OSSC Pro

[bobrocks95]

So the OSSC Pro HDMI output is 1.4, right? Even tho, I guess there could be a VRR add-on that takes advantage of this expansion port you are including in the OSSC Pro, am I wrong?

As far as I know (based on various information scattered on the net), VRR basically works by varying the size of the vertical blanking interval.

Yes, and that actually makes it detrimental to use VRR mode for sources that scan out at slow fixed rate, especially if the upper limit of the selected adaptive sync range is much above the source refresh rate. Enabling VRR also typically disables any low-persistence options offered by monitor as these are not truly compatible with each other. The only positive thing VRR could offer for classic consoles is the ability to handle refresh rate changes seemlessly while still holding framelock. Increased compatibility might be another thing, but well-designed monitors should display the range of VRR frequencies fine even in non-VRR mode.

If it's the type of feature where you can just pass a flag for the display to recognize it as a VRR source that would be really nice, but it's sounding like it's more involved than that I'm guessing?

[NJRoadfan]

Interesting choice of ADC. The ISL51002 looks like an updated version of the ISL98001 that is on my Epiphan DVI2PCIe card. It adds an additional input AND has built in color space conversion. The good thing is the chip is tolerant of oddball output like the NES and SNES. Adding automatic sampling clock/phase would be a nice feature (aka, something similar to the "auto" button on LCD monitors for VGA inputs)

Regarding outputs, its becoming common for designs to output DisplayPort internally and convert to the desired output (ex: Parade PS186 for HDMI 2.0b).

[Josh128]

In any case, if theres some technical reason as to why this mode would induce blur on a CRT ( I dont believe it does with MISTer or Retroarch), the tradeoff would be worth it to me.

Doubling the refresh rate induces both motion incoherency (2 strobes/scans per frame) and latency (half a frame at minimum) so it's not without tradeoffs. You can do motion interpolation to mitigate the former (remember the 100Hz CRT TVs) but that bumps the lag even more. That said, 240p 120+Hz output should be doable on the Pro model.

I'd love to see it in person, I have several VGA monitors in my possession. Ive heard the new Retroarch is capable of the exact same thing with certain video cards that can still output analog video, if anyone is familiar with it please point me in the right direction, as it would be a good indicator of what to expect from this OSSC Pro.

I was able to get the Retroarch 240p120 output working last night using a ~2013 era i5 laptop with an AMD GPU and a 17" Compaq VGA CRT and the results are amazing. Granted, Retroarch and the PC were very finicky about config options and files, and I was having trouble getting some games to center integer scale perfectly in the 320x240 frame (note the uneven pixels in the Bonk shot), but I was finally able to get Supergrafx GnG working with 100% perfect horizontal and vertical resolution and 60fps over 120Hz refresh. As I suspected, the motion was absolutely smooth with no visible judder or blurring whatsoever once I got the proper sync down in Retroarch. There was no need to insert a black frame every other frame (RA has the option) The colors, uniformity, clarity, contrast, and flicker-free image are simply amazing-- and this is from a throwaway monitor that was sitting in my non climate-controlled shed for the last 5 years or so. This is what I was hoping to see and why I said this type of functionality could render the hunt for rare and expensive monitors like BVM's pointless if it could be properly implemented. The TVL of even a run of the mill SVGA CRT like this one is ridiculous, easily on par with a BVM. These monitors can still be had for next to nothing, though as time goes on they too will become increasingly hard to find.

While I used a cellphone to snap some photos and they are not the greatest, the results speak for themselves. Marqs- PLEASE, PLEASE, PLEASE take a serious look at implementing this functionality into the new OSSC Pro, if its possible as you have said. I would even be willing to contribute funds to the development of this feature for the unit if that would help to get a proper implementation.

[Extrems]

I'm calling bullshit if you aren't seeing the effects of double-strobing.

[Josh128]

I'm calling bullshit if you aren't seeing the effects of double-strobing.

You can call it whatever you want, what I witnessed was flawless. What exactly am I supposed to be looking for? It looks just as smooth and stable to me as if it would be coming from a Dreamcast VGA in 480p 60, except there is actually LESS (ie ZERO) brightness strobing due to the 120Hz refresh. Literally zero visible anomalies compared to that. Its a perfect multiple of the frame rate, so why would you expect any judder or uneven frames?

I took some video with a 60fps Canon SL1 camera, though I dont know if that would even be capable of showing this supposed "double strobing". I'll upload it later if anyone is interested.

[bobrocks95]

Should be seeing something like this I believe (2nd from bottom of course)

[orange808]

I took some video with a 60fps Canon SL1 camera, though I dont know if that would even be capable of showing this supposed "double strobing". I'll upload it later if anyone is interested.

You'll probably want to mount a camera using a motorized rail to capture persistence blur accurately. You may also consider a new camera with global shutter, higher frame rates, and more robust features (exposure adjustment).

[fernan1234]

You can call it whatever you want, what I witnessed was flawless. What exactly am I supposed to be looking for? It looks just as smooth and stable to me as if it would be coming from a Dreamcast VGA in 480p 60, except there is actually LESS (ie ZERO) brightness strobing due to the 120Hz refresh. Literally zero visible anomalies compared to that. Its a perfect multiple of the frame rate, so why would you expect any judder or uneven frames?

It's there, but you can consider yourself happy and lucky if somehow you don't see it. Sometimes not having acute eyes is a blessing. Do you also not see duplicate images in moving objects/camera panning of 30fps video at 60hz?

[Nrg]

I'm curious on the gritty details of the ISL51002 vs ADV7842 situation as well, especially considering the ISL51002 + ADV7611 + ADV7513 is a 3-chip solution, but the ADV7842 + ADV7511 is a 2-chip. The ADV7842 + ADV7511 also lets you do things like a built-in HDMI receiver with a 10/12-bit color pipeline at 225 MHz, 2D/3D comb filtering, frame TBC, CVBS/YC input support, CVBS monitor output, etc.

Also it's worth remembering that the ADV video decoder ICs don't support custom resolutions when sampling the Composite/Svideo input.. so one can't get pixel-perfect results with those ICs (with retro consoles/computers), as they always sample at fixed number of samples per scanline based on the PAL/NTSC standard, so for example at 720 pixels/scanline for PAL (or 768 pixels/scanline).

In fact I haven't found a single video decoder IC that would allow configuring the number of pixels per scanline for composite/svideo sources. (for analog RGB sources the number of samples per scanline obviously is configurable in most ICs, just like in OSSC).

Does someone know of a pure *analog* method/circuit to decode/convert composite/svideo to component?

[marqs]

Does someone know of a pure *analog* method/circuit to decode/convert composite/svideo to component?

I have a test board built around TDA8376 (extracted from an old CRT) but haven't yet had time to play around with it properly.

[Josh128]

You can call it whatever you want, what I witnessed was flawless. What exactly am I supposed to be looking for? It looks just as smooth and stable to me as if it would be coming from a Dreamcast VGA in 480p 60, except there is actually LESS (ie ZERO) brightness strobing due to the 120Hz refresh. Literally zero visible anomalies compared to that. Its a perfect multiple of the frame rate, so why would you expect any judder or uneven frames?

It's there, but you can consider yourself happy and lucky if somehow you don't see it. Sometimes not having acute eyes is a blessing. Do you also not see duplicate images in moving objects/camera panning of 30fps video at 60hz?

My eyes are fine for every eye test Ive ever taken though. Im 40 and have never needed glasses (yet).

Sure I can see the effect at 30fps. Ive been gaming since 1984 and every TV I ever gamed on was 60Hz NTSC. In Mario 64, using the semi-first person view and panning around is a very easy way to see it. In Streets of Rage, call the car and its very easy to see in the 30fps animation of that game. 30fps is a much less smooth update than 60fps, and much easier to discern each individual frame than 60. I can see individual frames at 60 on a 60hz screen as well, its just much less apparent because its twice as fast.

Maybe it depends on the content, but Im telling you guys theres no apparent difference in flicker, judder, or blur in this GnG on the 120Hz monitor vs GnG on my 60Hz Sony Trinitron. If there is indeed a measureable effect, its much more subtle than say, DSE on an LCD panel or contrast between an LCD and an OLED. Its absolutely negligible in comparison.

**EDIT-- It is indeed present, but my man its very subtle. Its almost unnoticeable and you need to really be looking for it on fast moving content to be able to discern it. Its far from a deal breaker, IMO.

[energizerfellow‌]

Does someone know of a pure *analog* method/circuit to decode/convert composite/svideo to component?

What's the chipset used in the upcoming Koryuu?

[ldeveraux]

I took some video with a 60fps Canon SL1 camera, though I dont know if that would even be capable of showing this supposed "double strobing". I'll upload it later if anyone is interested.

You'll probably want to mount a camera using a motorized rail to capture persistence blur accurately. You may also consider a new camera with global shutter, higher frame rates, and more robust features (exposure adjustment).

You expect him to go out and buy a new camera and rail system because people on Shmups don't believe him?

[Xyga]

eh? not sure understand Josh, you have set RA to output to 240@120p, does that necessarily mean the emulated source rate is force-doubled to that?

[Xer Xian]

I think the question here is what's wrong with line-doubling 240p and adding full-intensity fake scanlines? The OSSC does this very well and you'd be hard pressed to tell it apart from true 240p on a multisync CRT. And this is with none of the lag and the double-image effect that result from frame doubling.

[rama]

I don't think a regular CRT computer monitor like this will show any of the discussed problems with frame doubling.
It might be a problem with modern displays, but a CRT interpolates it away, which is part of its good image for retro stuff.

[Josh128]

eh? not sure understand Josh, you have set RA to output to 240@120p, does that necessarily mean the emulated source rate is force-doubled to that?

What happens is, the content is still sent to the display at the source animation rate of 60fps, but each frame is sent twice to the VGA port and is displayed twice by the monitor. This is necessary to get true 240p on a 31+KHz multi-scan CRT.

I think the question here is what's wrong with line-doubling 240p and adding full-intensity fake scanlines? The OSSC does this very well and you'd be hard pressed to tell it apart from true 240p on a multisync CRT. And this is with none of the lag and the double-image effect that result from frame doubling.

I would counter that by saying that an SVGA CRT monitor is superior in almost every way to even the best LCD monitors when it comes to contrast, motion blur, viewing angles and aspect ratio. The only thing that comes close to it for contrast, motion blur, and viewing angles is plasma and OLED (minus the motion blur).

Also, Ive seen very few implementations of full-intensity fake scanlines that look as good as what Im seeing here. Theres just the right amount of bloom and brightness in this true 240p image with its authentic scanlines that just looks amazing.

As far as lag, Marqs said he believes this function could be implemented without motion interpolation (basically what Im testing here with RA) for only about 8ms of additional lag. Lets be real, thats basically nothing-- 5 years ago people were excited about the 16ms lag of the Framemeister and that doesnt even count the additional 8 to 16ms of additional lag that only the best LCD sets at the time could muster.

I dont mean or want to highjack this thread with a debate about this, so for additional testing of this mode with RA I should probably just create a new thread about it if the mods agree. In the end, its all about whether this type of scandoubling can be achieved with similar results on the new OSSC Pro, so it is definitely relevant here.

Ive taken some 120fps slow motion videos with an iPhone and will be upping the full speed, and slow-mo ones. They dont show anything. That said, I carefully studied Arthur throwing a lance in real-time on the CRT, and I can indeed perceive the effect in question, but man its really not bad at all-- I didnt even notice it yesterday, but it is there. IMO it is not even perceptible except under very fast moving objects in the right light/dark areas on the monitor. If motion interpolation could be implemented to fix it at the cost of an additional frame as an option, fine-- but being totally honest its very, very subtle and almost unnoticable until you really look for it as is, if given the option to pick either or, that would be great-- but if I had to choose one or the other? I'd choose the raw uninterpolated motion, ESPECIALLY if it saved 16ms or more of lag.

I don't think a regular CRT computer monitor like this will show any of the discussed problems with frame doubling.
It might be a problem with modern displays, but a CRT interpolates it away, which is part of its good image for retro stuff.

Its present, as I now see, but man its so very subtle. Its more apparent on some content than others, but its not bad at all.

[NormalFish]

what visual difference would there be between "real" and "fake" scanlines in a 480p frame?

**Welp, Fudoh has me covered below

[Kez]

I would counter that by saying that an SVGA CRT monitor is superior in almost every way to even the best LCD monitors when it comes to contrast, motion blur, viewing angles and aspect ratio. The only thing that comes close to it for contrast, motion blur, and viewing angles is plasma and OLED (minus the motion blur).

Xer's suggestion does work on a CRT. Line double the OSSC, set scanlines to 100%. What you end up with is essentially a 31khz 240p image (with timing differences, but no refresh rate/framerate mismatch). Every other line is blank.

[Fudoh]

I think the question here is what's wrong with line-doubling 240p and adding full-intensity fake scanlines? The OSSC does this very well and you'd be hard pressed to tell it apart from true 240p on a multisync CRT.

it's not this good. When you double content to 480p, the CRT will still add in it's own scanlines betwen each source line of pixels, so the scanlines you see end up being more intense than what you get by feeding 240p120 to the screen. By feeding 480p with 100% scanlines you basically get "active line - natural dark line - fake dark pixel line - natural dark line - active line", so the ratio between active and non-active lines is a bit off compared to a source signal without fake scanlines.

My last experiments with 120Hz on a CRT have been quite some time ago. I had 480p120 running on FW900 fed by a DVDO and I also don't recall any negative effects caused by the double strobing of the frames, but this has been 480p60 content back then and I don't think I tried with classic 2D stuff.

[Josh128]

I would counter that by saying that an SVGA CRT monitor is superior in almost every way to even the best LCD monitors when it comes to contrast, motion blur, viewing angles and aspect ratio. The only thing that comes close to it for contrast, motion blur, and viewing angles is plasma and OLED (minus the motion blur).

Xer's suggestion does work on a CRT. Line double the OSSC, set scanlines to 100%. What you end up with is essentially a 31khz 240p image (with timing differences, but no refresh rate/framerate mismatch). Every other line is blank.

Well, you are correct here-- I hadnt considered that. It might be the next-best solution (it has its own trade-offs as Fudoh described above) if the 240p 120 scan doubling functionality doesnt pan out.

[fernan1234]

Regardless of disadvantages, OSSC Pro support for 240p@120Hz would be a fantastic addition to its already rich set of features. Most definitely on CRTs, but also on upcoming flat panels that will have BFI options at both 120 and 60 refresh rates (hopefully, unless the manufacturers mess things up as usual and butcher customization), which would resolve motion issues at the expense of some brightness loss, but the option would be there.

[Josh128]

My last experiments with 120Hz on a CRT have been quite some time ago. I had 480p120 running on FW900 fed by a DVDO and I also don't recall any negative effects caused by the double strobing of the frames, but this has been 480p60 content back then and I don't think I tried with classic 2D stuff.

You are correct in your observations about the double strobing, IMO. Its very subtle, unnoticeable except on very fast moving objects-- and even then you really have to be looking for it.

I've upped two short YT videos of the Supergrafx GnG demo loop taken by an iPod Touch 7th gen in 120fps slow-mo mode. One of them is at the full 120fps speed (sorry, Youtube reduced it to 60 fps), and the other one is produced and output by the iMovie app (which defaults to 30fps, hence the YT 720p30 rate on this one), but that is how it is normally done-- all the frames that were captured are displayed--hence this video is 3x the length of the full speed video.

Full Speed - https://www.youtube.com/watch?v=_efcKe6TMQE

Slow-Mo - https://www.youtube.com/watch?v=0x_Q9MeEKG8

[orange808]

Ghouls and Ghosts scrolls so slow, it looks crystal clear on an OLED or a DLP. Not really a good test pattern.

As for the scanlines, all VGA monitor scanlines are "too smooth". They don't the same sawtooth appearance you get from a 15kHz consumer television or professional monitor.

Boot up Super Mario Bros and run. That tiled rock ground should be absolutely crystal clear at absolutely at all times. It isn't when I feed a double strobe. There's also the issue of keeping the double strobe in sync with the "console".

It's like stretching 4:3 to 16:9 or using scaling filters. It's a matter of taste. Make no mistake, though, the motion resolution is decreased and lots of people see the blur.

[orange808]

My last experiments with 120Hz on a CRT have been quite some time ago. I had 480p120 running on FW900 fed by a DVDO and I also don't recall any negative effects caused by the double strobing of the frames, but this has been 480p60 content back then and I don't think I tried with classic 2D stuff.

What model of DVDO did you use? I'd like to try it out with less latency than my other programmable video processors.

I've only attempted this with the Corio2, Barco ImagePro II, and editing modelines in emulators.

[energizerfellow‌]

I'm curious on the gritty details of the ISL51002 vs ADV7842 situation as well, especially considering the ISL51002 + ADV7611 + ADV7513 is a 3-chip solution, but the ADV7842 + ADV7511 is a 2-chip. The ADV7842 + ADV7511 also lets you do things like a built-in HDMI receiver with a 10/12-bit color pipeline at 225 MHz, 2D/3D comb filtering, frame TBC, CVBS/YC input support, CVBS monitor output, etc.

Also it's worth remembering that the ADV video decoder ICs don't support custom resolutions when sampling the Composite/Svideo input.. so one can't get pixel-perfect results with those ICs (with retro consoles/computers), as they always sample at fixed number of samples per scanline based on the PAL/NTSC standard, so for example at 720 pixels/scanline for PAL (or 768 pixels/scanline).

In fact I haven't found a single video decoder IC that would allow configuring the number of pixels per scanline for composite/svideo sources. (for analog RGB sources the number of samples per scanline obviously is configurable in most ICs, just like in OSSC).

Does someone know of a pure *analog* method/circuit to decode/convert composite/svideo to component?

There's still some analog composite/S-video video encoders still kicking around, e.g. Rohm BH7236AF, but analog decoders and transcoders like the TDA3562A seem truly dead now.

From digging around for decoders it looks like digital decoders from Analog Devices, TI, AKM, and Rohm/LAPIS are pretty much all that's left? They all look to be putting out digital ITU-R BT.656 with fixed number of samples per line?

Speaking of the Rohm BH7236AF chip, it has RGBs pass-through in addition to a CVBS/YC encoder, so in theory you could make a box that's SCART pass-through for capture, but has composite/S-video for a monitor. Also an advantage as there's lots of people out there with CRTs that don't have component inputs.

[Josh128]

Ghouls and Ghosts scrolls so slow, it looks crystal clear on an OLED or a DLP. Not really a good test pattern.

As for the scanlines, all VGA monitor scanlines are "too smooth". They don't the same sawtooth appearance you get from a 15kHz consumer television or professional monitor.

I completely disagree with you on the look of a VGA monitor my man. This thing looks terrifyingly good and extremely similar to what is shown of photos of BVMs Ive seen on the net. For instance, take this BVM photo below I found from a quick search.

If you dont like the look of a BVM, sure, you most likely wont like this. The horizontal lines are definitely not as course as say, a Sony Trinitron 27" NTSC TV, but then even a Trinitron, as good as it looks (I have one and yes its great), has nothing on the quality of this image in my opinion, when you consider clarity, convergence, and geometry.

Full size BVM : https://imgur.com/TOVKbAk


The high TVL of the BVM above looks extremely similar to what Im seeing here on my set below.

Full size SVGA: https://imgur.com/mG4ggTG

Full size SVGA: https://imgur.com/onzoLfI

Boot up Super Mario Bros and run. That tiled rock ground should be absolutely crystal clear at absolutely at all times. It isn't when I feed a double strobe. There's also the issue of keeping the double strobe in sync with the "console".

It's like stretching 4:3 to 16:9 or using scaling filters. It's a matter of taste. Make no mistake, though, the motion resolution is decreased and lots of people see the blur.

The tiled rock does not blur whatsoever. There is, however, a perceived effect of a slight double image when moving very fast, but it is merely an optical illusion, as it is not captured by a high speed camera.
As far as keeping the double strobe in sync with the console, that should be extemely easy to accomplish as it will be exactly double the rate of the sync signal generated by the console. The OSSC Pro could theoretically just double the incoming sync frequency, split it to H & V sync, and the monitor will simply lock onto that signal.

Motion resolution is not decreased whatsoever as evidenced by the 120fps slo-mo video I posted earlier. The "perception" of double images on the fast moving content is simply an illusion due to the inability of the human optical nerves, receptors, and brain to be able to properly discern the individual frames. In any case-- this whole issue can be alleviated by inserting a black or dark frame for every displayed game frame. This also should be easy to accomplish on the hardware. RA supports it and I tried it and it looks fantastic-- just a tad darker than without and the "strobing" effect we are discussing is completely vanquished.

[zarkFR]

Hi guys,

Sorry if it sounds dumb, but could the OSSC Pro have, in its upscaler mode, an Antialiasing algorithm for 480p (and maybe 720p?), something similar to the mClassic from Marseille ?

(Despite what people say, AA from mClassic is quite good (I have one). Maybe not 100$ worth, but still a very noticeable improvement - and yes the gamma change is shit)

Also, third picture shows only one 3.5 audio jack. Could this connector output passthrough audio from scart (if scart is selected as input) or component (if comp is selected as input) ?

(sorry for bad english !)


Z.

[Xer Xian]

The double-image effect of repeated frames on a CRT cannot be alleviated with BFI. The fade to black is the exact reason why the effect is present on CRTs (and why it is absent on sample&hold display). If you feed 240p120 to a modern display and activate 120hz BFI you will substitute one kind of vision artifact (image persistence) with another (double-image effect).

I think the question here is what's wrong with line-doubling 240p and adding full-intensity fake scanlines? The OSSC does this very well and you'd be hard pressed to tell it apart from true 240p on a multisync CRT.

it's not this good. When you double content to 480p, the CRT will still add in it's own scanlines betwen each source line of pixels, so the scanlines you see end up being more intense than what you get by feeding 240p120 to the screen. By feeding 480p with 100% scanlines you basically get "active line - natural dark line - fake dark pixel line - natural dark line - active line", so the ratio between active and non-active lines is a bit off compared to a source signal without fake scanlines.

Yes, but what makes you think that the ratio will not be equally off when displaying 240p120? Now I haven't tested this, but what I would expect is that the natural 240p blanklines in 240p120 will be just as thick as the combination of the simulated 240p blanklines + thin 480p blanklines. The width of the scanlines on any given CRT (and for any given focus setting) should stay the same regardless of the resolution being displayed after all.

Of course, a CRT displaying 240p120 will look brighter than when displaying 480p60 with 100% scanlines - but that's due to the increased refresh rate and the fixed decay time of the phosphors.

[Bassa-Bassa]

As for the scanlines, all VGA monitor scanlines are "too smooth". They don't the same sawtooth appearance you get from a 15kHz consumer television or professional monitor.

The high TVL of the BVM above looks extremely similar to what Im seeing here on my set below.

The pics you just posted precisely confirms orange808 statement, though. There are differences, and the VGA lines look smoother. But if someone prefers hi-end BVM's picture vs. a good arcade/TV set's, I don't see many reasons not to prefer the VGA monitor's, tbh. Size matters, anyway. At around 17'', this may work. Go over 20'', and it's totally overkill for these graphics.