OSSC (DIY video digitizer & scandoubler)

[citrus3000psi]

He could've asked more nicely, but I agree with tusecy, an option to disable LEDs would be nice.

Then add the feature, test it, and make a PR (pull request).

I'll try to be the banana and update my test environment this week and make the changes.

[ZellSF]

Tried the reverse LPF function on my really shitty Super Famicom and it was a noticeable improvement. Not close to the 1CHIP's image quality though. Still will be nice whenever I want to play Super Turrican. Though with a revised Super Turrican rom hopefully being leaked, maybe that one will work properly on 1CHIP SNES consoles (I sort of doubt it).

You're really overestimating my abilities (or being really rude).

Certainly not the latter!

I could maybe do it if I had weeks and a JTAG flasher.

[eric90000]

Is it generally OK to use the OSSC with a 25 feet long HDMI 2.0 cable? If so, It'd save me having to buy a new DVI-HDMI cable. It's an amazon basics cable, it's super thick.

[RocketBelt]

He could've asked more nicely, but I agree with tusecy, an option to disable LEDs would be nice.

Then add the feature, test it, and make a PR (pull request).

I'll try to be the banana and update my test environment this week and make the changes.

I achieved this using a small piece of black electrical tape.
I am the banana and I claim my prize.

[tpp]

Here's the 3 profiles I use for PS1 games...

Terrific - thanks for sharing! I didn't realize that the menu screens were using different timings than the gameplay screens - I had been testing on the menu screens since distortions on the letters are easy to spot. I'm curious about how that's possible if anyone in the know can explain it. I know the PSX GPU is capable of several different modes, but I thought the OSSC settings would be identical for a given mode since the dot clock is fixed per mode. So I figured 256x240 would be 256x240 no matter the game or where you're at in it.

I tested Castlevania a bit, and found the difference between generic 4:3 and optimized 256x240 quite small. I can really only see it by looking closely at the border between two columns of pixels of different colors. In optimized mode, it's a dead straight vertical line. In generic mode, there's a tiny bit of rounding of the active lines (non-scanlines), with the brighter color pixel "rounding out" a bit against the darker one. Not sure to what extent that's a product of the OLED display tech or the 4K upscaling on my set. If anything, I think I slightly prefer the generic mode - it feels like a tiny, welcome nod towards the original CRT tech. I certainly won't fret about playing in generic mode in games like Castlevania that switch timings on menu screens. But perhaps other games will tilt more strongly towards optimized mode.

Thanks again.

[nmalinoski]

Is it generally OK to use the OSSC with a 25 feet long HDMI 2.0 cable? If so, It'd save me having to buy a new DVI-HDMI cable. It's an amazon basics cable, it's super thick.

Looking at Wikipedia's page on HDMI, it should work fine as long as it's certified Premium High Speed (the minimum for HDMI 2.0), like this one; basically, if it passes certification, it should work.

[kel]

Just finished trying out the new Latency tester feature and got some unexpected results.

Sony 32W705b shows up as ~23ms on the bottom right in game mode which seems a bit high for this set. Anything lower than the maximum backlight setting will time out and it also times out for the middle and top left box regardless of the backlight setting.

Samsung 51F8500 Plasma shows up as 41ms on the bottom right in PC and game mode which seems quite low for this set as I have at one point borrowed a Leo bodnar tester to use on this set and it showed up as 65ms and ~75ms respectively for the above modes which also matches certain online reviews. Strangely the top left box shows a reading of 1ms

Panasonic P50VT65B Plasma shows up as 39ms on the bottom right in game mode which is close to the 42ms result that certain online reviews state. This set seems to be the closest match to known readings out of the three.

I'm also a little bit confused how the latency tester feature works compared to the Leo bodnar tester, mainly with regards to the fact that with Plasma screens the Leo bodnar tester will give the same results regardless of where the reading is taken on the screen due Plasma screens updating the whole frame at once but the OSSC latency tester feature gives a different result depending on where the reading is taken.

[tpp]

I didn't realize that the menu screens were using different timings than the gameplay screens...

On this note, is there any easy way to at least see the native resolution of the PSX output? Whether through the OSSC or an emulator? You'd think this information would be available running the game in ePSXe, but I can't seem to find a way to display it. It would be nice to have verification that I'm at least trying the correct 240p mode.

[orange808]

Sony 32W705b shows up as ~23ms on the bottom right in game mode which seems a bit high for this set. Anything lower than the maximum backlight setting will time out and it also times out for the middle and top left box regardless of the backlight setting.

~23ms at the bottom right would be about 6 or 7ms of lag. That's a conspicuously low number.

Samsung 51F8500 Plasma shows up as 41ms on the bottom right in PC and game mode which seems quite low for this set as I have at one point borrowed a Leo bodnar tester to use on this set and it showed up as 65ms and ~75ms respectively for the above modes which also matches certain online reviews. Strangely the top left box shows a reading of 1ms

Interesting results. What's the native resolution of the panel?

[jayde6]

Terrific - thanks for sharing!

I'm always happy to help.
I tried the same thing at first, adjusting stuff at the menu. I remembered hearing that the Saturn version changed stuff up when you went into the menu and thought for kicks I'd try it in the game itself. The weird thing is that the title screen is fine but not the file menu screen. The map, enemy text, item text and dialog text are fine but not the menu. I wonder what specifically they needed to change for the menu as text looks great and legible at whatever the actual game is set to.

Have you checked out the 480i and 480p modes yet? I can't tell any difference between 480ix2 and 480ix4 (can't get 480ix3 to work at all) and the 480px2 mode vs passthrough on my LG tv looks identical.

Was just curious, since you have an LG as well, mines a few years older and not oled, but I can't imagine they changed the scaler too much between 4k tvs.

[Syntax]

Is it generally OK to use the OSSC with a 25 feet long HDMI 2.0 cable? If so, It'd save me having to buy a new DVI-HDMI cable. It's an amazon basics cable, it's super thick.

Looking at Wikipedia's page on HDMI, it should work fine as long as it's certified Premium High Speed (the minimum for HDMI 2.0), like this one; basically, if it passes certification, it should work.

Don't get caught up in the stupid HDMI naming wank.
There's 2 types of HDMI out there.
The original 20 wire that had 4 unused wires which some manufacturers would omit and the new HDMI with all 20 wires used.

The speed thing is total bs.

I make my suppliers guarantee 20 wires via email so I can get a refund if they are 16 or less.

[marqs]

I'm also a little bit confused how the latency tester feature works compared to the Leo bodnar tester, mainly with regards to the fact that with Plasma screens the Leo bodnar tester will give the same results regardless of where the reading is taken on the screen due Plasma screens updating the whole frame at once but the OSSC latency tester feature gives a different result depending on where the reading is taken.

OSSC tester counts the time between first white pixel scanned out and sensor detect while Bodnar tester always starts counting at the top of the frame. It means the numbers are comparable only with top-left results, with OSSC tester showing 0.5 frame lower latency on center position and around 1 frame less on bottom-right. Interesting results - how big resistor you have as Rc? I've only tested with OLED TV and 2 desktop LCD monitors where the results were as expected.

~23ms at the bottom right would be about 6 or 7ms of lag. That's a conspicuously low number.

Nope, you don't need to subtract 16ms like with Bodnar tester. 23ms is 23ms here.

[Xyga]

If orange's not right then I don't get how this works, it doesn't make sense to me that the lag at the bottom end woud be the same as the top's, since 60Hz sample-and-hold displays need 1 frame time to draw from the top to the bottom, the lag's different across the screen that's how it's supposed to work, so what is it I'm missing
I don't get that in the wiki either;

If display panel refresh rate matches input refresh rate, latency is typically uniform across all positions.

[marqs]

it doesn't make sense to me that the lag at the bottom end woud be the same as the top's, since 60Hz sample-and-hold displays need 1 frame time to draw from the top to the bottom, the lag's different across the screen that's how it's supposed to work, so what is it I'm missing

It takes ~16ms for a CRT beam to travel from top-left to bottom-right, so are you implying it also has 16ms display latency? The number gets included in input latency only if the source is vsynced, i.e. does not change content during scanout.

[Galdelico]

I'm always happy to help.
I tried the same thing at first, adjusting stuff at the menu. I remembered hearing that the Saturn version changed stuff up when you went into the menu and thought for kicks I'd try it in the game itself. The weird thing is that the title screen is fine but not the file menu screen. The map, enemy text, item text and dialog text are fine but not the menu. I wonder what specifically they needed to change for the menu as text looks great and legible at whatever the actual game is set to.

As odd as it may sound, plenty of games back then did this. That's because retro consoles - take a look at the Mega Drive, over here at SEGARetro - support multiple screen resolutions, both in progressive and interlaced mode.
So - once again, if we consider the Mega Drive - you can have a game like Sonic The Hedgehog, which runs perfectly fine when you set your OSSC to optim. 320p, and another game such as Virtua Racing that indeed requires an optim. 256p profile, in order for it to get displayed accurately (you may also want to switch from 4:3 to 8:7, via the 256p screen ratio toggle in the OSSC's Output menu, since the graphics will most likely look too 'wide').
That said, you can encounter all kinds of hybrids too: there are games (i.e. Super Street Fighter II - The New Challengers) that boot up with the SEGA logo in 320p, and then switch to 256p, and other ones (see Splatterhouse Part 3) that keep swapping from 320p (boot-up, introduction, in game cutscenes...) to 256p (actual gameplay). For the Mega Drive/Genesis specifically, there's this list made by Firebrandx, which is very helpful to give you an insight on this matter.

Saturn and PlayStation do the same (but I'm not aware of comprehensive lists for those systems, unfortunately): games that run at 320p, 256p, swapping from 320p to 480i... I'm still bouncing from optim. modes and generic ones, mainly because I'd need plenty of OSSCs, in order to have enough profiles available for each system I own.

Have you checked out the 480i and 480p modes yet? I can't tell any difference between 480ix2 and 480ix4 (can't get 480ix3 to work at all) and the 480px2 mode vs passthrough on my LG tv looks identical.

I can tell a difference on my Asus 1080p monitor. 480i Linex4 looks definitely sharper than x2, even though (of course) it retains all the interlacing artifacts you'd also get in line double. Linex3 doesn't give me any picture either.

[paulb_nl]

If orange's not right then I don't get how this works, it doesn't make sense to me that the lag at the bottom end woud be the same as the top's, since 60Hz sample-and-hold displays need 1 frame time to draw from the top to the bottom, the lag's different across the screen that's how it's supposed to work, so what is it I'm missing
I don't get that in the wiki either;

If display panel refresh rate matches input refresh rate, latency is typically uniform across all positions.

He explains it in the first sentence:

OSSC tester counts the time between first white pixel scanned out and sensor detect while Bodnar tester always starts counting at the top of the frame.

I like this method because you can say the display has ..ms lag compared to a CRT.

[Xyga]

@marqs: Well I thought the lag people expect to read is the display's/total latency yes, typically of a 60hz lcd for most users, and in three places at the top, middle and bottom. With the LB we can check that and read the top's or substract 8 or 16 ms to know the input latency of course.
What I understand orange was saying is that if it's 23ms at the bottom then the - input - lag at the top should be of 6~7ms, I'm sorry from what you said it seemed to me you said it's not the case and so it confused me completely about what it all implies and what the OSSC's lag test figure actually shows if it's not the same as the LB tester.
I have yet to try it myself (waiting for components) maybe it will become cleared after I've experienced it.

@paulb_nl: isn't that the same thing though giving more accurate readings maybe?

PS/ ok one precision: I do count the duration of 1 frame draw as delay or simply time before completion.
If there's a difference on what people count as time or not, well I don't know what tosay, apparently there's two schools and for some only the input delay counts as real delay.

[jayde6]

I'm still bouncing from optim. modes and generic ones, mainly because I'd need plenty of OSSCs, in order to have enough profiles available for each system I own. .

I hear ya there haha.
Thanks for the great info! I encountered needing separate 320 and 256 profiles for Genesis already but figured it was just an out of the norm game, I had no idea it was something that affected so many titles. Looking at the list I realize now that on multiple occasions I've seen the flickering pixels on the sega logo or the title screen of a genesis game, thought I was in the wrong profile, and toggled the same profile while it was transitioning, thinking I corrected the issue.

That list is definitely going to come in handy.

[kel]

OSSC tester counts the time between first white pixel scanned out and sensor detect while Bodnar tester always starts counting at the top of the frame. It means the numbers are comparable only with top-left results, with OSSC tester showing 0.5 frame lower latency on center position and around 1 frame less on bottom-right. Interesting results - how big resistor you have as Rc? I've only tested with OLED TV and 2 desktop LCD monitors where the results were as expected.

Thanks for clearing that up, I understand now how the latency tester works compared to the Leo bodnar tester. I just couldn't work it out due to the strange readings that I seem to be getting. I kinda prefer it this way as I think that this method is much fairer to plasma TVs. The Leo bodnar started the counter at the top when the frame is not yet drawn but plasma draws the frame at once which I thought was kind of unfair due to the fact that it is comparing the start of a frame to a fully drawn frame.

I have used a 150k ohm resistor as RC as recommended on the Wiki page. maybe I'm doing something wrong, I will have another try today now that I understand what I should expect.

[paulb_nl]

@paulb_nl: isn't that the same thing though giving more accurate readings maybe?

If you have a LCD display with 10ms of lag compared to CRT then the Bodnar tester will report 10, 18, 26ms at top, middle, bottom. The OSSC will report 10, 10 ,10 ms.

With a plasma that refreshes the whole screen at once with 10ms lag compared to CRT, Bodnar will report 26, 26, 26. The OSSC will report 26, 18, 10.

For me the OSSC reports the true input lag.

[Xyga]

Okay thanks for clearing that out. I'm part of the crowd that thinks the half-life of the frame, or if you want the location-time around the middle of the screen where the eye can first perceive the new frame appearing over the old frame, is the 'extreme point' where it is logical to measure lag, since it is also where most of the action of video games takes place and therefore the eye focuses on.
I've been used to LB and its three readings and the reason behind those (not sure the majority of the people interested in lag readings got the meaning of the reason behind the three bars but heh) so the OSSC's testing tool completely confused me here, this is pretty simple but I wasn't expecting this at all haha.

[tpp]

Have you checked out the 480i and 480p modes yet? I can't tell any difference between 480ix2 and 480ix4 (can't get 480ix3 to work at all) and the 480px2 mode vs passthrough on my LG tv looks identical.

Was just curious, since you have an LG as well, mines a few years older and not oled, but I can't imagine they changed the scaler too much between 4k tvs.

480ix4 looked a touch sharper, but I didn't examine it closely. I expect I'll play most 480i titles in passthrough - I don't play fighters and just about every major fast-action PS2 game has been properly remade for PS3/4 at this point. 480ix3 doesn't get me a picture either. I haven't tested 480p at all.

[Hoagtech]

I received the newer 1.6 OSSC and got a chance to use to compare it to an older OSSC I own.

That being said I have a couple questions.

Bucko stated on his site that 1.6 has a built in "Analogue to Digital" converter. Does this work both ways as a DAC analogue output? Can I throw away my Tendak?

Also I wasn't able to get sound out of the HDMI output to my monitor . I connected the old Headphone male cable coming from my Surround into the audio output of my Tendak because I am using it on a RGBHV monitor so I have no way of passing audio through the HDMI to D-Sub converter.

I thought it should work but no sound was coming through the Tendak to my surround. Is an option or a switch to switch between A1 out / A2 in?

Also my 480p seems to have an issue handling black ranges on older televisions. I tried switching component Colorspace from rec 601 to 701 but I feel like there should be an option to have limited color range tv sets handle color ranges that are darker than black like the wii and xbox and GC?

[orange808]

If you have a LCD display with 10ms of lag compared to CRT then the Bodnar tester will report 10, 18, 26ms at top, middle, bottom.

Not really, though.

Using the Leo Bodnar, my monitor and television report different amounts of delay versus CRT at each testing box--and the real lag versus CRT declines as the frame is refreshed ("drawn from top to bottom").

I assume the panel "draws" each frame faster than a CRT, so the panel is able to recover some processing time by refreshing the frame faster than the beam would.

So, the actual input lag is lowest at the bottom of the screen for those displays, because the panels can work faster than 60Hz.

----

I'm looking forward to getting the OSSC tester running myself. Should be fun.

[orange808]

Has anybody tried the lag test when passing through 480i? It would be nice to get some deinterlacing delay estimates without a camera.

[citrus3000psi]

Here is the input jack for the latency tester.

[paulb_nl]

Using the Leo Bodnar, my monitor and television report different amounts of delay versus CRT at each testing box--and the real lag versus CRT declines as the frame is refreshed ("drawn from top to bottom").

Yea I should have said 60Hz LCD. Marqs has described on the wiki that faster displays will have the least lag with the bottom box. Its similar to plasma.

Here is the input jack for the latency tester.

Nice. Will you just route the wires in front of the OSSC to the button? Im planning to route the wires with a connector to the back of the OSSC.

BTW Just for fun you can also use the latency tester as reaction speed tester. Just press button 1 when the white box appears. I got around 140-170ms

[RaphM]

I got my OSSC this weekend and just finished setting everything up / digging through the settings etc... .
I was especially hyped about the line5x mode, finally got it to work and wanted to share some insights:

* My TV: Sony kdl55w755C (which is known for low input lag in game mode, 1080p screen), no other devices but a gscartsw_lite in the chain
* "Allow TVP HPLL2x" set to off

Initially 5x mode was not recognized, but with the above setting I was able to "trick" my TV into accepting the mode by switching between the "Line5x modes":
Usually a single "Generic 4:3" -> "320x240 optim." -> "256x240 optim" -> "Generic 4:3" cycle is enough and the TV will suddenly accept the image (and it also never dropped again, even on console resets).
Performing this trick is bit annoying though, maybe somebody got some advanced scaling settings that could help s.t. the output is accepted initially?

But anyway, I am super happy with the OSSC and am now connecting all my 240p consoles through it.
The Framemeister is still used for Shenanigans like 7x zooming into the Super Gameboy and interlaced sources (which is basically only my PS2), but for all other uses the super low input lag (I'm reaching ~1 frame results in the manual lag test of the 240p test suite) and perfect color processing of the OSSC make it my new #1 choice .

[bobrocks95]

I got my OSSC this weekend and just finished setting everything up / digging through the settings etc... .
I was especially hyped about the line5x mode, finally got it to work and wanted to share some insights:

* My TV: Sony kdl55w755 (which is known for low input lag in game mode, 1080p screen), no other devices but a gscartsw_lite in the chain
* "Allow TVP HPLL2x" set to off

Initially 5x mode was not recognized, but with the above setting I was able to "trick" my TV into accepting the mode by switching between the "Line5x modes":
Usually a single "Generic 4:3" -> "320x240 optim." -> "256x240 optim" -> "Generic 4:3" cycle is enough and the TV will suddenly accept the image (and it also never dropped again, even on console resets).
Performing this trick is bit annoying though, maybe somebody got some advanced scaling settings that could help s.t. the output is accepted initially?

But anyway, I am super happy with the OSSC and am now connecting all my 240p consoles through it.
The Framemeister is still used for Shenanigans like 7x zooming into the Super Gameboy and interlaced sources (which is basically only my PS2), but for all other uses the super low input lag (I'm reaching ~1 frame results in the manual lag test of the 240p test suite) and perfect color processing of the OSSC make it my new #1 choice .

This is super interesting since I think it's the first time the low-lag Sonys have been proven to work (at least to some degree) on a setting above 2x?

I'm sure 480px2 remains broken, but for your x5 setup do the NES/SNES work?

[Xyga]

interesting indeed !