The magic of NTSC's artifacting...

[phaeron]

4 hours ago, MrFish said:

Would it be possible to add an NTSC option for this in Altirra, seeing it's possible on real hardware?

Needs more research. The existing chroma blending implementation is specific to the 1-H delay that was common in PAL displays and designed specifically for that video standard. It also has the visible artifact of moving chroma down by half a scanline, which results in a decreased saturation edge along the top of objects. This type of chroma filtering was not standardized in NTSC and not present in contemporary displays. None of the displays I have show this effect for an NTSC signal, modern or not.

 

There are other types of filters, including a two-line filter, which at least avoids the vertical shift. However, they still globally reduce chroma vertical resolution. An adaptive filter avoids this but such filters are more complex to implement and much harder to get information on. Implementing a chroma filter that allows cross-blending of chroma between lines in an interleaved Atari screen mode but doesn't show the reduced saturation edges would require some work.

 

If anything, the myth is that this is universally a characteristic of PAL, as there are PAL displays and capture devices that don't do this blending at all and show horizontal stripes for solid colors.

 

[+MrFish]

2 hours ago, phaeron said:

Needs more research. The existing chroma blending implementation is specific to the 1-H delay that was common in PAL displays and designed specifically for that video standard. It also has the visible artifact of moving chroma down by half a scanline, which results in a decreased saturation edge along the top of objects. This type of chroma filtering was not standardized in NTSC and not present in contemporary displays. None of the displays I have show this effect for an NTSC signal, modern or not.

 

There are other types of filters, including a two-line filter, which at least avoids the vertical shift. However, they still globally reduce chroma vertical resolution. An adaptive filter avoids this but such filters are more complex to implement and much harder to get information on. Implementing a chroma filter that allows cross-blending of chroma between lines in an interleaved Atari screen mode but doesn't show the reduced saturation edges would require some work.

 

If anything, the myth is that this is universally a characteristic of PAL, as there are PAL displays and capture devices that don't do this blending at all and show horizontal stripes for solid colors.

 

If it's too much work, no problem; I can live without it; I was just thinking it would be nice to mimic what, at least, some people can see on their sets. I know at least one other person who mentioned seeing it on his NTSC set: @Allas, developer of some ABBUC contest software. I have no idea what kind of monitor he used.

 

The set I'm using is a JVC BM-H1300SU (13 inch, CRT, studio monitor).

 

 

JVC BM-H1300SU - Specifications.pdf

 

[Faicuai]

1 hour ago, MrFish said:

 

If it's too much work, no problem; I can live without it; I was just thinking it would be nice to mimic what, at least, some people can see on their sets. I know at least one other person who mentioned seeing it on his NTSC set: @Allas, developer of some ABBUC contest software. I have no idea what kind of monitor he used.

 

The set I'm using is a JVC BM-H1300SU (13 inch, CRT, studio monitor).

 

 

JVC BM-H1300SU - Specifications.pdf 10.96 kB · 0 downloads

 

 

Glad to see we are both seeing the exact same results... over two entirely different video-paths, and terminal displays, as well.

 

I've come to the conclusion that NOT being able to reproduce this effect over composite input may probably be a shortcoming or deficiency, instead of just a favorable mishap.

 

We'll see what others see...

 

(PS: I like that JVC... 750 lines horizontal... not bad at all!)

Edited July 5, 2020 by Faicuai

[+MrFish]

Actually, I just noticed that the specs I posted above don't include information about the what filters available on the set. It has comb and notch filters, which are both selectable and can be used simultaneously. I don't have my stuff setup right now, so I'm not sure what I was using before. I would like to fire it up and toggle some of the options with some of these images/games that use blending, to see the differences on both s-video and composite.

 

[Faicuai]

3 minutes ago, MrFish said:

Actually, I just noticed that the specs I posted above don't include information about the what filters available on the set. It has comb and notch filters, which are both selectable and can be used simultaneously. I don't have my stuff setup right now, so I'm not sure what I was using before. I would like to fire it up and toggle some of the options with some of these images/games that use blending, to see the differences on both s-video and composite.

 

Consider using Flop's image as a test, since we have the results from several sources on this thread.

 

Let's see the filtering effect on it, and compare with samples already posted.

[+MrFish]

Speaking of artifacting, here's a trick I was able to coax my monitor into doing. It's a way (at least on my set) to get moderate artifacts when using s-video.

 

S-Video Artifacts

 

[Faicuai]

14 hours ago, MrFish said:

Speaking of artifacting, here's a trick I was able to coax my monitor into doing. It's a way (at least on my set) to get moderate artifacts when using s-video.

 

S-Video Artifacts

 

YES!

 

The first time I noticed a bit of "color creep" on B&W y/c output was looking closely at System's Information wallpaper (when you boot it with enough base-ram). Then with a couple of utilities for testing Artifacting, with pre-configured odd-and-even horizontal patterns... And then I saw it again yesterday, when looking at Fractari's B&W output... It is not color aliasing per-se, but it is a luminance-driven induction of (residual) color-response.

 

And as for what we are seeing here via Composite. here's PM-2.0, ran on my Vizio LCD TC, Sony Bravia large-screen TV, and [VP950b Monitor + DVDO iScan HD+ processor], ALL of them via COMPOSITE inpute:

 

VIZIO:

 

 

Sony Bravia LCD display:

 

 

Viewsonic VP950b (95% of NTSC gamut) + DVDO iScan HD+:

 

Notice that on the Bravia's picture, my iPhone's white-balance struggled hopelessly in the presence of strong natural back-light invading the living room. Image shifted all the way to the cool side, non-correctible during capture (it appears much warmer and closer to the other two, to the naked eye).

 

So there you have it:

 

Three different implementations, three different sources, three different years... all yielding the exact same result...

 

This is NO coincidence.

Edited July 5, 2020 by Faicuai