Calling all Harmony Cart users, for science!

[RevEng]

Some games flicker, others don't. Help me and my fellow homebrewers learn how larger scale flicker impacts 2600 gamers like you, by running the 2600 ROM attached here and identifying which statements you agree with in this poll. Feel free to select multiple statements if they apply to your experience.

 

Please participate only if you can run the ROM on real hardware. I know it's less convenient, but I don't want Operating System issues to cloud any results. Also, there's one section for LCD users and another for CRT users, so please only answer the one appropriate for you.

 

Your participation and input is greatly appreciated. Thanks!

2rectangles.bin

[Thomas Jentzsch]

How about using other flash carts?

[RevEng]

Indeed, and thanks for pointing that out. All other flash cart users are welcome to participate!

[Andromeda Stardust]

I have an LCD and a CRT, so I can test both...

[+Random Terrain]

I downloaded it to my Harmony cart and viewed it on my CRT (old TV). Both flicker, but the one on the right would make me suicidal if I had to look at it for very long. The one on the left isn't so bad, but I wouldn't want to play a game if everything was flickering like that. The one on the left is good enough for a title screen. The one on the right makes me want to punch a Puppymonkeybaby.

[chavert]

vcs(harmony) + Tv LCD :

 

One of the 2 rectangles displayed appeared to flicker

I percieved flickering, and I feel negative feelings toward it

If I had to stare at one rectangle for a length of time, I would prefer the one on the left

[+DrVenkman]

I tested using a real CRT, a mid-80's vintage Apple Color Composite Monitor. And boy howdy did the rectangle on the right flicker like a sonofabitch. My eye on this monitor showed no flicker on the left. Interestingly, a regular 30 fps video taken with my iPhone didn't show any flicker at all - I guess the camera sensor was pretty well in synch with the flicker, despite my own eyes and brain perceiving it clearly. However, regular snapshot showed it plain as day:

 

 

 

I need to go back and run a video with 240 fps slow motion and see what that shows.

 

Slow-motion video shows both images flickering, the one on the right about twice as fast as the one of the left. Regular-speed video shot at 60 fps also shows flickering on the right, though not the left.

Edited February 13, 2016 by DrVenkman

[RevEng]

Totally appreciate all of the tests and comments so far!

 

[...]Interestingly, a regular video taken with my iPhone didn't show any flicker at all - I guess the camera sensor was pretty well in synch with the flicker, despite my own eyes and brain perceiving it clearly.[...]

Very interesting side-info here. It seems folks can't use game footage to see if a game will be too flickery for their tastes. Good to know!

[Andromeda Stardust]

If anyone hasn't figured it out yet, the left rectangle is a gray placebo. The rectangle on the right is solid white and flashes every other frame. Concerning the flicker running through LCD display, produces a very fine venetian blinds effect courtesy of the TV's built in deinterlace filter.

 

I might even recommend a third flicker test, which I like to call "simulated 240i," used in the Harmony menu. This involves rendering every odd scanline line on one frame, and every even line on the next. Much like interlaced 480i, it flickers but is not noticeably bothersome on CRT, however the venetian blind effect will be noticeably more pronounced on an LCD display, producing much thicker horizontal bars.

 

I downloaded it to my Harmony cart and viewed it on my CRT (old TV). Both flicker, but the one on the right would make me suicidal if I had to look at it for very long. The one on the left isn't so bad, but I wouldn't want to play a game if everything was flickering like that. The one on the left is good enough for a title screen. The one on the right makes me want to punch a Puppymonkeybaby.

I would want to punch a puppymonkeybaby regardless of what is on my display. Doesn't matter. If I'm in the middle of the final Bowser fight playing Mario and a puppymonkeybaby shows up, I will stop playing, without even taking time to pause the game, and puppymonkeybaby will get punched, drop kicked, and punted through the uprights, even if it costs me the game and the Princess' welfare. But the rectangle on the right is pure evil, at least viewed on a CRT, and makes me want to scream and huff a kitten. And I would never, ever bring harm a cute, adorable kitten!

http://uncyclopedia.wikia.com/wiki/Kitten_Huffing

[+DrVenkman]

Totally appreciate all of the tests and comments so far!

 

 

Very interesting side-info here. It seems folks can't use game footage to see if a game will be too flickery for their tastes. Good to know!

 

 

Okay, I uploaded some video to Dropbox, hope the links work right and compression doesn't ruin the files:

 

This is 60 fps 1080p video:

 

https://www.dropbox.com/s/s7xi8kyuppghapn/60fps.mov?dl=0

 

This is 240 fps 720p video:

 

https://www.dropbox.com/s/xwrouu7020ku04d/240fps.mov?dl=0

 

I didn't bother uploading 30 fps video because the camera syncs up with the flicker and you can't see it. Or at least, my camera syncs - perhaps others would be out of sync and pick it up. I know my eye sure does on this ancient all-analog monitor!

[+Random Terrain]

If anyone hasn't figured it out yet, the left rectangle is a gray placebo.

So is the irritating one on the right making the one on the left look like it's slightly flickering?

[Andromeda Stardust]

So is the irritating one on the right making the one on the left look like it's slightly flickering?

Well technically, the 60Hz flicker is slightly above most people's perception, but 30Hz is well below it. One of our CRTs can tolerate NTSC 50Hz (bottom cropped off), and I tried playing some PAL N64 Tazmanian Devil game on my Everdrive 64, and got massive headache from trying to play PAL ROMs on it. 60Hz CRT refresh does not normally affect me. When I used to use CRT monitors (VGA), I would set the desktop resolution to 1152x864 and 70Hz frame rate, which caused notably less eye fatigue than at 60Hz.

 

Back to the VCS, the left rectangle is illuminated every frame at half brightness (60 times per second); the right rectangle is illuminated every other frame at full brightness (30 frames per second). On an LCD monitor, the deinterlace filter divides each frame into even and odd scanlines, and you get venetian blinds with any sprite that flickers at 30Hz.

[Thomas Jentzsch]

So is the irritating one on the right making the one on the left look like it's slightly flickering?

It could be the normal 60Hz flicker, which is noticeable to most people.

 

Also the right flickering frame influences the whole picture on a CRT. How much depends on the CRT. E.g. some are pumping, when the screen changes its overall brightness.

[RevEng]

Okay, I uploaded some video to Dropbox[...]

Thanks! Interesting to watch the 240fps and visibly see the scanlines being traced. I think the takeaway from all of them is that none of the video seems to faithfully reproduce the in-person experience.

 

[edit] One more take-away from the 240fps video. Even on the left rectangle, the phosphors are quite dim just prior to the scanline coming in. The "hang time" of the phosphors seems to be significantly less than one 60Hz period.

[Joe Musashi]

[edit] One more take-away from the 240fps video. Even on the left rectangle, the phosphors are quite dim just prior to the scanline coming in. The "hang time" of the phosphors seems to be significantly less than one 60Hz period.

The persistence varies with the type of phosphor being used. For CRT displays it's typically "short":

 

https://en.wikipedia.org/wiki/Phosphor

https://figshare.com/articles/_Phosphor_persistence_of_CRT_monitor_/607162

 

For comparison, a 60Hz frame is about 16ms long. You can also see this in this video:

 

 

The emission curve is actually a pulse. (Which is also the reason why motion on a CRT looks smoother that on a TFT, which has a stable emission.)

[Thomas Jentzsch]

Very interesting video. Red seems to have a significantly longer persistence. Is that normal?

 

The graph shows a pretty rapid rise and fall (just 1ms each for 80%). I wonder if the brightness is linear to the voltage.

[RevEng]

Agreed, very interesting! I had no idea the hang time was that short.

 

I do know that the brightness of the phosphors themselves isn't linear to voltage. That's what gamma correction is supposed to address.

[Andromeda Stardust]

Very interesting video. Red seems to have a significantly longer persistence. Is that normal?

 

The graph shows a pretty rapid rise and fall (just 1ms each for 80%). I wonder if the brightness is linear to the voltage.

Red is longer wavelength and lower energy than say blue, so it is reasonable to assume it persists longer. But it really depends on the composition of the phospher. Generally the brightness tapers off rapidly, then a much slower decay at lower emission levels. That is why you can see phospher trails on old arcade games with black backgrounds like Joust and Galaga. I have some dimmable LED candelabra bulbs in my chandelier at home, and you can see the yellow afterglow for sometime after the lights are cut off. Even my bedroom CRT screen glows faintly after I cut my bright CFL room lighting at night. The CFL globes also glow faintly for a few seconds, but but the time my pupils dialate enough to see them, the effect is mostly gone.

[Thomas Jentzsch]

I wonder how a better phosphor emulation in Stella should look like.

 

From the graph it seems, that within one frame (~16ms), all persistence afterglow is completely gone. Which would mean, that a flickering object has zero afterglow in the next frame(s). But that's not what your eye sees on a CRT.

 

Also the LCD has its own (heavily varying between different models and settings) persistence, which makes things even more complicated.

[Mr SQL]

RevEng, the rectangle on the right in your demo is painful to view, it looks worse than any classic Atari game element. What is it 10 hz? I'm guessing the one of the left is 30 hz because while it looks fine to me, some people have seen flicker.

 

I think StarBlitz confuses the flicker issue further (even when working properly in the emu, I'll have to post a video of that other type of "flicker") because the framerate for full screen animation is 30 FPS, giving the city animation the characteristic feel of classic Television; all of the objects are at 30 hz in the game and most people (if they can see minimal flicker at all) only see it on the city.

 

btw, try turning on phosphor in Stella with StarBlitz and you will see the display break; phosphor doesn't break the display at 30 FPS on the real hardware otherwise NTSC Television would turn blurry and painful to look at, it just mitigates flicker.

 

[Mr SQL]

I wonder how a better phosphor emulation in Stella should look like.

 

From the graph it seems, that within one frame (~16ms), all persistence afterglow is completely gone. Which would mean, that a flickering object has zero afterglow in the next frame(s). But that's not what your eye sees on a CRT.

 

Also the LCD has its own (heavily varying between different models and settings) persistence, which makes things even more complicated.

 

To emulate phosphor Stella would have to run at 120 FPS.

 

To emulate it better still, 240 FPS

[RevEng]

RevEng, the rectangle on the right in your demo is painful to view, it looks worse than any classic Atari game element. What is it 10 hz? I'm guessing the one of the left is 30 hz because while it looks fine to me, some people have seen flicker.

The rectangle on the left is displayed every 60Hz frame. The rectangle on the right is displayed at 30Hz.

[Lynxpro]

Please define "real hardware".

 

Shouldn't testers report what model of 2600 they're using and what type of output?

 

What about 7800 owners using the Harmony?

 

How about Atari 5200s with the Piggyback Parasite module with the Harmony plugged in?

 

What about Atari Flashback 2 and 2+ models with cartridge slots?

 

And then in the further subcategories, the Coleco Gemini, the Colecovision with 2600 adapter, the Intellivision 2 with 2600 adapter, and the illegal clones...

[RevEng]

Please define "real hardware".

Anything you'd usually plug a cart into to play 2600 homebrew, since that's our area of interest.

 

Since the video signal timing is driven by software, they'll be near identical. Certainly identical for the purposes of flicker detection.

[Mr SQL]

The rectangle on the left is displayed every 60Hz frame. The rectangle on the right is displayed at 30Hz.

Wow someone can still see flicker on CRT at twice the frequency of NTSC television; must have trouble playing Atari games

 

How did you get the one on the right to do that? It looks far less that 30 hz, I'll have to try the real hardware now, I didn't see it flicker in Stella.