" PVM and dual output.
*The most technically accurate and easy to grab a photo as long as I can get the systems side by side.

In this quest I definitely spent upwards of $200 on hardware that I'm not using day to day.


I just had a really dumb idea that might be hacked to work. The idea of the Leo Bodnar tester is that it outputs a detectable signal and the receiver has a photosensor to calculate the timings.

That more or less describes my understanding of hit scanning for a light gun. I know the timings have to be too precise to work with existing games.

BUT, given that the 240p test suite is custom software written in a time when input lag is a known thing. Could a test in the suite be written such that it continuously polls the light gun sensor to progressively narrow down timing of display lag.

It wouldn't be as accurate as a split signal to a CRT. But it would be WAY cheaper, and if it worked at all I bet it would be much more accurate than the manual test.

Just a thought.

[Fudoh]

If you still got the Leo Bodnar tester, add a HDMI to VGA converter and a scan converter and you can test all TVs with analogue 15khz input - composite, s-video, component or RGB. You can grab cheap scan converters like the Extron VSC units from ebay for very little money.

[DirkSwizzler]

If you still got the Leo Bodnar tester, add a HDMI to VGA converter and a scan converter and you can test all TVs with analogue 15khz input - composite, s-video, component or RGB. You can grab cheap scan converters like the Extron VSC units from ebay for very little money.

Sure, but I'm hoping that I may have stumbled across a better lag test for the gamer on a budget. Because I had a nes and a light gun before I started down this road at all.

[Guspaz]

There isn't any real reason for somebody to need to know the latency exactly, even a single decimal place in terms of number of frames of lag is probably overkill, and you should be able to do that with the zapper just fine. Something like this:

1) Screen starts out black with user pointing the gun at the screen (because we need to get a reading from the zapper on a black screen but we don't know how much latency there is so it needs to be black when the trigger is initially pulled)
2) User pulls trigger
3) NES reads zapper value to get black level
4) NES displays 24-pixel tall strip across the middle of the screen
5) NES begins polling the zapper 10 times per frame
6) As soon as the NES detects the change, it calculates the latency as roughly (n - 4.5) * (5/3) milliseconds. My math might be off, that's supposed to be the number of time slices elapsed minus the minimum delay from the box being in the middle of the screen times the number of milliseconds per slice.

The 10-times-per-frame thing might be tricky, I think on the NES you'd have to cycle count to time that, but maybe you could just use an MMC3 scanline counter to make it a bit easier.

[Artemio]

If you want the most precise and simple method, it is the fast Fournier transform with the dreamcast microphone.

However, it is dependant on the TV lagging the audio in sync with the video.

I did point Pinobatch to this thread, since he is in charge of the NES version of the suite

[Guspaz]

Alternatively, the XBox 360 supports 15 kHz video, and Rock Band 2 has a built-in audio and video latency test that uses a photodiode and microphone built in to the guitar. I'm not sure how accurate the test is.

[PinoBatch]

My Zap Ruder tech demo cycle-counts from the start of the frame to photodiode activation in order to determine how far up or down it's pointed. Operation Wolf and Top Hunter do the same thing.

But the real problem that keeps the Zapper from seeing LCDs at all is that it's designed to distinguish light from a TV from light from a light bulb. It does this by passing the signal from the photodiode through a demodulator sensitive to 15.7 kHz. The portion of the CRT in front of the photodiode flickers at 15.7 kHz as a few adjacent horizontal lines are being scanned. An incandescent bulb flickers at 0.12 kHz, a fluorescent tube at 0.06, and the demodulator filters these out.

[citrus3000psi]

It might be possible to replace R3 in the light gun with something much higher. People have lowered R3 to be able to work with 31Khz sets.

[PinoBatch]

If you can show that Zap Ruder responds to your 31.5 kHz, 45 kHz, or faster CRT, and you're willing to try other test ROMs, I might make one test ROM.