OSSC (DIY video digitizer & scandoubler)

[marqs]

I've yet to find a single display that works with Sonic 2 splitscreen and OSSC.

I had it syncing correctly on Eizo and Philips PC monitors (didn't write down their models back then). That MD used csync taken from CXA chip and fw was earlier revision but that should not make any difference. Seems like I have to get one MD for in-depth testing.

[BuckoA51]

Okay, I'll try it on a bunch more displays too.

[ikari_01]

I just assembled the kit and goofed around a bit with a SNES. Linedouble mode suffers from aliasing due to the 256 pixel horizontal resolution vs. output resolution but linetriple mode looks absolutely gorgeous. Far less noise than the Framemeister and evenly spaced scanlines (the Framemeister scanlines always look somewhat "paired" to me). I'll continue testing with more consoles and another display and add my findings to the compatibility list.
Thanks for making this! It's already more fun to play around with than the Framemeister

One thing I noticed though is that the device responds very poorly to the IR remote. Often I have to press a button 5-6 times for it to register properly. Any clues? Soldering looks good on the IR receiver.

[Mantrox]

Marqs, if you can, when the first batch of pre-assembled kits are ready to go and you inform the folks on the waiting list, please post here warning that you have done so.
I don't want to skip getting the OSSC because of some crazy spam filter


...you were probably already going to do this anyway...

[BuckoA51]

but linetriple mode looks absolutely gorgeous.

It's stunning isn't it? I'm bogged down with taxes at the moment but after a bit more testing on the weekend I really need to write up about the line triple modes and re-review the near-final firmware.

One thing I noticed though is that the device responds very poorly to the IR remote. Often I have to press a button 5-6 times for it to register properly.

Mine seems to do this sometimes then not others, try holding the remote at a different angle, I've no idea why that works but it seems to.

[marqs]

One thing I noticed though is that the device responds very poorly to the IR remote. Often I have to press a button 5-6 times for it to register properly. Any clues? Soldering looks good on the IR receiver.

The first PCB prototype (a year ago) had unreliable IR most likely due to signal integrity issues. The IR diode on current PCBs is only ~1cm away from the FPGA and should work reliably, but it's still not perfect. I'm getting an ignored keypress (no led blink) every now and then, but not anymore such cases where a button needs to be pressed 5-6 times to get registered. I'd first check the 0805 cap next to it (part of supply RC filter). The IR code detection logic could be also made more robust, e.g. by adding a digital LPF if unreliability is really caused by noise.

Marqs, if you can, when the first batch of pre-assembled kits are ready to go and you inform the folks on the waiting list, please post here warning that you have done so.
I don't want to skip getting the OSSC because of some crazy spam filter


...you were probably already going to do this anyway...

Yeah, I should have written an update to people waiting the first batch already a month ago (original availability estimate) but forgot that among other things. Not that I could have given a hard delivery date / price back then, but still.

Now that I'm finally getting first samples of that batch tomorrow (and rest in 1-2 weeks), things look a lot better. I'll write an update on the weekend, so people waiting the 1st batch should check their spamboxes if nothing shows up by Monday.

I'm bogged down with taxes at the moment but after a bit more testing on the weekend I really need to write up about the line triple modes and re-review the near-final firmware.

I'm not sure if the current firmware can be called near-final, even though it's stable and includes all the basic features. There's still room for improvement and new features, and I hope I can concentrate on adding those once the remaining DIY-kit orders and 1st batch of pre-assembled boards have been shipped. I originally planned to get v1.00 out before that, but preparation of kits+materials and order process for pre-assembled boards have taken far more time than I anticipated.

[ikari_01]

Haha, okay, apparently the IR signal is too strong It doesn't work well when I point the remote directly at the ossc (be it from 1, 10, or 50cm). It works perfectly every time when I put the ossc on my wooden desk, hold the remote UNDER the desk and point it at the wall so the IR light has to bounce off the wall behind the desk, then off the back of the room, then back to the ossc.

[BuckoA51]

I'm not sure if the current firmware can be called near-final, even though it's stable and includes all the basic features. There's still room for improvement and new features, and I hope I can concentrate on adding those once the remaining DIY-kit orders and 1st batch of pre-assembled boards have been shipped. I originally planned to get v1.00 out before that, but preparation of kits+materials and order process for pre-assembled boards have taken far more time than I anticipated.

Perhaps final is a bad choice of words since these things are never really done. However compared to the first firmware I tested the new firmware is very good and I wouldn't hold up production any longer.

[lettuce]

So i should be ok for line tripling with my Sony KDL-42W653?

[cleeg]

A thought just occurred to me, if we line triple a 240p source that then converts to 480i (a la Saturn), what would happen to the picture?

Also, what happens to a 480i source that is line doubled? Do we get 960p? Forgive me if this one has been answered, it's a big thread!

[Guspaz]

Since each field in a 480i signal only has 240 lines, if you double it, you get 480 lines, so 480i doubled is 480p.

Remember, 240p does not actually exist: it is basically just a timing hack of 480i to get the two fields to appear in the same position instead of offset.

[cleeg]

Since each field in a 480i signal only has 240 lines, if you double it, you get 480 lines, so 480i doubled is 480p.

Remember, 240p does not actually exist: it is basically just a timing hack of 480i to get the two fields to appear in the same position instead of offset.

That makes sense, thankyou. It's that hack which creates the scanlines then?

[Guspaz]

Scanlines are the lines of the picture: the black part is the gap between the scanlines.

A CRT display works by tracing a single point across the screen in rows, and the screen glows where the beam hits. Depending on various factors (like how sharply focused the beam is and how high resolution the phosphors on the screen are), you can either have no gap between the scanlines, or you can have big black solid gaps.

A normal standard definition signal is 480i. This means that you've got 60 fields per second, and each field contains 240 lines, and they alternate between "even" and "odd", with every other one slightly offset. It looks sort of like this:

Field1:

111111
000000
111111
000000

Field 2:

000000
111111
000000
111111

And so on. This happens fast enough that at a distance, you probably just see whole images, but close up (or if you're sensitive to it) you can see it as a flicker.

Without getting too technical, the way that the TV knows to draw on different lines is basically by controlling the timing so that when the beam moves back up to the top of the screen, end ends up at a different place each time.

Retro consoles output a 480i signal, but they mess with the timing so that the beam ends up back at the same place each time, so instead, you get this:

Field1:

111111
000000
111111
000000

Field 2:

111111
000000
111111
000000

So it's still a 480i signal, but the display is made to always put the scanlines in the same place. We call this "240p" but it's really just messed up 480i. The reason that the gaps between scanlines are much more noticeable in 240p is because the display never draws the odd scanlines, it's only every drawing the even ones, so there is always those gaps (unless your display is really soft).

This is actually the reason why 240p is so hard on modern televisions. Because 240p is just a hack of 480i, most modern TVs see the signal as 480i, and they try to process it as such, trying to deinterlace the signal (which isn't interlaced) or just refusing to accept the non-standard timings.

480p would just look like this every frame:

111111
111111
111111
111111

[BuckoA51]

So i should be ok for line tripling with my Sony KDL-42W653?

Probably not, but I have encountered some televisions that will accept the linetriple modes, so you might get lucky.

A thought just occurred to me, if we line triple a 240p source that then converts to 480i (a la Saturn), what would happen to the picture?

OSSC falls back to it's 480i processing and does a quick bob deinterlace.

[telemetry]

A thought just occurred to me, if we line triple a 240p source that then converts to 480i (a la Saturn), what would happen to the picture?

OSSC falls back to it's 480i processing and does a quick bob deinterlace.

Correct me if wrong: the bob deinterlace is just taking each field's 240 lines and doubling them to 480.

So there's no contradiction in using line-triple mode with 480i - it would just be taking each field and scaling it to 720p (you'd still have flicking lines).

It's counterintuitive to think of 480 lines scaled to 720, but really it's 240 lines in one frame (whether it's 240p or 480i), and so it seems like you could line-triple 480i as well -- but only with the bob-type deinterlace. This seems like it would be a sharper alternative to pure 480i > 480p field-scaling.

[cleeg]

Thanks to both of you above for all the explanations, it's filled in a few knlowledge blanks.

[meneerbeer]

It's counterintuitive to think of 480 lines scaled to 720, but really it's 240 lines in one frame (whether it's 240p or 480i), and so it seems like you could line-triple 480i as well -- but only with the bob-type deinterlace. This seems like it would be a sharper alternative to pure 480i > 480p field-scaling.

Unfortunately, no, linetripling 480i is not that simple. A 480i field contains 262.5 lines. So you end up with 787.5 lines. Having half a line in your field gives you an interlaced signal, so you end up with a weird (720*2)=1440i like signal.

If you are able to generate a very accurate video clock so a 720p frame is exactly the same length as a 480i field, you would be able to do the bob deinterlacing. Linetripling unfortunately is not an option.

[telemetry]

It's counterintuitive to think of 480 lines scaled to 720, but really it's 240 lines in one frame (whether it's 240p or 480i), and so it seems like you could line-triple 480i as well -- but only with the bob-type deinterlace. This seems like it would be a sharper alternative to pure 480i > 480p field-scaling.

Unfortunately, no, linetripling 480i is not that simple. A 480i field contains 262.5 lines. So you end up with 787.5 lines. Having half a line in your field gives you an interlaced signal, so you end up with a weird (720*2)=1440i like signal.

If you are able to generate a very accurate video clock so a 720p frame is exactly the same length as a 480i field, you would be able to do the bob deinterlacing. Linetripling unfortunately is not an option.

Thanks for the clarification! Shows my ignorance of TV signals.

[Mantrox]

Form replied!
I'm in like Flynn on the first 50

Hope to get it soon, i have a stack of ps1 games i'm itchin' to play.

[Das Muel]

My E-mail landed in my spam folder. Form completed, Marqs.

[Nrg]

Order form completed

[Bahn Yuki]

Form completed, went to my junk folder!

Sent from my SAMSUNG-SM-G870A using Tapatalk

[Mantrox]

For those that have a unit already, can i get my harmony remote to work with the OSSC by adding the profile of a specific NEC product or do i have to go the with the old code programming?

Oh and another thing, having to playing ps1 games on a late model PAL PS2 (SCPH-9000x) as a primary goal at the moment, what should be the best way to setup everything considering i have a DVDO EDGE to add to the chain if needed?

Component YPbPr -> OSSC 480P w/scanlines -> TV/Monitor
Component YPbPr -> OSSC 480P w/scanlines -> DVDO (what resolution do i choose to output? 960P?) -> TV/Monitor
RGB SCART -> OSSC 480P w/scanlines -> TV/Monitor
RGB SCART -> OSSC 480P w/scanlines -> DVDO (what resolution do i choose to output? 960P?) -> TV/Monitor

Im confused, because the PVM detected the signal as 240P when i played a ps1 game, but the console and game are PAL....
Im lost...

[BuckoA51]

RGB SCART -> OSSC 480P w/scanlines -> TV/Monitor

This will give you the least amount of input lag, scaling/picture quality depends on your TV

RGB SCART -> OSSC 480P w/scanlines -> DVDO (what resolution do i choose to output? 960P?) -> TV/Monitor

Use the OSSC in line triple mode. Output the DVDO at 1080p and use the zoom control to adjust if necessary until everything looks even. This will give the nicest picture but adds about 25ms input lag to NTSC games (not sure on PAL games).

[Xyga]

The vp50pro should stay at the usual 6~8ms, can't wait to try the combo with OSSC, the DE/EE will put this on orbit.

[Mantrox]

RGB SCART -> OSSC 480P w/scanlines -> TV/Monitor

This will give you the least amount of input lag, scaling/picture quality depends on your TV

RGB SCART -> OSSC 480P w/scanlines -> DVDO (what resolution do i choose to output? 960P?) -> TV/Monitor

Use the OSSC in line triple mode. Output the DVDO at 1080p and use the zoom control to adjust if necessary until everything looks even. This will give the nicest picture but adds about 25ms input lag to NTSC games (not sure on PAL games).

Thank you for the tips.

I only have component available at the moment. The RGB scart cable i have is atrocious.
Any recommendations on that other than the obvious choice of the retrogamingcables one?

25ms is like a frame and a half of lag? hummm that's probably tolerable.

My current monitor DELL U2715H is 1440P native instead of 1080p and has ~12ms of input lag, but with a non-native signal the lag will probably increase and it will degrade the image as well.
I'll have to wait, try it out and see, there's too many variables.

[BuckoA51]

The vp50pro should stay at the usual 6~8ms, can't wait to try the combo with OSSC, the DE/EE will put this on orbit.

Hopefully I should be able to try that myself soon, potentially it's a match made in heaven, though Fudoh did say he had issues with it (couldn't get it to frame lock) though that was on an earlier OSSC firmware.

25ms is like a frame and a half of lag? hummm that's probably tolerable.

It's about what the Mini does, so yeah not the end of the world especially if your display is low lag too. You could always install a DVI switch to change between direct to the display for really timing critical games like shmups and to the Edge for other games where picture quality is more important than input lag.

My current monitor DELL U2715H is 1440P native instead of 1080p and has ~12ms of input lag, but with a non-native signal the lag will probably increase and it will degrade the image as well.

It's actually been proven using Leo's lag tester that, in most cases, there's no difference between a native and non-native resolution signal in terms of input lag. As long as the input signal is progressive, of course. Obviously what happens to the image quality wise depends entirely on your monitors scaler.

[blizzz]

What's the final price for the pre-assembled boards? The form says "X EUR (excluding VAT)", does that mean we have to add the VAT for Finland to the price? That would be an extra 39€ if I'm not wrong.

[Fudoh]

though Fudoh did say he had issues with it (couldn't get it to frame lock) though that was on an earlier OSSC firmware.

indeed - and line tripling mode only.

[BuckoA51]

Hmm, line triple seems to frame lock here quite nicely on the Edge, I'll double check that later though.