OSSC (DIY video digitizer & scandoubler)

[marqs]

Is there a release date for the first batch yet?

Pilot batch was already released back in April, and next batch of 100 boards should be out in a few weeks.

Does the OSSC accept a 15.6 KHz signal on the VGA input connector ?

Some PCBs like System 12 from NAMCO have a JVS output section with a VGA connector.

Yes, 15kHz is supported on all inputs. Howevre, VGA input is not routed through LPF chip so the sources should have good signal quality. BTW, fw. 0.70 will have vsync threshold option which should improve compatibility of Taito F2/F3 boards. The fw image is currently being tested by a few selected people, and public release should be in a few days.

[rtw]

Does the OSSC accept a 15.6 KHz signal on the VGA input connector ?

Some PCBs like System 12 from NAMCO have a JVS output section with a VGA connector.

Yes, 15kHz is supported on all inputs. Howevre, VGA input is not routed through LPF chip so the sources should have good signal quality. BTW, fw. 0.70 will have vsync threshold option which should improve compatibility of Taito F2/F3 boards. The fw image is currently being tested by a few selected people, and public release should be in a few days.

Thank you for the information, looking forward to the F2/F3 fix

[blizzz]

I've spent a couple days with the OSSC and compiled my impressions into a little article. The unit I had was running firmware 0.70 and had the audio add-on installed.



https://blz.la/ossc

[BuckoA51]

Nice overview, note that the aspect ratio in line triple should be fixed in a future firmware.

[Xyga]

Nice overview, note that the aspect ratio in line triple should be fixed in a future firmware.

@Marqs: On that topic would it be possible to keep the slightly wider~incorrect 4:3 output as an option ?
I kind of like it with some sources and games.

[ZellSF]

I've spent a couple days with the OSSC and compiled my impressions into a little article. The unit I had was running firmware 0.70 and had the audio add-on installed.



https://blz.la/ossc

Good article.

I feel I must be that guy (partially since I make this mistake pretty often myself): it's separate, not seperate.

[blizzz]

note that the aspect ratio in line triple should be fixed in a future firmware.

Good to know. I've added that info to the text.

On that topic would it be possible to keep the slightly wider~incorrect 4:3 output as an option ?
I kind of like it with some sources and games.

I have to agree, the slightly wider AR does look nice. When you're looking at 16:9 or 16:10 screens for hours each day the correct 4:3 actually looks a bit too small.

it's separate, not seperate.

What makes it worse is that it's also spelled separat in German. Also fixed a couple other spelling mistakes.

[RGB32E]

I've spent a couple days with the OSSC and compiled my impressions into a little article. The unit I had was running firmware 0.70 and had the audio add-on installed.

Nice write-up!

Due to the way the signal is created it is not a standard conform 720p video and your Sony TV will most likely not accept it.

Which model of Sony TV are you using?

My Sony KDL-55XBR8 accepts/syncs line tripled SNES, Genesis, Saturn, TG16, NESRGB, and 2600RGB to name a few. These are all NTSC native sources on a NTSC/U TV though. Perhaps PAL and PAL->NTSC converted sources don't play as nice and/or the multi-region TVs sold in Europe are somehow less tolerant? I'm interested in narrowing down which specific models/years have line tripled compatibility issues, as it appears you might be painting with a broad brush. IIRC Fudoh's German Sony HX9 had issues with line tripled sources as well. There was a spreadsheet floating around somewhere...

Looking forward to future FW releases!

[blizzz]

Which model of Sony TV are you using?

It's a Sony Bravia KDL40EX655.

The spreadsheet is here: https://docs.google.com/spreadsheets/d/ ... =923290594
According to the data in it no tested Sony TV accepted the 720p output of the OSSC. Or is the info outdated?

[RGB32E]

Which model of Sony TV are you using?

It's a Sony Bravia KDL40EX655.

The spreadsheet is here: https://docs.google.com/spreadsheets/d/ ... =923290594
According to the data in it no tested Sony TV accepted the 720p output of the OSSC.

Thanks! That makes sense given that data.

Or is the info outdated?

Possibly? I haven't updated added my results to the spreadsheet. I also have a KDL-40EX400 I need to test too!

[mechacrash]

I've spent a couple days with the OSSC and compiled my impressions into a little article. The unit I had was running firmware 0.70 and had the audio add-on installed.



https://blz.la/ossc

Really helpful thanks! I actually didn't know the different playstation models had different oscillation frequencies - a quick look at the PS2 schematics simply shows "54MHz" for all models, does this mean this isn't a problem for PAL modded PS2s?

Also thanks for checking out the GBI-ULL, shame that it doesn't work. Is it that the OSSC itself has issues, or your TV fails to take the output from the OSSC?

[blizzz]

I actually didn't know the different playstation models had different oscillation frequencies - a quick look at the PS2 schematics simply shows "54MHz" for all models, does this mean this isn't a problem for PAL modded PS2s?

It's probably not a problem anymore in the PS2 / GCN generation, since even the PAL consoles could output at 60Hz 480i without mods.

Also thanks for checking out the GBI-ULL, shame that it doesn't work. Is it that the OSSC itself has issues, or your TV fails to take the output from the OSSC?

There's not a huge difference between the LL and ULL version, so it's not a big problem. I believe the OSSC couldn't sync to the ULL version.

[Harrumph]

My Sony KDL-55XBR8 accepts/syncs line tripled SNES, Genesis, Saturn, TG16, NESRGB, and 2600RGB to name a few.

I encourage everyone with an OSSC to report TV linetriple compatibility in this thread in the videogameperfection forum.
This should help gather data on compatible TVs without putting strain on Marqs & Bucko who maintain the google spreadsheet.

Also, nice writeup Blizz, and thanks for contributing your data!

[Shuco13]

@blizzz
Really interesting and informative review, thanks a lot. It's a shame you didn't have the possiblity to compare the 1080p output of the Framemeister with the OSSC.
I think you have exaggerated on the color inaccuracies of the FM, especially given that the range issue was fixed by the last firmware. Furthermore you completely forgot to mention that the FM has numerous color settings to adjust the picture whereas the OSSC has got none (at least that you did mention).

[BazookaBen]

I've spent a couple days with the OSSC and compiled my impressions into a little article.

Great read! Question about the input lag part: what mode were you running to your LED monitor? Line-double? Triple?

[blizzz]

I agree that I may be too harsh on the Framemeister in that regard, but it has been such a big issue for me up until the latest firmware update. I will update the text so that it doesn't sound like a bigger difference than it actually is.



That is a graph of the colors I get from the RGB color bar screen of the 240p Test Suite with both the Framemeister and OSSC using my 1CHIP SNES. The OSSC is almost perfect, with only the blue channel being a tiny bit lower than the rest. That's within the natural spread of the video encoder, so it might just be my SNES. The Framemeister (with YCbCr output) on the other hand has quite a bit of difference between the channels. When I set the Framemeister to RGB output it gets even worse, with the green channel being way too high and red and blue a bit too low (Red: 207, Green: 254, Blue: 219). This is with the same console and cables. My settings for the Framemeister are: Brightness = 25, Gamma = 19, Black = 2, A/D Level = 130. That's brighter than the settings FBX recommends.

I just think that the OSSC does a really great job with the colors.

For the 1080p vs 720p comparison I can give you the screenshots I took. (480p and 720p screenshots taken with the SC-500N1 in 4:2:2; 1080p taken with the Elgato HD in lossy 4:2:0)
OSSC: 480p, 720p
Framemeister: 480p, 720p, 1080p

@BazookaBen, Both scalers were set to Line Triple / 720p. RetroRGB also confirmed my results. He'll post his review of the OSSC soon too.

[NormalFish]

Great article, Blizzz. The impressions will be increasingly helpful as the quirks of the device are better understood and dealt with, so I'm always anxious to read new impressions.

Your footage for 480i w/ bob deinterlacing actually looks fine to me. I'm pretty easily impressed when it comes to deinterlacing, though. Recently was playing 480i games in Picture mode on my mini and it took me an embarrassing amount of time to realise.

Also puts me a bit more at ease seeing your success with an Asus monitor. That's what I'll be using so I'm optimistic things will work. Was there anything that didn't work with your monitor?

[Shuco13]

@blizzz
Thanks for the clarification and the comparison shots. I think I get the difference now, it's pretty obvious. I also want to congratulate to the first real showcase of lag comparison between the two scalers.

That is a graph of the colors I get from the RGB color bar screen of the 240p Test Suite with both the Framemeister and OSSC using my 1CHIP SNES. The OSSC is almost perfect, with only the blue channel being a tiny bit lower than the rest. That's within the natural spread of the video encoder, so it might just be my SNES. The Framemeister (with YCbCr output) on the other hand has quite a bit of difference between the channels. When I set the Framemeister to RGB output it gets even worse, with the green channel being way too high and red and blue a bit too low (Red: 207, Green: 254, Blue: 219). This is with the same console and cables. My settings for the Framemeister are: Brightness = 25, Gamma = 19, Black = 2, A/D Level = 130. That's brighter than the settings FBX recommends.

I think the problem with the FM, as usual, is that the "out-of-the-box" settings aren't anywhere near as accurate as the OSSC's and far from universal for every input. My claim is though that, provided you have a meter and appropriate test patterns for every input, you can achieve the same results for color bars as with the OSSC using the numerous settings given. That this is something 99% of users might not be capable of is a different story.

[blizzz]

Your footage for 480i w/ bob deinterlacing actually looks fine to me.

There's nothing wrong with bob deinterlacing, even with high movement there are no artifacts. It's just not the best looking deinterlacer because of the shimmering.

Also puts me a bit more at ease seeing your success with an Asus monitor. That's what I'll be using so I'm optimistic things will work. Was there anything that didn't work with your monitor?

No issues with my Asus monitor. No input lag and it lets you scale the 480p output to 4:3. The only minor issue I found was that it was a tiny bit slower than my TV or capture card to resync after resolution changes.

My HP on the other hand doesn't have a 4:3 scaling option, so the 480p output is a bit too wide. Not a huge issue though, since both monitors accept the line triple mode.

I think the problem with the FM, as usual, is that the "out-of-the-box" settings aren't anywhere near as accurate as the OSSC's and far from universal for every input. My claim is though that, provided you have a meter and appropriate test patterns for every input, you can achieve the same results for color bars as with the OSSC using the numerous settings given.

I've tried it. Increasing the brightness with the A/D Level slider is not an option since it crushes the brighter shades. The Gamma option works, but even when I set it to max (63) the whites are not as bright as on the OSSC. But this discussion probably belongs in the Framemeister thread.

[ZellSF]

More on DVDO and OSSC: if you have this make sure to disable both Cadence Detection and HDCP before testing, both give sync issues. I know most people here will disable them eventually anyway, but they might try to test with them on thinking them to be harmless.

Just wasted an hour before I figured I had sync issues because I left Cadence Detection on on the OSSC input.

On OSSC's 480i bob deinterlace: I think it looks fine on stuff that's in motion, horrible on static 2D UI elements. Personally not ever going to use it and can't wait for 480i passthrough (so I can get rid of my dedicated PS2 transcoder).

Not sure how useful going "this doesn't have XRGB-mini bug X" is, the default assumption should be that an entirely different product shouldn't share the same software bugs, more important to mention if they have any bugs in common.

There's not a huge difference between the LL and ULL version, so it's not a big problem. I believe the OSSC couldn't sync to the ULL version.

Are you sure? What does the OSSC LED and LCD display say when running the ULL version? I want to try this soon, if I can just get Codejunkies to send the right product (they sent me the wrong region SD media launcher disc).

[blizzz]

I only tested GBI briefly. The ULL didn't show on either my capture card nor my monitor, so I switched to the LL version which worked without problems. Maybe someone else can test the ULL and check if the red LED lights up?

[Guspaz]

I believe that the ULL and LL versions are identical in every way except:

- LL runs at a slightly higher framerate, making it compatible with more displays
- LL adds an extra frame of lag, to solve the ULL tearing issue
- LL doesn't have the tearing issue
- LL only supports 240p, not 480p (ULL supports both).
- LL may have a one-frame stutter every 10 seconds (since it runs 0.1 FPS faster than the GBA)

[bobrocks95]

I will say, ULL never had screen tearing for me- it seems very hit or miss depending on display or maybe cube. LL definitely doesn't have it though and is a great compromise.

[BazookaBen]

The screen tearing is very infrequent. Maybe once every half hour. And it's gone within a minute.

[Guspaz]

The frequency/severity/duration differs from cube to cube. It was tolerable on my NTSC cube, but I've not had a chance to test my PAL one extensively to see what happens.

[marqs]

I quickly checked details of GBI LL and ULL. While LL runs at 59.8261Hz (=27MHz/(858*526) I think), ULL runs at 59.7276Hz which cannot be properly generated with 525+-1 scanlines if GC uses fixed 27MHz pixel clock. It needs to either use some uncommon line count (breaks display compatibility) or adjust scanline length dynamically (potential sync issues like with SNES).

[RocketBelt]

On OSSC's 480i bob deinterlace: I think it looks fine on stuff that's in motion, horrible on static 2D UI elements. Personally not ever going to use it and can't wait for 480i passthrough (so I can get rid of my dedicated PS2 transcoder).

How would you describe it in comparison to dvdo game mode 1?

[BuckoA51]

It looks very similar to that. It's quite suitable for some games, especially faux 480i games with scanlines turned on, but not really suitable as a general purpose deinterlace.

[Fudoh]

It looks very similar to that

my answer would have been: they couldn't look any more different. Total opposites of each other. But to be fair: there's a difference in game modes between the VPxx processors. DVDO's goal was to eliminate line tremble as much as possible, so, the fields are shifted even more towards each other than on the OSSC. Also you get the whole package of diagonal interpolation on the DVDOs, where you get none on the OSSC.

[BuckoA51]

I'm thinking of the DVDO Edge to be honest, haven't used game deinterlacing much in the 50 Pro yet.