Has anyone Genlocked 2 Ataris together?

[Rybags]

With all this talk on video enhancements lately, just wondering... has anyone tried combining video from 2 A8 machines together?

 

I'd imagine it should be reasonably simple - one possibility would be to take the chroma from one the luma from another.

 

Sync of course would be the issue - I think you could probably use the first machine to assert RESET on the second, then release it such that the timing of the two Antics was cycle-exact.

 

It would be of limited use for games as extensive rewrites would be needed to enhance existing stuff, but might be an interesting project for graphical demos.

 

APAC and CIN without "blinds" or flicker.

Normal Graphics 15 stuff in 4 colours * 4 luminances = 16 colours.

"Stereo" Pokey without addons.

 

Or, maybe even use luma from both - have the luma from the second machine "detuned" such that it fills in the gaps of normal 0 to 2 values, which could potentially give *shitloads* of possibilities.

[JayoK]

Sync of course would be the issue - I think you could probably use the first machine to assert RESET on the second, then release it such that the timing of the two Antics was cycle-exact.

 

Sounds like a good idea, but would it actually be possible to guaratneed that the Antic were in perfect sync? For example, could one Atari run at say 1.780000MHz (as an example) and another at 1.78001000 through natural deviations within the system? Would this 1Khz make a difference? Possibly not.

[Rybags]

That's a good point - even 1 cycle drift per second would see the display become rather unusable fairly quickly.

 

I'd still like to try it though - problem is I have only 1 Atari PSU so I'll have to bodge an adaptor to run the other off the PC.

[roland p]

Sync of course would be the issue - I think you could probably use the first machine to assert RESET on the second, then release it such that the timing of the two Antics was cycle-exact.

 

Sounds like a good idea, but would it actually be possible to guaratneed that the Antic were in perfect sync? For example, could one Atari run at say 1.780000MHz (as an example) and another at 1.78001000 through natural deviations within the system? Would this 1Khz make a difference? Possibly not.

 

I guess you could fix that by generating one clock signal and feed it to the two ataris.

[Rybags]

You could. But then it gets into the realm of heavily modifying the machine for "little" benefit.

 

Mixing 2 machine's video would be more of a novelty thing than something greatly useful, although in theory you could do games and serious apps which took advantage of it but I doubt the demand would be that great.

[MEtalGuy66]

Hmm. forgive me if Im worng, but based on my experinece with gen-locks, THATS WHAT IT DOES... It syncs the two displays perfectly. Ive genlocked multiple AMIGAs together many times and never had any problems with "scan-shifting", etc... In fact most available GENLOCK devices were made for the professional video (or post production video editing) industry, and any amount of "display shift" would make them pretty useless as far as producing consistant professional quality results..

 

Not to say there hasnt been sold some crap devices in the "home video" category that could be worse, performance-wise...

 

What Ive often thought of doing is using the chroma output from one macine, and the luma output from another, to the same monitor... In this case, your considerations about "clock/scan sync" would be VERY relevant... And yeah, it's be interesting to see 2 atari's clocked off of 1 source..

 

I wonder if you could just remove the oscilators from both machines, and hook a signal generator to both at the same time..

Edited August 27, 2008 by MEtalGuy66

[+CharlieChaplin]

With all this talk on video enhancements lately, just wondering... has anyone tried combining video from 2 A8 machines together?

 

I'd imagine it should be reasonably simple - one possibility would be to take the chroma from one the luma from another.

 

Sync of course would be the issue - I think you could probably use the first machine to assert RESET on the second, then release it such that the timing of the two Antics was cycle-exact.

 

It would be of limited use for games as extensive rewrites would be needed to enhance existing stuff, but might be an interesting project for graphical demos.

 

APAC and CIN without "blinds" or flicker.

Normal Graphics 15 stuff in 4 colours * 4 luminances = 16 colours.

"Stereo" Pokey without addons.

 

Or, maybe even use luma from both - have the luma from the second machine "detuned" such that it fills in the gaps of normal 0 to 2 values, which could potentially give *shitloads* of possibilities.

 

 

Well,

don`t know anything about mixing two machines video... but mixing two machines sounds is easy. There are already two demos available that support this: 1) Perestroyka Demo and 2) Sky Network. Each demo has (afaik) 5 sounds to listen to. You have to press 1-5 on computer 1 and SHIFT 1-5 on computer 2 (simultanously!) to hear the sounds in stereo... of course therefore the computers should be connected to two tv`s/monitors or to a hifi-system...

 

-Andreas Koch.

[Mathy]

Hello guys

 

Would it be possible to create an interlaced picture by having one computer do the first half and the second computer doing the second half of the picture?

 

Greetings

 

Mathy

[Rybags]

I think the Amiga Genlock would work by allowing for drift, ie - it probably brings it's VSync forward or back slightly if the source isn't matched to it's rate.

 

Interlaced picture from 2 Ataris?

 

Doing that would require external hardware to alter the frame from each machine slightly so it matches the timing of interlaced ones, ie - the VSync gets skewed by half a scanline.

 

No point really the way I see it as you could achieve the same effect by just modifying the output from one machine (something I was also interested in trying).

My idea there was to have an external box which could override or suppress the sync signal level - and the Atari would do it via software control.

 

I might give the Genlock idea a go later. Simply using colour from one machine and luma from another, without any extra intervention.

 

It would be a trial/error thing to get the displays matching HSync, but you should be able to tell fairly quickly just what oscillator drift (if any) there is.

[MEtalGuy66]

I think the Amiga Genlock would work by allowing for drift, ie - it probably brings it's VSync forward or back slightly if the source isn't matched to it's rate.

 

Interlaced picture from 2 Ataris?

 

Doing that would require external hardware to alter the frame from each machine slightly so it matches the timing of interlaced ones, ie - the VSync gets skewed by half a scanline.

 

No point really the way I see it as you could achieve the same effect by just modifying the output from one machine (something I was also interested in trying).

My idea there was to have an external box which could override or suppress the sync signal level - and the Atari would do it via software control.

 

I might give the Genlock idea a go later. Simply using colour from one machine and luma from another, without any extra intervention.

 

It would be a trial/error thing to get the displays matching HSync, but you should be able to tell fairly quickly just what oscillator drift (if any) there is.

 

They have (and have had for ages) completely external Gen-lock hardware that can take two completely separate and discrete NTSC signals, and sync them into one rock solid NTSC output. It doesnt matter what the sources of these signals are.. In fact, external devices like this even pre-date the "expansion devices" made for the AMIGA..

[ijor]

Genlock is, by definition, to have all the sources synchronized. This could be done either with an external sync/clock generator that feeds all the sources, or with one master feeding the slave(s).

 

This doesn't mean you can't "combine" multiple video sources that aren't originally synced. And this sometimes is also called "genlock", but strictly it isn't.

 

If the two or more sources are not related, such as an independant display of two computers, then combining them is not too difficult. You just buffer one frame for each source, and combine them when they are all available. Eventually you need to discard frames from the slower sources, and you might need or want then to apply motion compensation, but still not too difficult with current technology.

 

However, if both sources must be in sync because they are related somehow in time, then without a true genlock, things get much more complicated. It is still doable as long as you have enough memory. The longer the display, the more the sources become apart in time, and the more frames you need to buffer. Because you need to "remember" the display from the faster sources until getting the corresponding frames from the slower ones.

[+Allan]

http://www.a8maestro.com/atari/info/studio/studio.htm

 

Allan

[DamageX]

They have (and have had for ages) completely external Gen-lock hardware that can take two completely separate and discrete NTSC signals, and sync them into one rock solid NTSC output.

I believe it's called a time base corrector.

[thorfdbg]

Hmm. forgive me if Im worng, but based on my experinece with gen-locks, THATS WHAT IT DOES... It syncs the two displays perfectly. Ive genlocked multiple AMIGAs together many times and never had any problems with "scan-shifting", etc... In fact most available GENLOCK devices were made for the professional video (or post production video editing) industry, and any amount of "display shift" would make them pretty useless as far as producing consistant professional quality results..

 

Except that the Amiga had the hardware and the connectors to make this easy, i.e. Agnus had the option not only to generate, but also to receive the sync signals from an external source. That's not the case for the Atari.

 

Not to say there hasnt been sold some crap devices in the "home video" category that could be worse, performance-wise...

 

What Ive often thought of doing is using the chroma output from one macine, and the luma output from another, to the same monitor... In this case, your considerations about "clock/scan sync" would be VERY relevant... And yeah, it's be interesting to see 2 atari's clocked off of 1 source..

 

I wonder if you could just remove the oscilators from both machines, and hook a signal generator to both at the same time..

 

That seems plausible and possible, but solves only half of the problem. You also need to ensure that the HSync and the VSync of the two machines appear at exactly the same time. Hopefully, once this is established, both machines will remain in sync as the timing on both machines should be identical - so it *might*

be sufficient to reset both machines with one switch.

 

So long,

Thomas

[Rybags]

Yes, the "exact" H/V sync issue I went over in the fist post.

 

I'd use the master machine to Reset the second. Finding the time to do it would be trial/error but wouldn't take long to work out (probably minutes).

 

The main concern of course would be drift if the 2 main oscillators were in the slightest bit different.

 

I'm pretty certain Antic establishes a new VSync regime when reset - I remember with my XL on the old TV it was easy to get a quick screen roll most times you held/released the Reset key.

[analmux]

Or, what about a totally different approach:

 

With nowadays screen devices it should be easier to mix consecutive frames, like in a pairwise blending.

 

F.e.: Let the atari display colours (in gtia 11) on the odd frames, and display lums (in gtia 9), then mix together and only let the screen device update the screen 25 times per second instead of 50.

 

Shouldn't such a thing be possible by just an external device (with digital buffers to make the mixing work) ??

[Rybags]

My capture card does that - blends frames such that stuff like TIP doesn't flicker, but updates at 25 FPS as a result.

 

No idea about TVs - I still got ye olde LG CRT so can't vouch for the newer stuff.

[+poobah]

I wonder if you could just remove the oscilators from both machines, and hook a signal generator to both at the same time..

 

Now that's interesting... I'd imagine the actual startup sequence timing isn't completely deterministic, You'd probably have some jitter since the electronics aren't going to be exactly the same, but I bet it would be very very minute.

 

{poobah resists urge to borrow sig gen from work....}

[ijor]

I wonder if you could just remove the oscilators from both machines, and hook a signal generator to both at the same time..

 

That seems plausible and possible, but solves only half of the problem. You also need to ensure that the HSync and the VSync of the two machines appear at exactly the same time. Hopefully, once this is established, both machines will remain in sync as the timing on both machines should be identical - so it *might* be sufficient to reset both machines with one switch.

 

Now that's interesting... I'd imagine the actual startup sequence timing isn't completely deterministic,

 

I think poobah is right. Resetting both machines at the same time won't help, because Antic video sync doesn't seem to be affected by reset. Try pressing reset and, depending on your monitor/tv, you would see that sync is still working.

 

So it is not about reset, but about power-up, and Antic power-up timing is likely to differ across machines.

 

That's the reason that systems that support genlock receive a sync signal, not a clock. Some receive a clock as well, other derive the clock from the external sync using a PLL. But you must use a common video sync, not (only) clock.

[Rybags]

I think Antic's VSync generation is affected by Reset though.

 

I can distinctly remember my 800XL being able to cause the screen to roll back at times when you held Reset - of course the 400/800 never really did since System Reset is only an NMI. Although, thinking about it some more, it might be generated by the fact that the 6502 is in a holding pattern and not storing the DList address each frame which can sometimes cause glitches.

 

To test it, I'll have to do some more 192 KHz sound card sampling - it should be able to tell some of the inside story.

[MEtalGuy66]

I think Antic's VSync generation is affected by Reset though.

 

I can distinctly remember my 800XL being able to cause the screen to roll back at times when you held Reset - of course the 400/800 never really did since System Reset is only an NMI. Although, thinking about it some more, it might be generated by the fact that the 6502 is in a holding pattern and not storing the DList address each frame which can sometimes cause glitches.

 

To test it, I'll have to do some more 192 KHz sound card sampling - it should be able to tell some of the inside story.

 

Well, I have a 500mhz 4-channel digital scope, that could capture that quite nicely, but I dont have a signal generator..

 

I can tell you however, that it is amazing how close the clock signals are generated between different machines..

 

Here is some crap Ive been working on to document the "phase 2 clock enigma" that seems to plague various atari expansion configurations..

[a8maestro]

Heh,

 

That is mine. Using an atari for the external video source is something I have not tried.

 

I can if interested. The device does exhibit some jitters though.

 

Rick D.

 

 

 

http://www.a8maestro.com/atari/info/studio/studio.htm

 

Allan

[Rybags]

How does the device decide on "priority" ?

 

Does it just overlay dark areas?

 

Also, it seems to buffer the data and mix later.

 

I'm going to try some feasibility tests a little later - I've got 3 A8 machines - 800XL, 130XE and XEGS.

 

To test how closely synched the clocks are, I'lll try sampling the luma from each machine for about 5 minutes each, and see how accurate the VBlank intervals are over that period - should be sufficient to calculate how much drift exists between each one.

[+poobah]

How does the device decide on "priority" ?

 

Does it just overlay dark areas?

 

Also, it seems to buffer the data and mix later.

 

I'm going to try some feasibility tests a little later - I've got 3 A8 machines - 800XL, 130XE and XEGS.

 

To test how closely synched the clocks are, I'lll try sampling the luma from each machine for about 5 minutes each, and see how accurate the VBlank intervals are over that period - should be sufficient to calculate how much drift exists between each one.

 

should be interesting. even if there is no drift, there is still the sync issue. not much use if 2 machines are always exactly 23 scan lines off.

[Rybags]

Yes, that's why it'll need one machine to Reset the other at a precise time.

 

I've done an initial sampling - weird to say the least. It drifts immediately (130XE).

 

The 10th VBlank interval occurs at around 0.20054 seconds instead of 0.2. By the 5 minute mark, it's at 5.2051. Now to get the 800XL out and see if it has similar characteristics.