Page 3 of 3

Re: First-Timer: XG2431 Pursuit Camera Photos By Hand-Wave Smartphone

Posted: 24 Nov 2021, 02:00
by Chief Blur Buster
woland wrote:
24 Nov 2021, 01:49
Hi Chief, you've mentioned in numerous posts now that "CS:GO will not work well with this (bad mouse math at 1600dpi and 3200dpi)" - what exactly happens in csgo at 1600-3200 dpi?
The motion becomes a bit more wonky / less accurate. CS:GO is a very old game; it uses an engine almost 20 years old so it's probably using integer math or single-precision floats when doing mouse mathematics. 3200dpi 1/8sens does not move the mouse at the same speed as 400dpi original sensitivity. Etc. Many esports gamers complain it feels bad at high DPI.

Newer games properly do subpixel mouseturns (e.g. turning 1dpi mouseturns 0.1dpi -- it simply is antialiased motion.

Re: First-Timer: XG2431 Pursuit Camera Photos By Hand-Wave Smartphone

Posted: 24 Nov 2021, 14:01
by teo
re-reading both the QFT how-to forum post and the xg2431 web article, I believe I'm following the instructions correctly. the native 240hz VT is 1125, but every VT I've tried above 4200 (including 4525) result in either no image, or black screens every few seconds. I also don't have any cross-talk on the last row of my "60"hz--that's entirely artifact in my image.
60_settings.JPG
60_settings.JPG (72.39 KiB) Viewed 230 times
120_settings.JPG
120_settings.JPG (70.24 KiB) Viewed 230 times

in looking at the panning map test, at 240hz the clarity difference between the strobing modes stands out a bit more but they're still very small increments and overall screen brightness is more noticeable to me. reading text is actually the only place within a game I feel like I perceive a meaningful clarity improvement with any of the 240hz strobe presets--albeit in an artificial environment as I drove around in a circle looking at some background signs in rocket league. I don't really think of myself as a fixed-gaze type player, but it's entirely possible that what I think I do vs what my eyes really do is quite different.

now, the panning map test at 6(4)hz and 120hz with QFT/custom strobe settings is clear as day. I did give both of those refresh rates a bit of latitude on screen brightness (15% persistence) but it's impressive nevertheless. that only reinforces my surprise that sonic mania (vsync on with a game pad) did not show similar (perceived) performance with my 64hz QFT/custom pure xp vs unstrobed. I went back and confirmed via monitor OSD that I'm at the correct refresh rate. I actually found a section with text that I could reliably run past at speed--sure enough, it's much sharper with strobing on. I still don't experience that "aha" moment when I enable pure xp, but I'll give it another try with other games at 120hz.

I've been playing valorant and apex legends (although the arena mode where framerate is far more stable) at 240hz with pure xp on normal. I have been bouncing around between vsync on, vsync off uncapped*, vsync off 240 cap. I definitely notice the fluidity gain with vsync on, but I also think it feels a bit less responsive. I'm not convinced that lag perception isn't placebo (vsync lag at 240hz should be what, at most 8ms?). I've been using a viper 8k at 3200 dpi, generally at 2000hz because apex doesn't like 8k. overwatch seems like a good test candidate so I'll add that in.

*this is 260-300fps, so not exactly overkill

Re: First-Timer: XG2431 Pursuit Camera Photos By Hand-Wave Smartphone

Posted: 25 Nov 2021, 01:48
by Chief Blur Buster
teo wrote:
24 Nov 2021, 14:01
I'm not convinced that lag perception isn't placebo (vsync lag at 240hz should be what, at most 8ms?).
It is not placebo.

8ms is very human detectable, and even 1ms can be human detectable in some cases, depending on lag context.

Please read The Amazing Human Visible Feats Of The Millisecond, a post I wrote.

Back to topic, for a lower lag clone of VSYNC ON, you can can use RTSS Scanline Sync or the superior Special K Latent Sync (new clone of RTSS Sync), though anticheat might flag Special K — so your mileage may vary.
teo wrote:
24 Nov 2021, 14:01
I've been using a viper 8k at 3200 dpi, generally at 2000hz because apex doesn't like 8k. overwatch seems like a good test candidate so I'll add that in.
You’re good, then.

In these special cases, you can move the radio button to the “Hz” and then edit the Hz lower. Usually this is acceptable once you’ve created a too-high-Hz QFT mode, so you can usually edit it into a lower Hz. Try that.

Follow existing QFT instructions, get your Hz as low as you can by making VT as big as possible, and once that fails, move the radio button to the Hz and then edit the Hz (e.g. edit 64 to 60 without changing any other numbers).

Re: First-Timer: XG2431 Pursuit Camera Photos By Hand-Wave Smartphone

Posted: 26 Nov 2021, 20:35
by teo
Chief Blur Buster wrote:
25 Nov 2021, 01:48
It is not placebo.
to be clear, my statement was in reference to my individual perception rather than the entire human (or game playing) population. I suspect my reaction time is on the slower side, and certainly a standard deviation or two below an esports professional/"good" player. my in-game performance (even within a match) can also fluctuate wildly which makes me think that there's a lot of inconsistency in the human portion of my personal latency chain. I know that vsync introduces additional latency, but I have a healthy skepticism of my conscious perception limits--i.e. I'm not sure if I feel that latency because I "should" or because I really do. at the same time I know I don't have to "feel" a change to benefit from it and maybe I'm not giving myself enough credit.

Chief Blur Buster wrote:
25 Nov 2021, 01:48
Back to topic, for a lower lag clone of VSYNC ON, you can can use RTSS Scanline Sync or the superior Special K Latent Sync (new clone of RTSS Sync), though anticheat might flag Special K — so your mileage may vary.
in any case, I'm getting more confident that I'm more tear-sensitive than blur-sensitive so I'll look up special k latent sync. a few months ago I played around with RTSS scanline sync just to see how it works, and I also felt like it added a touch of latency vs no sync at the same frame rate. from what I understand of that technique, shouldn't that not be true?

Chief Blur Buster wrote:
25 Nov 2021, 01:48
Follow existing QFT instructions, get your Hz as low as you can by making VT as big as possible, and once that fails, move the radio button to the Hz and then edit the Hz (e.g. edit 64 to 60 without changing any other numbers).
to make sure I'm understanding: when manually lowering HZ after having set the VT to its functional large limit, the horizontal frequency then goes down. this is lowering the scan rate, so I'm (very slightly) losing some of the benefit of QFT though it's still much better than the stock 60hz mode.

Re: First-Timer: XG2431 Pursuit Camera Photos By Hand-Wave Smartphone

Posted: 27 Nov 2021, 13:28
by Chief Blur Buster
teo wrote:
26 Nov 2021, 20:35
Chief Blur Buster wrote:
25 Nov 2021, 01:48
It is not placebo.
to be clear, my statement was in reference to my individual perception rather than the entire human (or game playing) population. I suspect my reaction time is on the slower side, and certainly a standard deviation or two below an esports professional/"good" player. my in-game performance (even within a match) can also fluctuate wildly which makes me think that there's a lot of inconsistency in the human portion of my personal latency chain. I know that vsync introduces additional latency, but I have a healthy skepticism of my conscious perception limits--i.e. I'm not sure if I feel that latency because I "should" or because I really do. at the same time I know I don't have to "feel" a change to benefit from it and maybe I'm not giving myself enough credit.
That's all fair and there's a lot of error margins that hides these latencies below the noise floor.

So a lot of these milliseconds don't matter, but other milliseconds do massively.

Interesting fact:When it comes to strobe backlights, even 10 microseconds is human visible, since 1.0ms MPRT strobe flash versus 1.01ms MPRT strobe flash (10 microsecond difference) is a 1% difference in photons, which is tantamount to a bigger brightness difference than RGB(253,253,253) versus RGB(255,255,255). So sometimes 10 milliseconds isn't human visible, but 10 MICROSECONDS is human visible -- depending on WHICH technological milliseconds or microsecond is being done. ;)
teo wrote:
26 Nov 2021, 20:35
Chief Blur Buster wrote:
25 Nov 2021, 01:48
Back to topic, for a lower lag clone of VSYNC ON, you can can use RTSS Scanline Sync or the superior Special K Latent Sync (new clone of RTSS Sync), though anticheat might flag Special K — so your mileage may vary.
in any case, I'm getting more confident that I'm more tear-sensitive than blur-sensitive so I'll look up special k latent sync. a few months ago I played around with RTSS scanline sync just to see how it works, and I also felt like it added a touch of latency vs no sync at the same frame rate. from what I understand of that technique, shouldn't that not be true?
This is true, but we need to use the world's lowest latency tear-free sync technologies available to fixed-Hz non-VRR strobed modes, and that generally is the scanline sync methods.

teo wrote:
26 Nov 2021, 20:35
Chief Blur Buster wrote:
25 Nov 2021, 01:48
Follow existing QFT instructions, get your Hz as low as you can by making VT as big as possible, and once that fails, move the radio button to the Hz and then edit the Hz (e.g. edit 64 to 60 without changing any other numbers).
to make sure I'm understanding: when manually lowering HZ after having set the VT to its functional large limit, the horizontal frequency then goes down. this is lowering the scan rate, so I'm (very slightly) losing some of the benefit of QFT though it's still much better than the stock 60hz mode.
Correct, but you're still getting close. You're getting 240/64ths of the benefit instead of 240/60th of the benefit. So you're only getting 3.75x better scanout rather than 4x better scanout. Your screen is refreshing in about 1/225sec instead of 1/240sec. We should long-term figure out why your specific configuration is not doing a full 1/240sec QFT at 60Hz, since my XG2431 on my desk is doing it fine. However, the difference between 1/225sec QFT for 60Hz and 1/240sec QFT for 60Hz is extremely tiny since that's just a (4.44ms - 4.16ms = 0.28ms difference in VBI size).

At 60Hz VT4500 on a 1080, you're got a blanking interval of (4500-1080)/4500ths = 3420/4500ths of 1/60sec = about 12.6ms VBI (to hide IPS 1ms GtG90% more completely). Your VT reached only 4200 so your blanking interval time in milliseconds is (4200-1080)/4200 = 3120/4200ths of 1/60sec = 12.4ms VBI.

So your VBI difference for your mode is 12.4ms versus my 12.6ms. That becomes a mere rounding error in the works of hiding 1ms IPS GtG in the dark period between refresh cycles.

So, I wouldn't worry about it much if you're using a sub-100Hz mode with an above-4000 vertical total...

You're pretty much crosstalk free at 60Hz, as seen in your crop of your pursuit:
60hz-crosstalk-free-pursuit.png
60hz-crosstalk-free-pursuit.png (3.64 MiB) Viewed 92 times

Re: First-Timer: XG2431 Pursuit Camera Photos By Hand-Wave Smartphone

Posted: 28 Nov 2021, 10:09
by Discorz
teo wrote:
24 Nov 2021, 14:01
We don't have a good quality 240Hz full screen strobe crosstalk pursuit. Can you do one? Something like you did here. Someone did it before, but its not a good quality image.

Re: First-Timer: XG2431 Pursuit Camera Photos By Hand-Wave Smartphone

Posted: 28 Nov 2021, 17:42
by teo
Discorz wrote:
Yesterday, 10:09
We don't have a good quality 240Hz full screen strobe crosstalk pursuit. Can you do one?
I took a few stationary shots at 1/240 shutter speed for pure xp normal, extreme, and ultra. I still don't quite understand how/why shutter artifacts occur, or how to mitigate them. I took three of each (except for normal, which I took another 3 after decreasing ISO) which you can find here. let me know if there's something simple I'm missing which would result in cleaner shots.

Re: First-Timer: XG2431 Pursuit Camera Photos By Hand-Wave Smartphone

Posted: 29 Nov 2021, 02:40
by Discorz
teo wrote:
Yesterday, 17:42
I took a few stationary shots at 1/240 shutter speed
Oh, that doesn't look good. I don't recommend taking stationary images. But for those who know how to look, something can still be visible out of these.

I believe these should be representable, but correct me if I'm wrong:

Normal
IMG_1834 column 4 5 7
IMG_1839 column 6 7

Extreme
IMG_1831 column 6 7 8
IMG_1832 column 1 2 3 4 5 6 8