OSSC (DIY video digitizer & scandoubler)

[Galdelico]

For non-experts like me that are curious about why sampling adjustments matter - here's a concise, but neat explanation

Very interesting, thanks! ^_-

Edit - Reposting your link because of the new page.

[lui]

Rotating the image 90 degrees via a frame synchronizer, should this be done before or after the OSSC's processing?

[nmalinoski]

Rotating the image 90 degrees via a frame synchronizer, should this be done before or after the OSSC's processing?

If you're using multiple inputs on the OSSC, I would think connecting any kind of video processor after the OSSC would result in a simpler setup. (And it's probably easier to find HDMI video processors these days, compared to RGB/YPbPr ones.)

[Blair]

@Blair
That meme made my afternoon
I was pretty much taking a guess at what I was seeing in your screen shot was ghosting or ringing.

It's hard to tell what's going on really, in the gameplay screen shot the right edges of the top of the cave are green, and the same goes for some edges in your Sonic title screenshot. Is it just the game?

lol, thanks (for some reason I've been thinking about Star Trek quotes lately)

Yeah there's something weird going on either the GameCube or my component cable. unfortunately I can't just swap out another component cable to test, as their rarer than hens teeth these days. I might swap out my GameCube as I have another in storage to test.

I'm really into the 480px2 mode with up-sample 2x on, I think it makes the colours look so much nicer on the DVDO and less ringing to boot.

Yeah, definitely has a different look to it than normal pixel repetition, surprised that the "my life in gaming" guys didn't even bother to mention it during their PS2 video. FBX, will you be experimenting with profiles based on normal pixel repetition and/or up sample 2X modes? Does that affect how your phase tuning technique works?

I think Thomago, (in another thread) mentioned that it's necessary to use up-sample 2x in order to get proper results with certain games. like God of war 2?

I was a fan too but recently learned that result can vary somewhat depending on the game. I didn't spot anything too off-putting on Blair's vid but have a look at this Outrun 2 captures (I posted them a few pages back to praise 2x480p but they backfired on me ):

I guess that, in this specific case, the in-game AA is completely defeated by line-doubling.

I'm glad you didn't see anything too terrible in my videos, lol. (I also just put up some new ones) But I think you're 2X shots look pretty darn good especially the up-sample 2x one. The "native" shot looks like somebody smeared Vaseline all over it. I wonder if the CII has difficulties with certain games/signals when it comes to the PS2. (Similar to how the HD 3000 can act up with direct PS2 signals) or maybe just needs calibrating? regardless I'd be interested to see a similar comparison with something like the Xbox or Dreamcast in that same scenario.

[Xer Xian]

I'm glad you didn't see anything too terrible in my videos, lol. (I also just put up some new ones)

Ahah sorry that probably sounded somewhat negative, I was referring specifically to line2x480p mode not your videos in general It's quite the opposite actually, often I find your videos/pictures to be the best expression of the underlying tech they showcase. I have tested line multiplied 480i on my Electron IV and got mostly crap results, while it was great on some shots you posted here sometime ago, and it's the same for the videos you posted today - I've had a VP50 (but non pro) for a short while and don't think I could ever get it to look as good with 480p material (regarding the FM vid though - that one looked a bit oversaturated to me).

As for the CII, I don't think it has any trouble with the PS2, it just looks a bit soft when fed digital sources for some reason (all the Outrun captures were from an OSSC+CII combo). It may be limited to my specific model and/or firmware, I don't know. Anyway, I feel that the CII can have the sharpness setting cranked up several notches before showing ringing artifacts, so it's not really a problem. If you're interested, I've just captured a sample from a Dreamcast as well - OSSC set to sample 480p as VGA and upscaled to 1080p with the CII (sharpness set to +5.. I barely touched this setting, plenty of room left):
480p: http://image.ibb.co/kaM3VG/480p.jpg
960p upsampled: http://image.ibb.co/cOd6jb/960p_upsample.jpg
960p: http://image.ibb.co/hBDvcw/960p.jpg

Not exactly the same frames (did Ryo change his hairstyle in between the first two shots or what ), but it's clear that 480p is slightly softer on the DC as well. I've got a few other upscalers that I could test this with, but the only ones that can touch the upscale quality of the CII do not accept 960p unfortunately. Also I'm not entirely sure I prefer the sharper look, but anyway it's barely noticeable, so.. whatever.

[djc5166]

I'd be pretty pissed off if 4x mode were dropped. It's my favorite mode for both the OSSC and the Framemeister when it comes to 240p content.

If you did video content, it should be 3x.

This is the first video I've made of a game with the fully calibrated OSSC and new workflow to get the sharpest/highest quality image out of it. https://www.youtube.com/watch?v=xJD79qZk0as

Oh I gotcha, for 4k upscaling, yeah 3x is ideal. I was speaking for us folks in the stone age with 1080p displays.

Nub owner of a B7 here,

Why is 3x ideal for 4k over higher modes?

[Zappyraccoon]

I don't notice any benefit setting sharpness to 5 on the Crystalio. I find it much more of a noticeable difference to set smooth scaling to -3 and leaving sharpness at 0.

[Zappyraccoon]

Nub owner of a B7 here,

Why is 3x ideal for 4k over higher modes?

Probably because 720p is perfect integer scaling.

[Xer Xian]

But that would only make a difference if we assume that the TV will integer scale 720p/1080p to UHD, which is not guaranteed.

I don't notice any benefit setting sharpness to 5 on the Crystalio. I find it much more of a noticeable difference to set smooth scaling to -3 and leaving sharpness at 0.

I used the checkerboard frame of the OSSC starting test pattern as reference - in my case +5 was a small improvement (didn't want to overdo it), but it was there. Anyway, thanks for mentioning the 'smooth scaling' thing, I went into 'video settings' and totally overlooked it

[fafangus]

Whow, how many people use the OSSC + C2 combo here ?

[citrus3000psi]

Whow, how many people use the OSSC + C2 combo here ?

I use mine occasionally, but sometimes my CII doesn't like audio over hdmi, so I'll get blurbs etc. I'm sure its related to its unlocked frame rate.

[Sega90s]

Is there any Dreamcast VGA box that works better with the OSSC out of the many avaliable
options?

[Tapeworm]

Has anyone used one of these with something like a PVM 20L5 or a JVC DT-V1710CG? Curious what the results are like. I'm assuming best you could do is 240p LineX3, but something like 480p would have to stay 480p?

[Dochartaigh]

Has anyone used one of these with something like a PVM 20L5 or a JVC DT-V1710CG? Curious what the results are like. I'm assuming best you could do is 240p LineX3, but something like 480p would have to stay 480p?

Just wondering why you would want to do that when the 20L5 and similar does native 240p, 480p, etc.

[FBX]

I was wondering if Marqs could give us a comment on how the audio processing in the OSSC is handled? Reason I ask is I and several others have noticed audio issues, where the audio doesn't work at all, pops in and out on HDMI receivers, etc. It seems like the audio circuitry is somehow intermixed with the oddball refresh rates of optimal timing settings, causing most audio amps and processors to reject it. It might be a situation where audio must be routed independent of the OSSC in order to prevent the output issues from happening.

[Zappyraccoon]

I was wondering if Marqs could give us a comment on how the audio processing in the OSSC is handled? Reason I ask is I and several others have noticed audio issues, where the audio doesn't work at all, pops in and out on HDMI receivers, etc. It seems like the audio circuitry is somehow intermixed with the oddball refresh rates of optimal timing settings, causing most audio amps and processors to reject it. It might be a situation where audio must be routed independent of the OSSC in order to prevent the output issues from happening.

That's what I am wanting to do, though I am not sure if there are any automatic RCA switches that can detect only audio. I mean I'm sure there are but which ones are quality or ones that match what I'm looking for (preferably only the audio rca plugs with maybe spdif).

[marqs]

I was wondering if Marqs could give us a comment on how the audio processing in the OSSC is handled? Reason I ask is I and several others have noticed audio issues, where the audio doesn't work at all, pops in and out on HDMI receivers, etc. It seems like the audio circuitry is somehow intermixed with the oddball refresh rates of optimal timing settings, causing most audio amps and processors to reject it. It might be a situation where audio must be routed independent of the OSSC in order to prevent the output issues from happening.

Audio muxing to HDMI stream is fully handled by HDMI transmitter chip. Counting out sources which end up generating jitter on TMDS clock (mainly NES/SNES), audio issues are likely to be caused by audio regeneration logic on receivers' end. Audio packets are sent during hblank, so receivers may pose some restriction on its (or hsync portion) length that may not be met with tweaked timing parameters. Another possibility is that the HDMI TX chip does not auto-calculate N/CTS parameters (used in audio regeneration) correctly, but fortunately those could be overridden by FPGA-calculated values if necessary.

[FBX]

Audio muxing to HDMI stream is fully handled by HDMI transmitter chip. Counting out sources which end up generating jitter on TMDS clock (mainly NES/SNES), audio issues are likely to be caused by audio regeneration logic on receivers' end. Audio packets are sent during hblank, so receivers may pose some restriction on its (or hsync portion) length that may not be met with tweaked timing parameters. Another possibility is that the HDMI TX chip does not auto-calculate N/CTS parameters (used in audio regeneration) correctly, but fortunately those could be overridden by FPGA-calculated values if necessary.

Hmm. I guess if I had my way, a new version of OSSC would be designed to have all incoming audio processed independently of video, digitized, and then passed to a digital audio out port (be it TOSLINK or coaxial).

BTW, playing around more with the 256x240 optimal timing line4x modes, I discovered the 4:3 setting is perfectly aligned. So it's only the 8:7 mode that doesn't line up with the active game graphics when H. active is set to 256. One side is pinched while the other has overscan (and shifting the backporch just switches the side pinched). I've not checked this on the other linexx modes, just 4x at this point.

[Seraphic]

Whow, how many people use the OSSC + C2 combo here ?

I am too as it normalizes the signal for use with my TV. OSSC 1.6 and Crystalio II VPS-3800

I use mine occasionally, but sometimes my CII doesn't like audio over hdmi, so I'll get blurbs etc. I'm sure its related to its unlocked frame rate.

Forgot you had a Crystalio II also, lol.
I had that problem too, but only with the GameCube with GCVideo over HDMI which is why I added the toslink mod as a bypass as needed.

[FBX]

Has anybody found optimal H. samplerate yet for XBOX in 480p mode 2X? I've been looking, but so far it's eluding me.

[TestType]

Has anybody found optimal H. samplerate yet for XBOX in 480p mode 2X? I've been looking, but so far it's eluding me.

What are you using to find the optimal H. samplerates for 480p/i consoles, if you don't mind me asking? Wondering if I can get my GameCube and XBOX (and PS2) to look even better. I'm guessing you are using your capture card as you have previously mentioned, but would a checkerboard pattern not work in 480p/i for adjusting H. samplerate? Not sure if there is much in the way of image viewers on these consoles to view such an image though, except the XBOX.

[FBX]

Has anybody found optimal H. samplerate yet for XBOX in 480p mode 2X? I've been looking, but so far it's eluding me.

What are you using to find the optimal H. samplerates for 480p/i consoles, if you don't mind me asking? Wondering if I can get my GameCube and XBOX (and PS2) to look even better. I'm guessing you are using your capture card as you have previously mentioned, but would a checkerboard pattern not work in 480p/i for adjusting H. samplerate? Not sure if there is much in the way of image viewers on these consoles to view such an image though, except the XBOX.

I've not been worrying about interlace just yet, but 480p on the xbox is proving quite elusive. The closest I can find is a value of 840, and that generally seems to put 640 games in the right horizontal width, but the sampling still doesn't perfectly align with each pixel. It's just not making any sense to me on the XBOX.

[jayde6]

I've not been worrying about interlace just yet, but 480p on the xbox is proving quite elusive. The closest I can find is a value of 840, and that generally seems to put 640 games in the right horizontal width, but the sampling still doesn't perfectly align with each pixel. It's just not making any sense to me on the XBOX.

Does changing the "480p in sampler" to 640x480@60 help at all? Allow upsample2x?

[FBX]

Does changing the "480p in sampler" to 640x480@60 help at all? Allow upsample2x?

Already tried both. Neither helped. I'm fairly certain Xbox is meant to use DTV anyway.

[xadox]

I'd be pretty pissed off if 4x mode were dropped...

Is there a benefit using 4x instead of 5x?
My OSSC is directly connected to the TV and working fine with 5x @ 1920x1080.

[Kez]

Is there a benefit using 4x instead of 5x?

5x cuts off some pixels from the top and bottom. In most games this isn't much of an issue but 4x is still the sharpest mode that displays all 240 lines. Scanlines also look a bit better in 4x in my experience.

[marqs]

Hmm. I guess if I had my way, a new version of OSSC would be designed to have all incoming audio processed independently of video, digitized, and then passed to a digital audio out port (be it TOSLINK or coaxial).

Toslink/coaxial input ports unfortunately are getting more rare on TVs/monitors and there's no replacement dedicated digital audio interface in sight. In amps / AV receivers they'll be certainly around for the foreseeable future but so will be analog inputs too, somewhat defeating the purpose of using OSSC for audio digitization.

I've not been worrying about interlace just yet, but 480p on the xbox is proving quite elusive. The closest I can find is a value of 840, and that generally seems to put 640 games in the right horizontal width, but the sampling still doesn't perfectly align with each pixel. It's just not making any sense to me on the XBOX.

Does changing the "480p in sampler" to 640x480@60 help at all? Allow upsample2x?

"480p in sampler" just switches the preset mode for 480p content - with adv. timing adjustments their parameters can be made identical.

I'd be pretty pissed off if 4x mode were dropped...

Is there a benefit using 4x instead of 5x?
My OSSC is directly connected to the TV and working fine with 5x @ 1920x1080.

There wouldn't be much point to drop 4x or any other specific line multiplication mode - doing so would not free much resources.

[nmalinoski]

Hmm. I guess if I had my way, a new version of OSSC would be designed to have all incoming audio processed independently of video, digitized, and then passed to a digital audio out port (be it TOSLINK or coaxial).

Toslink/coaxial input ports unfortunately are getting more rare on TVs/monitors and there's no replacement dedicated digital audio interface in sight. In amps / AV receivers they'll be certainly around for the foreseeable future but so will be analog inputs too, somewhat defeating the purpose of using OSSC for audio digitization.

Perhaps I've been looking at the wrong TVs, but I've not seen a TV with S/PDIF inputs--only outputs--and those outputs are probably disappearing due to the increasing availability of the more-capable HDMI ARC.

In my case, I'd like to have TOSLINK inputs (an output would be a nice-to-have) on the OSSC, so it can handle audio switching for me. Currently, I have four consoles hooked up to am automatic component AV switch with TOSLINK, where analog/RCA audio gets digitized by the OSSC before hitting my receiver, and the digital/TOSLINK goes directly into my receiver, bypassing the OSSC. With this setup, if I want to switch from a console with analog/RCA audio to one of my PS2s, which are hooked up solely with TOSLINK, I have to go into the OSSC's settings and switch TX Mode from HDMI to DVI*.

I maybe could just route the RCA audio from the component switch directly to the receiver, leaving the OSSC to process only video, but that won't help me when my Toro or PlayStation SCART cable come in.


* I think this is because the OSSC always puts a digitized audio signal on HDMI, even if there's no audio signal or cable connected, and my receiver won't fall back onto TOSLINK as long as HDMI has audio.

[xga]

I'd really love to see a TOSLINK / S/PDIF input on the OSSC for consoles I have modded for digital audio also.

Would it even be technically possibly to add this feature to v1.6 OSSC's if there was space to solder the required hardware to the PCB or would this have to be a v1.7 board or something?

[FBX]

I'd really love to see a TOSLINK / S/PDIF input on the OSSC for consoles I have modded for digital audio also.

Would it even be technically possibly to add this feature to v1.6 OSSC's if there was space to solder the required hardware to the PCB or would this have to be a v1.7 board or something?

Adding a digital audio input to the OSSC isn't solving the problem of the OSSC's HDMI output audio being rejected by receivers/amps. My idea was to have a digital audio output to send to amps that is processed independently of the video signal. It would least provide a stable solution for those of us that need it. And while digital audio input on a TV is rare, it isn't rare on an amp/receiver.

Of course, my perfect world there would be digital audio input AND output on the OSSC.