Requesting Video Of Tv Out Cables With Crt Tv

[WizardStan]

Notaz has added DSS2 features to latest kernel so I'm trying it out. There are still some issues with the setup and there is some slow down displaying simultaneously to LCD and S-Video output (Note: Why would he say display to (external, as you claim) LCD and S-Video (leading to external LCD, as you claim)? It makes no sense.. It's clearly about internal LCD and TV-Out running out of sync on the Pandora). This is just the first crack at it and the code can only be improved from here.

He doesn't mention the LCD in question a single time. Its (has been?) a software / driver issue with DSS2.
The video is from early 09, DSS2 hasn't been in the angstrom kernel for too long neither and has been the root of many problems for me.

I never spoke about hardware issues neither - It (was?) a problem with the software side of the Pandora though.


I know that these TVs have to upscale the image internally - but they do it extremly fast, the overhead we see here is caused by doing something similar in the Pandora probably which was not optimized (back then?).

No. The original problem was that a cloned output between LCD and TV out caused a slow down. The fixes he's talking about here are to improve the speed when performing video out, and has nothing to do with the sync issues we saw in the more recent video. The reason he doesn't mention the LCD is because it was irrelevant: the slow down was the important thing. I don't even remember if the original TV out video we saw was to a CRT or LCD TV, nor if it exhibited the sync issues in this more recent video, just the slow down when switching to TV out.
Simply put, you misunderstood and do not know what you are talking about.

 

[kaprikawn]

It's a very common problem and much more likely to be the source of the lag than all the kernal-related stuff that you're quoting in a contrived effort to sound clever.

When an actual dev says here is a problem in the kernel, it's probably true. Telling someone who is already writing fairly complicated software for a system that you, someone with no actual experience at all, knows better than they do, is pretty damn rude. There are (or at least were) some software issues with the TV-out.

Don't be a condecending prick in a contrived effort to sound clever. It has the exact opposite effect.

I hadn't realised he was a dev, otherwise I would have taken a different tone. Doesn't automatically make him right though. In fact he's said something on this thread that is incorrect, he said that plugging it into a CRT wouldn't make a difference, when in fact it would, there's no scaling lag on a CRT.

There's a saying amongst doctors (I've seen it on TV so I know it's true) that if you hear hoofs, assume it's a horse, not a zebra, until you can prove otherwise. It means you should look for the most obvious explaination. As far as I'm aware, we've seen one TV-out demo (the Earthworm Jim one) and it displayed noticeable lag. The 'horses' explaination is that it's lag caused by the TV, it's the most likely scenario and the reason the OP is asking to see TV-out on a CRT to rule that out. The 'zebra' explaination is that it's something kernal/software related, possible, but much less likely. (Or it could be both of course but let's not get into that).

Personally, I'm happy to wait until release to find out what the situation is, I don't want to bother MWeston or anything and I don't think people should be posting requests to the Pandora team (personal opinion). So I don't think the OP should have made this thread, but it is the right idea, the first step in trying to find out what the lag was caused by is seeing TV-out on a CRT, that would be the most useful course of action in determining the problem.

And as a side note, I don't appreciate being called a prick. I think some people have agreed with some of your mod decisions (rightly so on the whole) but that's started to go to your head and now you're starting to be a bit rash with your posting style. Take that as you will, I'm just voicing my opinion because I think you need to take a look at how you conduct yourself considering you're a mod. But nothing was ever solved by arguing on the internet so I'll leave it at that, listen, don't listen, that's up to you entirely.

 

[MWeston]

I don't have a CRT based TV to test with or I would have done this for people (and myself) a long time ago. I dropped it off for recycling about a year ago to the day as it was taking up a lot of space doing nothing.

One thing I have noted in the past is that my laptop's S-Video output has the same lag as Pandora which leads me to believe all the delay is in the digital processing circuitry of my TV. From what I have read, my TV is actually quite good since the lag is only a frame or two. I watch movies from my laptop and have never felt it was distracting. You can see in that Youtube video linked earlier that the intro for Burn Notice plays quite nicely and the voice looks well synced up to me, even with the slight lag we know about.

 

[transmition]

I don't have a CRT based TV to test with or I would have done this for people (and myself) a long time ago. I dropped it off for recycling about a year ago to the day as it was taking up a lot of space doing nothing.

One thing I have noted in the past is that my laptop's S-Video output has the same lag as Pandora which leads me to believe all the delay is in the digital processing circuitry of my TV. From what I have read, my TV is actually quite good since the lag is only a frame or two. I watch movies from my laptop and have never felt it was distracting. You can see in that Youtube video linked earlier that the intro for Burn Notice plays quite nicely and the voice looks well synced up to me, even with the slight lag we know about.

If you jump to around 1:00 in the video, it appears that there is some lag between the audio and the video regarding the gunshot.

Maybe go over to a friend/neighbors house and use their CRT?

 

[DaMummy]

but internet friends are thousands of miles away......how do we resolve this mystery?

 

[Chip]

And as a side note, I don't appreciate being called a prick.

Then don't act like one. The tone of your post was way out of line, especially considering that Fox's post was largely correct. Like most of the devs, he's doing this in his spare time for free. None of them deserve to be told off as if they're idiots, because they're not.

 

[transmition]

but internet friends are thousands of miles away......how do we resolve this mystery?

Is this why they don't have girlfriends / other people do their video recordings for them? If they don't like playing through the viewfinder...

 

[Exophase]

Can anyone give an estimate to what they think the latency is? If it's only ~16-33ms like MWeston states then I don't think people would be so easily noticing it, especially since Youtube is only 30fps. I think for this to be the kind of problem everyone's making it out to be it'd have to be hundreds of ms, and I think that a TV that generates that in a way that there isn't a workaround for would be widely panned as being unusable for video games.

 

[kaprikawn]

And as a side note, I don't appreciate being called a prick.

Then don't act like one. The tone of your post was way out of line, especially considering that Fox's post was largely correct. Like most of the devs, he's doing this in his spare time for free. None of them deserve to be told off as if they're idiots, because they're not.

[Sigh]

I appreciate everything the devs are doing, it's all sterling work and I've supported them vocally on a number of occasions. This does not mean that everything they post on any given subject is correct information and that they are above reproach. I may not have a triple-digit post count and a shiny GP32X-certified avatar thingy, but that doesn't necessarily make me wrong if I disagree with a dev.

My post was out of line in tone, I shouldn't have put 'Fail' at the beginning of it and I accept how inappropriate that was. But I don't think it was way out of line [emphasis yours]. As out-of-line posts go I think mine was pretty mild.

"especially considering Fox's post was largely correct".

No, it wasn't. I stand by what I've said. I may not know a lot about kernal issues, but I know lots about TV scaling issues so I know that it's by far the most likely culprit. Factually speaking, I think I'm right and Fox is wrong and I'm arguing my point.

 

[Exophase]

No, it wasn't. I stand by what I've said. I may not know a lot about kernal issues, but I know lots about TV scaling issues so I know that it's by far the most likely culprit. Factually speaking, I think I'm right and Fox is wrong and I'm arguing my point.

Why do you think that only one of you can be right? The TV could be introducing a significant amount of latency, but that doesn't mean that it's the sole contributor. Latency is an additive effect. When considering total latency in a system you should always look at every link in the chain, not just expect one to be completely dominating it.

At any rate, JayFoxRox is probably not wrong, since he's shown that this is a documented issue in the DSS2 drivers.

 

[Polossatik]

Not going into the who's who's...

Just want to provide a little link for the people who are confused about all this "tv lag/delay" talk and want to read up a bit.

Any non-"standard tv resolution" tv (=720p,1080p... tv's) has this "lag", some more then others. Some sync the audio with the lag to mask the issue.
bottom line, it hurts "most" in older / cheap LCD tv's (slower cpu's) aldo an new expensive TV is not always a warranty for better results.
It only hurts when using upscaling and "enhancing" stuff, if you hook up a PC and use the TV as a display (VGA/DVI) it has no (noticable) "lag" (or at least that's not common).

Edit: this pure the tv side. I know nothing about the pandora side and yes it also apply to older "CRT HD" tv's.

 

[Asmo]

it also apply to older "CRT HD" tv's.

CRT HD TVs? That must be a rare item. A CRT based display should have the least 'lag' of any display being directly fed an analogue signal.

 

[Kramy]

My post was out of line in tone, I shouldn't have put 'Fail' at the beginning of it and I accept how inappropriate that was. But I don't think it was way out of line [emphasis yours]. As out-of-line posts go I think mine was pretty mild.

Fail.

You were correct, and now you're arguing with (and apologizing to) a mod that has told you you're a prick repeatedly.

I think you've won this argument. No need to feed the trolls.

The delay on my TV is approximately 160ms. This is approximately 1/6th of a second. At 24fps I suppose that's 4 frames.

This delay doesn't look that bad. I'm guessing that's all TV delay, and it looks to be under 160ms.

 

[WizardStan]

Can anyone give an estimate to what they think the latency is? If it's only ~16-33ms like MWeston states then I don't think people would be so easily noticing it, especially since Youtube is only 30fps. I think for this to be the kind of problem everyone's making it out to be it'd have to be hundreds of ms, and I think that a TV that generates that in a way that there isn't a workaround for would be widely panned as being unusable for video games.

Looks to be about 1 frame latency, so 33 ms seems to be about right.

 

[fischju2000]

Why, again, is it even a problem if some (or all even) TVs hooked to the Pandora lag behind a couple of frames? AFAIK the Pandora outputs the same image on the screen and through TV-Out. If video and sound from the composite are synced (even if they weren't, but that would be more steps) how difficult would it be to make the controls or LCD lag a couple of milliseconds in software so they matched up? Is it not, in the end, a non-issue? Even if you wanted multiplayer with one person using the Pandora controls/screen and another using a separate controller and a TV, wouldn't introducing artificial lag remedy the situation?

 

[Exophase]

It looks like more than 33ms to me. The fade out on the Pandora is nearly black while the TV is quite red (at 1:11); the granularity of the fadeout would be 60fps and I doubt it's taking so few frames for the whole thing. I think it's more like 2 or 3 frames of the Youtube video. But it'd be a lot easier to tell if I could actually step through it. What would be easiest if someone would run a frame counter like the one linked here. I think @60fps would be the most useful refresh, assuming Pandora can sync to that now.

 

[dflemstr]

It looks like more than 33ms to me. The fade out on the Pandora is nearly black while the TV is quite red (at 1:11); the granularity of the fadeout would be 60fps and I doubt it's taking so few frames for the whole thing. I think it's more like 2 or 3 frames of the Youtube video. But it'd be a lot easier to tell if I could actually step through it. What would be easiest if someone would run a frame counter like the one linked here. I think @60fps would be the most useful refresh, assuming Pandora can sync to that now.

I count 3.8 frames of latency in Avidemux (I count 3 frames between a fade-to-black on the Pandora and fade-to-black on the screen, and the screen is only at about 80% of the Pandora's darkness on the 3rd frame), and the video runs at 30 fps (using the H264 YouTube transcoded version to measure this).

Ergo, the latency is about 127 milliseconds, which is well above the 80 milliseconds of maximum latency a human can casually see without noticing it.

EDIT: updated figures with newly acquired information.

 

[gruso]

Doesn't Youtube have a capped FPS though? I'm out of my depth here to be honest, but I'm wondering if that 30FPS figure is something you want to be basing calculations on.

[edit]

CRT HD TVs? That must be a rareitem.

Reasonably rare, but they're around. Up until a couple of years ago you could buy a widescreen CRT with component in, that would do 720p. I'm hoping my next TV is one of these (second hand of course)!

 

[dflemstr]

Doesn't Youtube have a capped FPS though? I'm out of my depth here to be honest, but I'm wondering if that 30FPS figure is something you want to be basing calculations on.

But ask yourself this: does it make a difference how many FPS the video is running at? True, with 60 FPS, we would get a more accurate result, but if I measure 3.8 frames at 30FPS, the equivalent has to be *at least* 7 frames at 60 FPS. Which puts us at a latency of 117 milliseconds. 'Nuff said.

 

[Kramy]

Do not argue with dflemstr. He has facts which he wields like hammers. :lol:

I personally can tell the difference between Triple Buffering on and Off. (which at 60fps is a sub-50ms difference)

I'm sure many people will notice it, but there's not much that can be done if it's entirely the TV. (which I suspect it is, because consoles lag horribly too)

Conclusion: Worry more once we have actual Pandoras we can try on actual TVs. For now there's no solving it.