Aliasing Visibility Test bug?

NEW for 2017: Discussion about the testufo.com Blur Busters Motion Tests. Widely used by enthusiasts, display tweakers, YouTubers reviewers, monitor manufacturers and VR headset makers!
Post Reply
a_c_r_e_a_l
Posts: 36
Joined: 04 May 2020, 19:02

Aliasing Visibility Test bug?

Post by a_c_r_e_a_l » 26 May 2020, 15:33

Hi, I would like to share some observations during AA visibility test.

This is what I recorded with OBS:

https://streamable.com/76edn8

Look at left end and you can see it is looking quite ok but not as sharp as it supposed to be. On the right end it's looking completely different and jagging is out of control, kinda difficult to explain. I remember when I did this about 4 months ago and it was looking exactly like on this screenshot
AA test screenshot 1.png
AA test screenshot 1.png (6.3 KiB) Viewed 485 times
Any ideas?

I am freshly after format and all settings are default.

User avatar
Chief Blur Buster
Site Admin
Posts: 7699
Joined: 05 Dec 2013, 15:44
Location: Toronto, Ontario, Canada
Contact:

Re: Aliasing Visibility Test bug?

Post by Chief Blur Buster » 26 May 2020, 19:48

Which web browser are you using?

I use the browser / GPU driver default AA. Unfortunately, we've discovered that different browsers / different GPUs / different OSes, sometimes create some flaws.

Also, streaming compression can interfere with the AA quality.

In addition, if you video-record a TestUFO test, please see TestUFO Crediting Guidelines for what is allowed -- appreciated!
Head of Blur Busters - BlurBusters.com | TestUFO.com | Follow @BlurBusters on Twitter

       To support Blur Busters:
       • Official List of Best Gaming Monitors
       • List of G-SYNC Monitors
       • List of FreeSync Monitors
       • List of Ultrawide Monitors

a_c_r_e_a_l
Posts: 36
Joined: 04 May 2020, 19:02

Re: Aliasing Visibility Test bug?

Post by a_c_r_e_a_l » 26 May 2020, 20:09

Chief Blur Buster wrote:
26 May 2020, 19:48
In addition, if you video-record a TestUFO test, please see TestUFO Crediting Guidelines for what is allowed -- appreciated!
Sorry, I will follow that next time.

I tested it on Firefox, Chrome and Edge. All are giving same results. Right end of the line is displayed differently comparing to left end which makes no any sense. Like I mentioned, before everything was looking sharp and "stairs" were on both halves. Only what I found is that it may be related to stucked L2 Cache on GPU but so far nobody can confirm it and nvidia is disregarding that problem.

User avatar
Chief Blur Buster
Site Admin
Posts: 7699
Joined: 05 Dec 2013, 15:44
Location: Toronto, Ontario, Canada
Contact:

Re: Aliasing Visibility Test bug?

Post by Chief Blur Buster » 26 May 2020, 20:16

a_c_r_e_a_l wrote:
26 May 2020, 20:09
Sorry, I will follow that next time.
More about crediting options
And, if you are wanting to put these on a website of yours, why not EMBED TestUFO tests instead? It will automatically adapt to the browser visitor's screen resolution, and people can click on the embeds to launch the full TestUFO (bringing the visitor to Blur Busters). Embeds are a good quid pro quo.


a_c_r_e_a_l wrote:
26 May 2020, 20:09
Only what I found is that it may be related to stucked L2 Cache on GPU but so far nobody can confirm it and nvidia is disregarding that problem.
Fascinating! An L2 cache issue creating antialiasing issues?

Can you provide a link? I'd like to read up a bit more on this, and see if I can later program some countermeasures (including potential JavaScript software-based AA for 100% control over all pixels -- perhaps with a custom AA gamma correction slider). That said, there's a backlog of TestUFO tests to be invented -- however, this is a useful suggestion.

Also, make sure you are using 1:1 scaling -- and browser zoom 100% -- because a scaled video or a zoomed browser will show some AA artifacts beyond default look. View video in fullscreen mode on a 1080p display, not windowed. And if a viewer is using a different screen resolution, it will not look correct.
Head of Blur Busters - BlurBusters.com | TestUFO.com | Follow @BlurBusters on Twitter

       To support Blur Busters:
       • Official List of Best Gaming Monitors
       • List of G-SYNC Monitors
       • List of FreeSync Monitors
       • List of Ultrawide Monitors

a_c_r_e_a_l
Posts: 36
Joined: 04 May 2020, 19:02

Re: Aliasing Visibility Test bug?

Post by a_c_r_e_a_l » 26 May 2020, 21:01

Unfortunately I can't find any research regarding this but user SpeedUp10 wrote interesting conclusion:

https://www.nvidia.com/en-us/geforce/fo ... aa-tessel/

Test was done in full screen mode on native resolution 1:1 scaling so I am 100% sure that something is wrong and it makes me sick.

a_c_r_e_a_l
Posts: 36
Joined: 04 May 2020, 19:02

Re: Aliasing Visibility Test bug?

Post by a_c_r_e_a_l » 27 May 2020, 18:42

So what is your opinion about this? I am really curious.

User avatar
Chief Blur Buster
Site Admin
Posts: 7699
Joined: 05 Dec 2013, 15:44
Location: Toronto, Ontario, Canada
Contact:

Re: Aliasing Visibility Test bug?

Post by Chief Blur Buster » 11 Jun 2020, 13:24

a_c_r_e_a_l wrote:
27 May 2020, 18:42
So what is your opinion about this? I am really curious.
I was thinking: Have we fully confirmed that this is not a hidden scaling issue?

For example, your screenshot looks pixellated when you zoom: Zoomed version of your screenshot -- then use browser zoom. The AA disappear and it looks pixellated.

We also need to make sure that scaling is not a weak link -- we've seen different browsers behave differently.

If you're displaying using WIndows at any other setting than 100% DPI, TestUFO will scale up, creating scaling artifacts.

To compensate for non-1:1 mapping, I'll be modifying TestUFO to be more device pixel ratio aware, with support for zooming of selected tests (e.g. zoom the main testufo.com by 2x size). This will be useful for retina displays, and when Windows is configured to a DPI different.

There are multiple scaling weak links:
- GPU scaling
- Display scaling
- DPI setting (enter "Make Text Bigger" in search box)
- Browser zoom setting
- Accessibility zoom in OS
- Etc.

In other words; are we sure none is occuring?

I can make the TestUFO more DPI-aware on critical tests such as the aliasing visibility test.
Head of Blur Busters - BlurBusters.com | TestUFO.com | Follow @BlurBusters on Twitter

       To support Blur Busters:
       • Official List of Best Gaming Monitors
       • List of G-SYNC Monitors
       • List of FreeSync Monitors
       • List of Ultrawide Monitors

a_c_r_e_a_l
Posts: 36
Joined: 04 May 2020, 19:02

Re: Aliasing Visibility Test bug?

Post by a_c_r_e_a_l » 11 Jun 2020, 13:54

100% none is occuring, everything was scaled to 1:1, but what interesting is, I have some random 60Hz Asus as a replacement. Out of curiosity I checked AA test again and it was looking way way better than on my XL2546 which was showing AA off like there was AA on. Anyway I RMA my Benq and we will see because for sure there was a problem.

User avatar
Chief Blur Buster
Site Admin
Posts: 7699
Joined: 05 Dec 2013, 15:44
Location: Toronto, Ontario, Canada
Contact:

Re: Aliasing Visibility Test bug?

Post by Chief Blur Buster » 12 Jun 2020, 01:57

Sharpness adjustment? That can add ringing/banding artifacts even around AA.
Head of Blur Busters - BlurBusters.com | TestUFO.com | Follow @BlurBusters on Twitter

       To support Blur Busters:
       • Official List of Best Gaming Monitors
       • List of G-SYNC Monitors
       • List of FreeSync Monitors
       • List of Ultrawide Monitors

a_c_r_e_a_l
Posts: 36
Joined: 04 May 2020, 19:02

Re: Aliasing Visibility Test bug?

Post by a_c_r_e_a_l » 12 Jun 2020, 13:22

I was trying everything what I could with monitor settings, nvidia and drivers and unfortunately nothing helped. I will respond after I receive (I hope) new monitor and then we'll see if it's fixed or not. The only thing I didn't try was HDMI 2.0.

Post Reply