Payback 2015
category: general [glöplog]
More pics from the party!
(quantity over quality?)
http://tiny.cc/spk-payback2015
(quantity over quality?)
http://tiny.cc/spk-payback2015
oasiz: so what's the thing with the scaler boxes really, why are they needed, is it just to split the signal to both the big screen and the net stream? It got me thinking, if this amount of hardware and video expertise and equipment cannot display these brand new oldskool demos in full framerate to an audience full of oldskool framerate geeks, then the world is broken.
yzi:
I guess I'll just go with the longer response, bear with me :)
Those were scalers straight from US, I actually didn't get to test them before. Why I had four of them is because I happened to order four of them :)
I did however know quite a lot before ordering those as my friends have already used those while organizing speed running events.
Workflow was analogue, reason for that is being easier to split, easier to control trough matrix, no PnP ruining capture/compomachine/etc.. Let's just say that we had the splitters/matrix already sourced for that purpose.
Splitting HDMI and using HDMI matrixes is always asking for trouble unless you have more expensive gear, and then you're better off with SDI anyway but that's another story.
Truth is that very few scalers handle NTSC lowres "240p" properly, the ones I had were extron DVS304A. Even fewer of those "240p" compatible ones actually work properly with PAL stuff.
PAL scaling is just really tricky to work with, worst case is that the "288p" sometimes doesn't even sync at all in any form.
Although, had I known that it would have been such an issue, I would have used a different chain. This is why Jope brought his Sony scaler as a backup which has been a really reliable scaler so far, albeit with wrong field order issues for some reason this time around (As it interlaces the progressive image first and then tries to de-interlace it, like quite many do), I believe that the extron would have worked with PAL if it was properly interlaced first, sadly I didn't have a pass-trough device with me such as a DVD recorder.
So apologies for the messed up field order, I know it could have gone a lot better.
Unfortunately there is no "turnkey solution" for this kind of stuff so if you want proper 50Hz RGB that actually goes to screen & capture, you're best off using scalers and in some cases it's going to be a chain of devices. For the best quality you might even have to alternate between different devices.
I know that with speed running events we even had to deal with scalers that do a full crash, requiring a forced power cycle if "PAL60" hits it. State of proper video processing for older stuff is really bad, XRGB probably comes closest to hitting the mark but even the framemeister has issues when switching between lace-progressive as the delay can be huge. Things like this are a lot easier in NTSC-land.
There are efforts on making some simple linedoublers that would support PAL modes as well but usually those projects come and go as people lose interest and never want to publicly sell/share their design in the end.
But again, I am really bummed because I know that it could have gone a lot better.
So, scalers, all Extron DVS304A that I managed to get cheap and just picked from the postal office on thursday, good and reliable broadcast scalers but mostly NTSC-friendly.
Scaler #1 - Scaling screen/slide machine to capture, with capture & streaming going on you can't really afford sudden screenmode changes or you might still get remnants of the flickering buffer. In worst case, the driver can lock. This is something that can't be afforded so it's best to just abstract everything. This is actually the only scaler I originally had set up in that chain.
Scalers #2 and #3?
#2 - Slide machine, Pre-scaled for projector and also it acted as a tiny amplifier in the middle, mostly just amplification.
#3 - I don't know how this is even possible but straight output from the compo PC to the video matrix would result in complete signal loss, areas that are darker would go black and lose sync where brighter areas work better. But routing it trough a cheap chinese VGA splitter as a pass-trough would make it work magically. Problem with this is that chinese crap is crap so I just knocked up another scaler as a simple RGB passtrough. Fixed it and it looked much better than that power-leaky splitter.
I guess I'll just go with the longer response, bear with me :)
Those were scalers straight from US, I actually didn't get to test them before. Why I had four of them is because I happened to order four of them :)
I did however know quite a lot before ordering those as my friends have already used those while organizing speed running events.
Workflow was analogue, reason for that is being easier to split, easier to control trough matrix, no PnP ruining capture/compomachine/etc.. Let's just say that we had the splitters/matrix already sourced for that purpose.
Splitting HDMI and using HDMI matrixes is always asking for trouble unless you have more expensive gear, and then you're better off with SDI anyway but that's another story.
Truth is that very few scalers handle NTSC lowres "240p" properly, the ones I had were extron DVS304A. Even fewer of those "240p" compatible ones actually work properly with PAL stuff.
PAL scaling is just really tricky to work with, worst case is that the "288p" sometimes doesn't even sync at all in any form.
Although, had I known that it would have been such an issue, I would have used a different chain. This is why Jope brought his Sony scaler as a backup which has been a really reliable scaler so far, albeit with wrong field order issues for some reason this time around (As it interlaces the progressive image first and then tries to de-interlace it, like quite many do), I believe that the extron would have worked with PAL if it was properly interlaced first, sadly I didn't have a pass-trough device with me such as a DVD recorder.
So apologies for the messed up field order, I know it could have gone a lot better.
Unfortunately there is no "turnkey solution" for this kind of stuff so if you want proper 50Hz RGB that actually goes to screen & capture, you're best off using scalers and in some cases it's going to be a chain of devices. For the best quality you might even have to alternate between different devices.
I know that with speed running events we even had to deal with scalers that do a full crash, requiring a forced power cycle if "PAL60" hits it. State of proper video processing for older stuff is really bad, XRGB probably comes closest to hitting the mark but even the framemeister has issues when switching between lace-progressive as the delay can be huge. Things like this are a lot easier in NTSC-land.
There are efforts on making some simple linedoublers that would support PAL modes as well but usually those projects come and go as people lose interest and never want to publicly sell/share their design in the end.
But again, I am really bummed because I know that it could have gone a lot better.
So, scalers, all Extron DVS304A that I managed to get cheap and just picked from the postal office on thursday, good and reliable broadcast scalers but mostly NTSC-friendly.
Scaler #1 - Scaling screen/slide machine to capture, with capture & streaming going on you can't really afford sudden screenmode changes or you might still get remnants of the flickering buffer. In worst case, the driver can lock. This is something that can't be afforded so it's best to just abstract everything. This is actually the only scaler I originally had set up in that chain.
Scalers #2 and #3?
#2 - Slide machine, Pre-scaled for projector and also it acted as a tiny amplifier in the middle, mostly just amplification.
#3 - I don't know how this is even possible but straight output from the compo PC to the video matrix would result in complete signal loss, areas that are darker would go black and lose sync where brighter areas work better. But routing it trough a cheap chinese VGA splitter as a pass-trough would make it work magically. Problem with this is that chinese crap is crap so I just knocked up another scaler as a simple RGB passtrough. Fixed it and it looked much better than that power-leaky splitter.
we hope to get a micomsoft xcapture-1 for nordlicht capturing and streaming purposes as I heard it can do at least as well as the framemeister (coming from the same company).
yeah: setting up a proper video chain with the least signal loss/stutter and hassle/points of failure while keeping all the fancy like capturing/upscaling is somehow not that easy...
i think payback did a very good job at it almost all the time.
yeah: setting up a proper video chain with the least signal loss/stutter and hassle/points of failure while keeping all the fancy like capturing/upscaling is somehow not that easy...
i think payback did a very good job at it almost all the time.
Xcapture is good, I think it has the "single resolution output" feature on the drivers where it acts as an "el cheapo" scaler. At least the SC-512 does.
I use the SC-512 for capturing and streaming which is mostly the same deal.
You will still want something to deal with linedouble vs. interlaced modes however, this prevents it being a simple dumb input.
Capturing is nice but streaming is always a challenge.
Amarec has some options to set it up automatically but it's a pain to deal with.
However one good thing about it is that it can output the post-processed result (linedouble, de-interlace, etc..) straight in to it's own video output so you can sort of try to abstract what is being done for streaming software.
My advice is to use scalers if you plan on streaming AND using the projector. Also if you are using RGB: Finding powered splitters for RGB is a huge pain in the ass (there are 1001 scart splitters out there but all are passive).
If you have some questions then feel free to catch me on IRCnet/Freenode.
Also, I am not sure if Xcapture suffered from it but at least the SC-512 has issues with tiny color component bleeding when using lowres modes, 288p especially. In Amarec this can largely be fixed by using "Type-C" color fix. It's not a biggie but something you might want to get rid of when capturing.
I use the SC-512 for capturing and streaming which is mostly the same deal.
You will still want something to deal with linedouble vs. interlaced modes however, this prevents it being a simple dumb input.
Capturing is nice but streaming is always a challenge.
Amarec has some options to set it up automatically but it's a pain to deal with.
However one good thing about it is that it can output the post-processed result (linedouble, de-interlace, etc..) straight in to it's own video output so you can sort of try to abstract what is being done for streaming software.
My advice is to use scalers if you plan on streaming AND using the projector. Also if you are using RGB: Finding powered splitters for RGB is a huge pain in the ass (there are 1001 scart splitters out there but all are passive).
If you have some questions then feel free to catch me on IRCnet/Freenode.
Also, I am not sure if Xcapture suffered from it but at least the SC-512 has issues with tiny color component bleeding when using lowres modes, 288p especially. In Amarec this can largely be fixed by using "Type-C" color fix. It's not a biggie but something you might want to get rid of when capturing.
Thanks for the long explanation. I'll have to read up on the things you mentioned. I'm interested in this, because music-synced full framerate motion on CRTs has always been the single most important thing in demos for me, and it feels more and more difficult to get any of that anywhere. I'm dreaming of a system where oldskool fans could use something like a RasPi as a playback device connected to a TV, ideally CRT monitor of course, to watch oldskool content streams from e.g. parties, in full framerate so that the picture and motion would look like it's supposed to. The problem is, how to produce the stream to feed it, not to mention doing that in a live situation...
I've had the exact same idea as well.
Although I wouldn't aim for composite straight. UMSA which could even convert a 15KHz signal from a PC to something that the CRT would eat, giving RGB quality.
Pair that with some PAL / NTSC encoder in case you really want to have s-video.
Problems still arise with some systems as commodore stuff can vary somewhere between 49Hz ~ 51Hz at worst, while it might sync on many scalers/capture cards, it might still drop a frames or pad it with dummy frames. This is a huge problem already with stuff like SNES outputting 60.10Hz, making it almost impossible to get a 1:1 signal captured without drops.
If you are interested on discussing about this stuff then feel free to hit me up on IRC some time, I can expand some things in case you want to know more.
Although I wouldn't aim for composite straight. UMSA which could even convert a 15KHz signal from a PC to something that the CRT would eat, giving RGB quality.
Pair that with some PAL / NTSC encoder in case you really want to have s-video.
Problems still arise with some systems as commodore stuff can vary somewhere between 49Hz ~ 51Hz at worst, while it might sync on many scalers/capture cards, it might still drop a frames or pad it with dummy frames. This is a huge problem already with stuff like SNES outputting 60.10Hz, making it almost impossible to get a 1:1 signal captured without drops.
If you are interested on discussing about this stuff then feel free to hit me up on IRC some time, I can expand some things in case you want to know more.
It didn't help at all that the Extron that fed the stream + the projector somehow began to magically fail (my guess is overheating) just before the 8-bit + Amiga compo was about to begin.
I had verified the signal with only the Extron earlier that day and I could get the it to sync to the machines (MSX, Amiga) just fine. Then when we had the next up 8-bit slide up, I connected the first compo machine to the scaler and .. no sync. VITTU. As a result I just plugged in the DSC-1024HD into the Extron, got sync and now we had two scalers in the pipeline, both performing a forced deinterlace -> the field order got fucked up there.
If we hadn't had the video screen + internet stream stuff, I'd have connected the oldskool computers directly to the projector to begin with. For some reason modern projectors usually have no problems syncing to 240p60 and 288p50 and they even upscale very nicely with barely any banding artifacts.. Modern TVs are much more finicky.
One option would also have been to split the signal before the scaler feeding the stream/downstairs video screens, so that at least the main screen would have gotten p50. Unfortunately as Oasiz mentioned, the VGA splitter we had produced a very annoying visual hum to the signal so that was not an option either. Ah well hindsight 50/50, all in a weekend's demo compo mayhem. :-D
I had verified the signal with only the Extron earlier that day and I could get the it to sync to the machines (MSX, Amiga) just fine. Then when we had the next up 8-bit slide up, I connected the first compo machine to the scaler and .. no sync. VITTU. As a result I just plugged in the DSC-1024HD into the Extron, got sync and now we had two scalers in the pipeline, both performing a forced deinterlace -> the field order got fucked up there.
If we hadn't had the video screen + internet stream stuff, I'd have connected the oldskool computers directly to the projector to begin with. For some reason modern projectors usually have no problems syncing to 240p60 and 288p50 and they even upscale very nicely with barely any banding artifacts.. Modern TVs are much more finicky.
One option would also have been to split the signal before the scaler feeding the stream/downstairs video screens, so that at least the main screen would have gotten p50. Unfortunately as Oasiz mentioned, the VGA splitter we had produced a very annoying visual hum to the signal so that was not an option either. Ah well hindsight 50/50, all in a weekend's demo compo mayhem. :-D
Entry caps are uploading as I type this, check out: http://memorex.lerppu.net/payback/
While anyone could DL these, I'd suggest waiting for possible YT uploads as these are meant to be with little loss on image quality and take more space than a typical video would.
WinSCP is telling me that ETA ~35min but we know how these always go.
Don't touch those with .filepart in the end as those are still uploading.
I'll check tomorrow if I missed anything.
While anyone could DL these, I'd suggest waiting for possible YT uploads as these are meant to be with little loss on image quality and take more space than a typical video would.
WinSCP is telling me that ETA ~35min but we know how these always go.
Don't touch those with .filepart in the end as those are still uploading.
I'll check tomorrow if I missed anything.