V an H Blanking interupts - timings @ 50hz

GFA, ASM, STOS, ...

Moderators: simonsunnyboy, Mug UK, Zorro 2, Moderator Team

Rybags
Retro freak
Retro freak
Posts: 15
Joined: Sat Jan 31, 2009 2:57 pm

Re: V an H Blanking interupts - timings @ 50hz

Post by Rybags »

I'd hardly trust a capture card to verify or deny the existence of an interlaced signal.

2 of my previous capture cards (old Buster II, Zoran chipset and an ATI 9550 with VIVO) had combing effects like that practically all the time with old computers connected to them.

My current card (Compro hybrid DTV + analog) has a Deinterlace setting - all that does is perform blending to remove the combing effect. It still generates a 576 line display from my Atari 130XE by treating each "288p" Atari frame as seperate fields of a 576i display.

With a single setting in my capture software, I can generate the exact same effect in a video (sample here of a single frame while moving the cursor on the 130XE - no interlace tricks, just the way the capture card is behaving).
130XE cap1.JPG
You do not have the required permissions to view the files attached to this post.
Last edited by Rybags on Fri Feb 06, 2009 3:45 pm, edited 1 time in total.
ppera

Re: V an H Blanking interupts - timings @ 50hz

Post by ppera »

That was not the point of my post.
I just wanted to confirm what you wrote about non-existence of Field-info.

Then, linked capture is not capture of some normal interlaced video. By normal interlace we have 1/50th sec between fields.
But, as it is actually progressive video with repeating frames 2x (game has 25 fps) we may have displaying problems with wrong field order on displays working in interlaced mode.
Plus, visible interlace by horizontal movement is 2x bigger than by regular interlaced video, as time diff. is 1/25 sec.

So, it has nothing with 'trust capture card' .
But capture card is good to show us that displaying depends not only from input signal's type, but display itself - in case of capture card: it's settings. If use Virtual Dub for capture, may set almost everything.
I doub that your cards have mentioned blendings etc. in HW . All it depends from used driver and it's settings.

P.S. it is normal displaying, as capture card is intended for TV watching. But you may set it to show 50 fps, progressive in mentioned Virtual Dub for instance.
Last edited by ppera on Fri Feb 06, 2009 3:50 pm, edited 1 time in total.
Rybags
Retro freak
Retro freak
Posts: 15
Joined: Sat Jan 31, 2009 2:57 pm

Re: V an H Blanking interupts - timings @ 50hz

Post by Rybags »

You probably replied before I added that pic.

Your Nebulus capture was no-doubt just experiencing the same effect I was able to generate with a simple settings tweak in the capture software.
ppera

Re: V an H Blanking interupts - timings @ 50hz

Post by ppera »

Rybags wrote:You probably replied before I added that pic.

Your Nebulus capture was no-doubt just experiencing the same effect I was able to generate with a simple settings tweak in the capture software.
No, you don't get it. Interlace in case of Nebulus is something different. It is caused by displaying fields from 2 different frames of progressive video, and therefore interlace. If you change field order you get correct video, where even and odd lines are same. In case of your cursor movement they are never same as long as you move cursor - no field order swap will change it.
Rybags
Retro freak
Retro freak
Posts: 15
Joined: Sat Jan 31, 2009 2:57 pm

Re: V an H Blanking interupts - timings @ 50hz

Post by Rybags »

So, essentially what you're saying is that - all those years ago (1988?), they programmed Nebulus on the ST, and deliberately moved everything at 25 fps because they knew about the undocumented interlaced mode that nobody else did?

And, somehow, they were able to fluke getting the field order right such that nobody ever got combing effects? Despite the ST not having any means to communicate to a programmer which particular field the Shifter chip is outputting at a given time?

Too far-fetched, I think.
Rybags
Retro freak
Retro freak
Posts: 15
Joined: Sat Jan 31, 2009 2:57 pm

Re: V an H Blanking interupts - timings @ 50hz

Post by Rybags »

OK. Maybe my supposition was out a bit on Nebulus. 25 fps should work on interlace anyway since you have a static field which would negate combing.

So, let's forget that game for a minute. But consider probably 90% of other games out there that operate at a full 50 fps. Why is it that I've never experienced any combing effect when playing them?
ijor
Hardware Guru
Hardware Guru
Posts: 4012
Joined: Sat May 29, 2004 7:52 pm
Contact:

Re: V an H Blanking interupts - timings @ 50hz

Post by ijor »

ppera wrote:I don't care what writes on diverse sites. I believe to my eyes in first place...
Finally (I'm really bored with this all) I don't care about your opinion.
...so I don't have what to admit.
...
Your attitude, Ijor is very bad. You play judge here....as you are not some persona over others.
...
We are here to exchange our opinions, experiences.
What ???

You don't care about my opinion, you don't care about other references or other places, you only believe in your own eyes, etc. That's your concept of "exchanging opinions"?

You further seem to take the word "admit" as some kind of insult. Who is the one that is having a bad attitude here?
ppera

Re: V an H Blanking interupts - timings @ 50hz

Post by ppera »

To me more precise, I don't care about your opinion about me. Your technical postings are here welcome, but it means not that I will put theory behind practice. In case of posted interlaced pic example you remained silent. So, please present your opinion about it.
'Admit' is not insult. What is insulting is your twisting of my words.
ijor
Hardware Guru
Hardware Guru
Posts: 4012
Joined: Sat May 29, 2004 7:52 pm
Contact:

Re: V an H Blanking interupts - timings @ 50hz

Post by ijor »

ppera wrote:'Admit' is not insult. What is insulting is your twisting of my words.
That I did what ???!!! I twisted your words and that was insulting ???

I honestly think that you previously claimed that the ST outputs a (more or less) interlaced video signal, and that now you are agreeing it is not, but you are claiming (right or wong) that the TV is interlacing the display even when the incoming signal is not interlaced.

I might have misunderstood you. But even then, misunderstanding and "twisting" as insulting are two very different things.

I don't want to quote now all your posts (in this and in the older thread) that you were (seeming) to claim that the ST signal was interlaced. But I think it was obvious since, among other things, you were constantly asking for waveform captures of the ST signal. Why would you want to capture the signal if you were already agreeing with all of us that it is not interlaced? And I am pretty sure that most people here understood exactly as I did.

Actually, it is obvious that Rybags was thinking that you did were talking about the ST signal being interlaced. All I did is to explain to him that you are not claiming that (anymore).

Anyway, it doesn't make much sense to debate about what you exactly said, or meant in the past. If I misunderstood you, then I did. Talking about "twisting" and "insulting" are completely out of place.
To me more precise, I don't care about your opinion about me. Your technical postings are here welcome...
Really? Let's see what you said before:
ppera wrote:Finally (I'm really bored with this all) I don't care about your opinion. If you want not to accept what I say, and still will claim that even and odd lines go on same pos on TVs from later 80-es. it is your problem.
Seems to me you are clearly saying you don't care about my technical opinion. Seems clear to me you are saying it is my problem if I don't technically agree with you. Seems clear to me that if I don't accept what you say, then it is my problem. Or am I twisting your words again?
In case of posted interlaced pic example you remained silent. So, please present your opinion about it.
I was going to comment on that picture before, and even considered doing that now. I even considered asking Rybags to post pictures with his A8 software, both with and without interlacing. But it would be worthless, and honestly, I'm not interested in fights. I abandoned the old thread at some point for the very same reason. You seems to take things too personally. Sorry.
ppera

Re: V an H Blanking interupts - timings @ 50hz

Post by ppera »

There is no any fighting. There is just case that people has different experiences.
In any case, I'm waiting for your (what you intended, but not posted) comments about photo of moving mouse displaying on TV.
Even if it is personal :D
Rybags
Retro freak
Retro freak
Posts: 15
Joined: Sat Jan 31, 2009 2:57 pm

Re: V an H Blanking interupts - timings @ 50hz

Post by Rybags »

My hypothesis (of that pic).

Your TV is giving you interlace for free, whether you want it or not, whether your computer is outputting an interlaced signal, or not.

My theory: put another old computer on the same TV. Atari 800, Commodore 64, Plus/4, VIC-20. Atari 2600 running Space Invaders for that matter.
I reckon there's a good chance you'd get the exact same effect. Although the difference among all machines (I believe) is that the ST is the only one that is claimed to output a 313 line rather than 312 line signal in PAL mode.
ppera

Re: V an H Blanking interupts - timings @ 50hz

Post by ppera »

It is not hypothesis. It is what I talk. That output depends from displaying device too.
Rybags
Retro freak
Retro freak
Posts: 15
Joined: Sat Jan 31, 2009 2:57 pm

Re: V an H Blanking interupts - timings @ 50hz

Post by Rybags »

That's what we've been trying to say all along.

It's not the ST generating an interlaced signal, it's just that the TV isn't smart enough to know it's the defacto-standard type non-interlaced signal that practically all older computers generate.
Dio
Captain Atari
Captain Atari
Posts: 451
Joined: Thu Feb 28, 2008 3:51 pm

Re: V an H Blanking interupts - timings @ 50hz

Post by Dio »

I've done some research on this topic myself. I think I can confirm what a lot of the people here have said.

As far as I can see, if the computer generates a constant number of lines for every 'field', there really are no fields. There are just 50Hz non-interlaced frames. I believe that most CRTs did not generate an interlace offset in this case.

I actually did the experiment mentioned above many years ago on the ST: I took a 640x400 image, downscaled it to 320x400 and split it into two 320x200 images with alternate lines. I then swapped between the two at 50Hz, assuming this would let me see a higher-res image using the interlace. I also added the ability to switch the order of the two frames on the spacebar, in order to align them on the correct fields. It didn't work. Both ways round, the image was identical. That was on a standard TV. I didn't understand why at the time, but in my recent emulation work I worked out a theory.

My theory as to why interlace is suppressed is that on analogue TVs the flyback distance is actually averaged over a group of frames. So if you have alternate 313 and 312 frames, the flyback distance is 312.5 lines. You therefore start at line 0, display 313 lines and fly back to line 0.5, where you generate 312 lines and flyback to line 0 and start again. If the signal received is even on all frames, you don't get this effect and there is no interlace. This also explains why TVs tolerate the various choices of 312 and 313 lines on different computers and consoles, and probably why in the very early days you had to fiddle with the vertical hold to get a ZX81 to work.

(You could perhaps prove this experimentally by generating 312 / 312 / 313 lines and seeing if this enabled three-frame interlacing?)

It is quite clear that many digital TV's don't understand this protocol. I have two desktop monitors with comp in, but only one will talk to STs, Jaguars, Amigas, Megadrives, N64s and the like - the other just displays a blank screen, unless you feed it a console with an interlaced output such as a PS2. And my main Samsung TV, while it will get the signal, is clearly performing deinterlacing on a signal that isn't interlaced given the visual artifacts, and always has a 1-pixel X offset on what it thinks is the 'even' field. Ugly as hell. (The effect is the same with all the different computer systems I've tested).

And as previously said, I certainly wouldn't rely on video capture cards as solid evidence of what is going on with retro computer video. It is clear that many professional TV people are not aware of this case. If someone still has an old AG CRT TV (Trinitron or similar) then you might be able to take a macro picture with 1/25s exposure and see a different interlace pattern traced from an ST compared to a live signal. Unfortunately by the time I'd realised this was worth testing I didn't have a proper CRT any more :) .
ppera

Re: V an H Blanking interupts - timings @ 50hz

Post by ppera »

Dio wrote:...
My theory as to why interlace is suppressed is that on analogue TVs the flyback distance is actually averaged over a group of frames. So if you have alternate 313 and 312 frames, the flyback distance is 312.5 lines. You therefore start at line 0, display 313 lines and fly back to line 0.5, where you generate 312 lines and flyback to line 0 and start again. If the signal received is even on all frames, you don't get this effect and there is no interlace. This also explains why TVs tolerate the various choices of 312 and 313 lines on different computers and consoles, and probably why in the very early days you had to fiddle with the vertical hold to get a ZX81 to work.....
And as previously said, I certainly wouldn't rely on video capture cards as solid evidence of what is going on with retro computer video. It is clear that many professional TV people are not aware of this case. If someone still has an old AG CRT TV (Trinitron or similar) then you might be able to take a macro picture with 1/25s exposure and see a different interlace pattern traced from an ST compared to a live signal. Unfortunately by the time I'd realised this was worth testing I didn't have a proper CRT any more :) .
I don't know about computer which generates mentioned 313/312 lines video, or something similar. It would be probably bad on most of TVs. There is no averaging by regular TV standards. Fact is that whole interlaced TV system is established very long time ago, in era of electronic tubes. So, aim was to maximally simplify TVs, regarding high prices of components.
To achieve interlaced pic on simple CRTs , odd total count of lines was a must, and sync pulse in middle of frame must happen at half line. As I checked on one old B/W TV, there is a total overlap of even/odd lines with Atari ST.

I posted here some screens made with capture card as example of how interlaced pic. looks, when is really interlaced. It was not in purpose to proof anything about ST's video signal. We already concluded that Atari ST has 2x313 lines in 1 frame. It is proven with different testings.

I don't get what you mean under "different interlace pattern traced from an ST compared to a live signal" .
Here was couple proposal about how generating interlaced content withST (I say 'content' with intention, as video timing can not be changed. Except maybe using external sync. , for what ST(E) has some support, but I never tried it.)
Btw. I never had problems with ZX81 - it had stable sync. on TVs on which I tried. Obviously, syncro circuits are pretty different in diverse TVs.
Dio
Captain Atari
Captain Atari
Posts: 451
Joined: Thu Feb 28, 2008 3:51 pm

Re: V an H Blanking interupts - timings @ 50hz

Post by Dio »

What I meant with the interlace pattern trace is that in a real interlaced signal you might be able to see the fields shift by a half-line, while on a non-interlaced signal you might not see this and so the scanlines would be more visible. I suggest an aperture grille TV because it would be more visible there due to the uninterrupted vertical phosphor stripes (no horizontal shadow mask).

I really don't think that TV's were all built to a standard set in stone. In terms of what was legal and what wasn't the computer manufacturers were digging in the dark. Note the VCS documentation, for example, which states that the VCS' display 'works on the wide range of TVs we have tested'.

You were lucky with your ZX81 :) . I went through three TVs before I found one that was reliable :) .
To achieve interlaced pic on simple CRTs , odd total count of lines was a must, and sync pulse in middle of frame must happen at half line. As I checked on one old B/W TV, there is a total overlap of even/odd lines with Atari ST.
I presume by this you're talking about 'total lines' being 2 fields making up one frame: like I said, I reckon that if you only generate a constant line count per 'field' there really are no fields, just 50Hz frames.

I take your point about the half-line though - certainly, that's what's in the Jaguar documentation. So that probably explains the 312.5 flyback, rather than my 'averaging'.
Rybags
Retro freak
Retro freak
Posts: 15
Joined: Sat Jan 31, 2009 2:57 pm

Re: V an H Blanking interupts - timings @ 50hz

Post by Rybags »

In my experience playing around with the Commodore Plus/4...

it's TED chip is very versatile in that you can just tell it what horizontal position or even vertical position it "thinks" it's on within a screen.

So, you can generate oversize displays, early or late H and V-Sync, etc.

The important thing I found was to have a half-scanline as the last on the display every second frame. The TV didn't seem to care about the next frame, and happily displayed interlace.

I even tried bringing forward the half-scanline such that it occurred around 10 scanlines early - the TV still accepted it and displayed interlace. But of course, doing something radical like that would likely have mixed results dependant on how tolerant the TV/monitor is.


To reiterate what I said earlier - I think that with some cleverly timed register changes, the ST just might be able to fudge a signal in similar fashion, and as such, give a proper looking interlaced display on most TVs.
ppera

Re: V an H Blanking interupts - timings @ 50hz

Post by ppera »

Rybags wrote:...
To reiterate what I said earlier - I think that with some cleverly timed register changes, the ST just might be able to fudge a signal in similar fashion, and as such, give a proper looking interlaced display on most TVs.
I doub it. Very likely not possible, and video syncing is hardcoded in HW - Glue chip which generates pulses. It is possible to make overscan, even by pure SW, but it's made just by changing timing of DE signal in MFP.
With external sync. likely possible, on STE. Don't know if someone used it ever...
Gunstick
Captain Atari
Captain Atari
Posts: 294
Joined: Thu Jun 20, 2002 6:49 pm
Location: Luxembourg
Contact:

Re: V an H Blanking interupts - timings @ 50hz

Post by Gunstick »

you would need to go to 60 or 70hz for quite some time so that the screen counter runs out after half of the line
But still I don't know if this would trigger the HBL at thr right point and if the VBL will be correct.
unionapa
Atarian
Atarian
Posts: 1
Joined: Fri Dec 27, 2019 4:16 am

Re: V an H Blanking interupts - timings @ 50hz

Post by unionapa »

Just over a decade too late to join the discussion... As someone that extensively used the horizontal and vertical blanking interrupts in demo coding I'm pretty positive that the ST used with the 13-pin color monitor connection at 50Hz did 313 lines progressive scan at 50 frames per second (obviously). We coded our demos to draw new frames at 50Hz and oftentimes updating the memory just ahead of the shifter reading it (if I remember correctly). As far as I know, both Atari monitors (SM124 and SC1224) did 70Hz progressive scan on monochrome and 50Hz or 60Hz on color monitor (approximately!). The vertical blanking interrupt came 50, 60 or 70 times per second depending on mode. There was NO alternate frame aberrations when updating at the selected frame rate. Later ST models that included a modulator and composite video output may possibly have altered the signal to mimic PAL/NTSC interlaced as televisions of the era expected an interlaced signal and were unable to display progressive. All TV as far as I know in the 80s was interlaced, progressive scan only existed on computer monitors. The Atari monitors treated all input as progressive scan. Any TV would have treated any input as interlaced because any and all TV stations broadcast in that format. TV knew no such thing as progressive scan. Using the ST via composite video or RF modulator would correctly have been interpreted by the display as being interlaced video despite the fact that it really was not. Presumably, the resulting picture would have been somewhere between horrible and more horrible (ironically for the more expensive composite displays as they would have had the resolution to actually display odd and even frames). In many ways, Atari did everybody a disservice by ever supporting composite video and RF modulation. Thankfully it was the dying breath of composite video and the usage of TVs of that era for computer video. I found it bizarre that early HD TV standards even included interlaced video. And apparently I was not alone as 1080i video died almost immediately. I think 90s CRT displays could still do interlaced but I can't for the life of me recall anybody actually using it. But I digress. The ST used progressive scan on the monitor bus. All demos assumed progressive scan. Those that actually connected the ST to a television should have never moved on from the C64.
TomH
Atari User
Atari User
Posts: 40
Joined: Fri Mar 08, 2019 3:47 pm

Re: V an H Blanking interupts - timings @ 50hz

Post by TomH »

A classic analogue TV actually doesn’t know anything about either a progressive or an interlaced image. Other than to discriminate one type of sync from the other and communicate them down the right path it has no link between horizontal and vertical state whatsoever.

The horizontal offset is one sawtooth wave, the vertical offset is another. Both are the result of phase locked loops acting on the actual input, often simple logical flywheels.

Because both are independent, scans are slightly diagonal, not truly horizontal.

A side effect is that if you trigger vertical sync every 312.5 lines instead of an integral number, in one field it’ll fall at the end of a line, in the next it’ll fall in the middle, in the next it’ll be at the end again, etc.

Vertical retrace takes the same amount of time regardless, so all you’re changing is which part of the first line after retrace ends is at the top of the display. It actually doesn’t matter how long retrace takes, but pretend for a second it takes an exact number of lines. In that case if you triggered sync at the end of a line then the beam will start again at the top left of the display. If you triggered it half way through then it’ll start in the top centre.

Since lines are diagonal, starting in the top centre is exactly the same thing as starting half a line higher up. So, that’s interlacing for you.

You could just as easily go the other way and have a three-field interlace, a four-field interlace, etc. The TV wouldn’t see any issues. It just wouldn’t look good.

It’s also untrue that the TV did any kind of deinterlacing — that two fields were weaves together. Real phosphors are lit maybe 10% of the time. The entire display isn’t even lit during a single field, let alone lit long enough still to be emitting when the next field painting comes.

If you have a screen where neighbouring scans don’t quite meet, it’s because the manufacturers were worried about focus and over painting between each pair of lines in the same field. It’s not “leaving gaps” for the next field or anything like that. Indeed, some TVs don’t leave gaps at all.
ijor
Hardware Guru
Hardware Guru
Posts: 4012
Joined: Sat May 29, 2004 7:52 pm
Contact:

Re: V an H Blanking interupts - timings @ 50hz

Post by ijor »

unionapa wrote:Just over a decade too late to join the discussion...
This thread is not only very old, it is also obsolete. Years ago Pera (the one that it was claiming it was interlaced) agreed that the ST video output is progressive and not interlaced. We also now know exactly how the chipset generates the sync signals.

Something quite curious though, is that the chipset does have logic to generate interlace, but it doesn't work. See this other thread for more details: http://www.atari-forum.com/viewtopic.php?f=15&t=30303
Fx Cast: Atari St cycle accurate fpga core
Post Reply

Return to “Coding”