DLSS4. We need proper tests for motion quality of frame generation

Ask about motion blur reduction in gaming monitors. Includes ULMB (Ultra Low Motion Blur), NVIDIA LightBoost, ASUS ELMB, BenQ/Zowie DyAc, ToastyX, black frame insertion (BFI), and now framerate-based motion blur reduction (framegen / LSS / etc).
1000WATT
Posts: 479
Joined: 22 Jul 2018, 05:44

Re: DLSS4. We need proper tests for motion quality of frame generation

Post by 1000WATT » 29 Jan 2025, 10:06

In any case, these are just ideas for now.)
RonsonPL wrote: The test would be done at the same "car" speed, let's say it' 600km/h
at 30fps source, it will just be a frame before the bridge, and then a frame after the bridge. At 120fps source, it will have the frame before the bridge, a frame at the start of the bridge, a frame at the end of the bridge, and a frame after the bridge. Much more data for FG to interpolate something decent.
It's easier for us to make this car go 150 km/h and people with 120 hz.
Than 600 km/h and people with 480 hz.
And all this is on the condition that the gtg mprt 120 hz screen does not become a limiting factor.

There is no difference between panning, racing or any other movement.
I often do not clearly state my thoughts. google translate is far from perfect. And in addition to the translator, I myself am mistaken. Do not take me seriously.

1000WATT
Posts: 479
Joined: 22 Jul 2018, 05:44

Re: DLSS4. We need proper tests for motion quality of frame generation

Post by 1000WATT » 29 Jan 2025, 10:52

Chief, like all of us, wants to speed up the path to 1000hz.
For this, any mention or even just talk about 480hz and interpolation will be welcome.
RonsonPL wants to avoid unnecessary blurring in the generated frames like "taa" does.
I'm just chatting here.
And so far I don’t see a lot of people willing to participate in this.
RonsonPL I think you'll have to be the first tester, and it doesn't matter that you don't have a 480hz display.
The main thing is to understand that it is quite simple with the right tools. Which will make the car go 150 km\h.
And you will choose the materials for testing yourself, this will be the beginning.
I often do not clearly state my thoughts. google translate is far from perfect. And in addition to the translator, I myself am mistaken. Do not take me seriously.

1000WATT
Posts: 479
Joined: 22 Jul 2018, 05:44

Re: DLSS4. We need proper tests for motion quality of frame generation

Post by 1000WATT » 29 Jan 2025, 11:16

I'll try again)

We take benchmark.
1.FG off 120 fps - FG on 480 fps. We need a 480hz monitor.

2.FG off 30 fps - FG on 120 fps.
+ Tools
Cheat Engine - game speed x0.25. We don't need a 480hz monitor.

The artifacts must be identical.
I often do not clearly state my thoughts. google translate is far from perfect. And in addition to the translator, I myself am mistaken. Do not take me seriously.

1000WATT
Posts: 479
Joined: 22 Jul 2018, 05:44

Re: DLSS4. We need proper tests for motion quality of frame generation

Post by 1000WATT » 29 Jan 2025, 14:10

I came up with the perfect scenario for popularization.
We do tests in FG off 24 fps. Next I head to the Offtopic Lounge and create a topic. The human eye can't see more than 24 fps, and reputable people have confirmed this with tests. Naturally, I provide a link to this topic. Success is guaranteed for many years.
+ - 10 hairs on the Site Admin head. Not a big loss.
I often do not clearly state my thoughts. google translate is far from perfect. And in addition to the translator, I myself am mistaken. Do not take me seriously.

User avatar
Chief Blur Buster
Site Admin
Posts: 11944
Joined: 05 Dec 2013, 15:44
Location: Toronto / Hamilton, Ontario, Canada
Contact:

Re: DLSS4. We need proper tests for motion quality of frame generation

Post by Chief Blur Buster » 03 Feb 2025, 14:04

1000WATT wrote:
29 Jan 2025, 14:10
The human eye can't see more than 24 fps, and reputable people have confirmed this with tests. Naturally, I provide a link to this topic. Success is guaranteed for many years.
+ - 10 hairs on the Site Admin head. Not a big loss.
Fortunately, I have lost no further hair over this.

The hair loss from old 30fps vs 60fps debates was already pre-deducted years ago with TestUFO's granddaddy.

Image
Head of Blur Busters - BlurBusters.com | TestUFO.com | Follow @BlurBusters on: BlueSky | Twitter | Facebook

Image
Forum Rules wrote:  1. Rule #1: Be Nice. This is published forum rule #1. Even To Newbies & People You Disagree With!
  2. Please report rule violations If you see a post that violates forum rules, then report the post.
  3. ALWAYS respect indie testers here. See how indies are bootstrapping Blur Busters research!

User avatar
Chief Blur Buster
Site Admin
Posts: 11944
Joined: 05 Dec 2013, 15:44
Location: Toronto / Hamilton, Ontario, Canada
Contact:

Re: DLSS4. We need proper tests for motion quality of frame generation

Post by Chief Blur Buster » 03 Feb 2025, 20:17

1000WATT wrote:
28 Jan 2025, 21:22
1. the car speed will be 30 km\h FG off 30fps
2. the car speed will be 120 km\h FG off 120fps
BTW, for the record, I am not a fan of pre-framegen framerates of lower than flicker fusion threshold.

If I *have* to use framegen as a blur busting technology (where the temporal clarity exceeds spatial blurring by a BIG margin), I prefer framegen 80fps->240fps over framegenning from 30fps.

At GtG=0, for regular perfect framepacing, the stutter threshold is equalized with the flicker fusion threshold, as seen in TestUFO Stutter to Blur Continuum (look at the 2nd UFO for 30 seconds).

Likewise, framegen artifacts are much more visible of you're framegenning from extremely low base framerates. You can also try to make sure your spatial framegen (resolution upscalers) creates a new high base framerate. Before doing anything related to the temporal dimension, aka temporal framegen (temporal upscaling by any means; such as interpolation/reprojection/extrapolation/warping/etc).

It's explained at Developer Best Practices that you want a high base framerate in order to reduce the framegen artifacts by a gigantic amount, and tip the pros:cons ratio seismically.

Also framegen is less effective on impulsed displays, and is more beneficial if you personally prefer framerate-based motion blur reduction on sample-and-hold display. So the wow-ratio (pros versus cons) is more useful on OLEDs than on LCDs, or on impulsed displays that already reduce motion blur for you.

The pick-poison Pandora Box of framegen have VERY different "pros versus cons" when your base framerate is above flicker fusion threshold, because of the TestUFO Stutter to Blur Continuum.
Head of Blur Busters - BlurBusters.com | TestUFO.com | Follow @BlurBusters on: BlueSky | Twitter | Facebook

Image
Forum Rules wrote:  1. Rule #1: Be Nice. This is published forum rule #1. Even To Newbies & People You Disagree With!
  2. Please report rule violations If you see a post that violates forum rules, then report the post.
  3. ALWAYS respect indie testers here. See how indies are bootstrapping Blur Busters research!

RonsonPL
Posts: 138
Joined: 26 Aug 2014, 07:12

Re: DLSS4. We need proper tests for motion quality of frame generation

Post by RonsonPL » 04 Feb 2025, 16:02

Chief Blur Buster wrote:
03 Feb 2025, 20:17

If I *have* to use framegen as a blur busting technology (where the temporal clarity exceeds spatial blurring by a BIG margin), I prefer framegen 80fps->240fps over framegenning from 30fps.
Sadly, most people think about 15-40fps when they hear about any frame gen/interpolation.
I think NV should give it a different name, so people could separate the framegen goals. For people like us, going below 80fps is just unacceptable. But most casual gamers don't even know what's the difference between motion smoothness and motion quality.


Anyway. May I ask about the earlier "This is my only reply and call-out at this juncture."?

So far we only have pieces of the puzzle. Your opinion after CES 2025 is the only word on MFG on RTX 5xxx.
Meanwhile I can see people talking about transformer model taking much longer time, which makes me concerned about it being viable for HFR (especially 250->1000Hz can be impossible if calculating the fake frames takes longer than the time between the frames (pre-FG)

I also saw posts on PC tech forums, suggesting that MFG generates more frames if the GPU is less busy (like CPU bottlenecked, for example) - more if the GPU load is smaller. This again creates concerns about "how much time it actually cost".
Surely resolution will matter too.

Do you plan to do any tests or even just "first impressions" if you plan on getting a new RTX yourself, or maybe you know about someone preparing the tests?

PS. Adding a long post still has issues. I log in, I start typing. I'm getting logged out cause I write too much, and then when I try add a PM/post, I get login page and then the text is gone. Temp-draft created automatically could solve this maybe? I actually have 0 coding skills/knowledge so that's the only idea I came up with. (Oh, and I'm behind a VPN, this can be a factor too, might trigger some checks (verification page) or affect the session timers, maybe?

User avatar
Chief Blur Buster
Site Admin
Posts: 11944
Joined: 05 Dec 2013, 15:44
Location: Toronto / Hamilton, Ontario, Canada
Contact:

Re: DLSS4. We need proper tests for motion quality of frame generation

Post by Chief Blur Buster » 07 Feb 2025, 16:45

RonsonPL wrote:
04 Feb 2025, 16:02
Anyway. May I ask about the earlier "This is my only reply and call-out at this juncture."?
Workload. The social media / forum / etc consumes too much of my time, given the powderkeg they generally are.

There's a lot of things going on, and I need to find active work. Blur Busters is a 50% services-based company. Post-CES I am also recruiting new clients so I can keep Blur Busters running.

I'm thinking of finally pushing the Patreon forward already, now that I have some publicity on the Blur Busters Open Source Display Initiative. I would like to focus on fan projects more, but for that part of Blur Busters I may need that Patreon. Economic pressures is pulling me away from forums / social media at the moment.
Head of Blur Busters - BlurBusters.com | TestUFO.com | Follow @BlurBusters on: BlueSky | Twitter | Facebook

Image
Forum Rules wrote:  1. Rule #1: Be Nice. This is published forum rule #1. Even To Newbies & People You Disagree With!
  2. Please report rule violations If you see a post that violates forum rules, then report the post.
  3. ALWAYS respect indie testers here. See how indies are bootstrapping Blur Busters research!

Post Reply