Testing the new Stella TIA core

[+stephena]

It has to do with the fact that the display is shifted somewhat. It seems to be inherent in how an NTSC signal is encoded, so I'm not sure it would even make sense to disable them, even if we could.

[Thomas Jentzsch]

It has to do with the fact that the display is shifted somewhat. It seems to be inherent in how an NTSC signal is encoded, so I'm not sure it would even make sense to disable them, even if we could.

I didn't want to criticize the lines at all, only wanted to state that they are not depending any of the parameters.

 

BTW: PAL shows them too.

[+stephena]

Yes, this is an idiosyncrasy of the TV effects. Technically they're only defined for NTSC modes, but we apply them for PAL too. This is mostly because I've never seen a PAL TV before, and I don't know what the output should look like.

[DirtyHairy]

I am curious about the actual origin of those lines, too. My theory is that this is an artifact caused by the shadowing that occurs at vertical high-contrast edges, both with Blargg and with actual TV signals. The Draconian logo uses flickerblinds, so each single frame shows a pattern of eight colored pixels alternating with eight blank pixels. This produces edges, and my suspicion is that the shadowing caused by these edges creates the dark lines that can be observed The same effect can be seen with other ROMs that use flickerblinds.

Edited July 25, 2017 by DirtyHairy

[+RevEng]

The Draconian logo uses flickerblinds, so each single frame shows a pattern of eight colored pixels alternating with eight blank pixels. This produces edges, and my suspicion is that the shadowing caused by these edges creates the dark lines that can be observed

In terms of the non-emulated effect, the edges are just real-world square waves (luma) taking a slower aount of time to rise vs fall. Flickerblinds relies on signal edges to line up...

 

       ______
______|      |______  odd frame

______        ______  
      |______|        even frame

But in the real world, we get something like...

 

        _____
_______/     |______  odd frame

_______       ______  
       |_____/        even frame
    

When I created my first flickerblinds display I was a bit bummed over the black gaps. But I suspect most people don't get as distracted by them as programmers do.

[DirtyHairy]

Interesting, thanks alot. So is the effect that I described as "shadowing" around edges actually the finite raise time of the luma edges?

[Thomas Jentzsch]

In terms of the non-emulated effect, the edges are just real-world square waves (luma) taking a slower amount of time to rise vs fall.

Vice versa, isn't it?

 

Both are not immediate, but because the lines are dark, the fall must be faster than the raise. And depending on the difference, the lines are more or less visible.

 

BTW: I wonder if phosphor plays a role here too.

Edited July 26, 2017 by Thomas Jentzsch

[+RevEng]

I believe I have it the right way around. If the fall were slower than rise, we'd have brighter overlap between the lined up regions. If it's easier, consider which scenario has more average high, or more average low.

 

I was originally going to comment on phosphor - I'm not sure how much it contributes, because I haven't checked to see if the effect is present on LCD. Whenever we talk about CRT phosphor, it also involves persistence of vision too. So square waves here has to mean everything from electrons in TIA to nerve impulses. Confirmation of the effect on LCD would put the reason for the effect back to signal, since LCD doesn't strongly trigger persistence of vision like CRT does, and has no phosphor.

[Thomas Jentzsch]

If the fall were slower than rise, we'd have brighter overlap between the lined up regions. If it's easier, consider which scenario has more average high, or more average low.

Agreed. I got you the wrong way, probably just my non-perfect English.

[+RevEng]

Agreed. I got you the wrong way, probably just my non-perfect English.

Looking at my original non-perfectly worded English, I see where the confusion occurred. The blame is all mine.

[devwebcl]

is there a way to run faster Stella emulator ?

(1000% etc, depending on the host machine's speed)

 

I found that feature in other emu's but I can't find that option in Stella

[Thomas Jentzsch]

In Video Settings you can set the Framerate to 300 and disable VSYNC.