Flicker frequency testing

[Dionoid]

Hi, I'm experimenting with flickering sprites on an Atari 2600 + real CRT. I always assumed that -when it comes to flickering objects- it's better to display the object as often as possible to get the least amount of perceived flickering; so e.g. displaying a sprite 2 out of 3 frames (67% visible) is better than displaying it 1 out of 2 frames (50% visible).

 

However, in my perception, a constant flicker-frequency of 30Hz (i.e. showing a sprite every other frame) looks less flickering/annoying compared to showing it 2 out of 3 frames.

And showing it 3 out of 4 frames is even worse. Is this just me?

 

Here is a ROM showing 4 sprites. Anyone wants to test this ROM on a real CRT and give me feedback on which of the bottom 3 sprites shows the least flickering? (Note: the top sprite is just for reference, and is not flickering).

 

flicker_test.bin

flicker_test_moving.bin

 

 

[Edit: when moving the object horizontally, the effect is even more noticable]

 

Edited January 5, 2021 by Dionoid

[RevEng]

It's a very counter-intuitive result, but indeed I perceive the flicker getting worse for each square that's lower on the screen. (save the first, of course) Tested on real hardware with a CRT.

 

Nice discovery! ?

[Thomas Jentzsch]

Interesting and unexpected finding.

 

I have experimented with variable luma boosts (e.g. 0, 0, 4) for 3/4, but that didn't improve the flicker. The flicker frequency is effectively lower for 2/3 (20Hz) and 3/4 (15Hz) than for 1/2 (30Hz). That seems to be the reason.

 

BTW#1: It might look better if every sprite is displayed at 30Hz, even if it is not flickering. That avoids switching between flicker on and off.

BTW#2: I found that the best boost for $0a at 30Hz on my PAL TV is +4. 

Edited January 1, 2021 by Thomas Jentzsch

[+ZeroPage Homebrew]

12 hours ago, Dionoid said:

Here is a ROM showing 4 sprites. Anyone wants to test this ROM on a real CRT and give me feedback on which of the bottom 3 sprites shows the least flickering? (Note: the top sprite is just for reference, and is not flickering).

Thanks for making this! This will also be extremely useful for testing purposes on digital hardware for how well flicker translates through capture devices and then displayed on various streaming and video hosting services.

 

- James

[+Andrew Davie]

In order of preference.

1

2

4

3

 

#3 (2 out of 3) has a weird-ass strobing on the verticals that hurts my eyes.

I suspect it's related (like photosensitive epilepsy) to individuals' response to flicker frequency as to what's preferable.

 

 

Another effect I noticed is that the one I am looking at always looks MUCH (i.e., 2x) brighter than all the others.

 

 

 

[Mr SQL]

This is pretty cool it's good to see other programmers being open minded with variable Hz flicker tests, great tests @Dionoid  

 

[Just Jeff]

Part of your flicker on #2, has to do with your scanline count/frame rate of 266/59.1.  That pulse that happens every second or so will go away if you get that to 262/60.  I think I can see that on the others as well though its not as distinguishable. 

[Dionoid]

On 1/3/2021 at 4:23 PM, Just Jeff said:

Part of your flicker on #2, has to do with your scanline count/frame rate of 266/59.1.  That pulse that happens every second or so will go away if you get that to 262/60.  I think I can see that on the others as well though its not as distinguishable. 

I missed that the frequency wasn't at 60Hz; I fixed that and updated the ROMs.

However I don't see why that would give a pulse on a CRT every second?

[+ZeroPage Homebrew]

23 minutes ago, Dionoid said:

I missed that the frequency wasn't at 60Hz; I fixed that and updated the ROMs.

However I don't see why that would give a pulse on a CRT every second?

I'm guessing it has to do with the same issue if you try to film a CRT at the wrong shutter speed, you need all of the frequencies lined up perfectly.

 

- James