OSSC (DIY video digitizer & scandoubler)

[Mantrox]

It was late, i spent 14 hours at work and my brain was frozen apparently.
When i turned on the DVDO i went straight to the game mode and turned it off, instead of leaving it on.
I didn't even process what i was doing...

I'll get some better shots with settings in an album and post it, for some reference, later this week.

Checklist:
- OSSC linedouble -> Monitor
- OSSC linetriple -> Monitor
- OSSC linedouble -> DVDO Game mode On -> Monitor
- OSSC linedouble -> DVDO Game mode Off -> Monitor
- OSSC linetriple -> DVDO Game mode On -> Monitor
- OSSC linetriple -> DVDO Game mode Off -> Monitor

I hooked up the SNES to the BVM and switched to the OSSC+DVDO with game mode off and the lag was kind of atrocious.
I do complete Super Ghouls n' Ghosts runs from time to time, and when i do a double jump and fire a javelin on the first dog on the rock, i allways hit him on the BVM, but on the DVDO i hit the rock.
Again, game mode off... because i'm an idiot.
I'll have to redo the test and see what it's like with lowest lag settings.

[BuckoA51]

I get 6ms input lag with the DVDO, OSSC and SNES. Game mode on obviously.

[Mantrox]

With that value, on my setup with the dell monitor, the total chain average lag should be about 18ms.
Just over a frame, which should be fine.

[Xyga]

Nothing in the mailbox today either, Finland is too far away !

(I remember when I was working in e-commerce that anything shipping from or to most Northern-Europe destinations took ridiculously long paths with many warehouse stops, unless you paid for really expensive third party carrier solutions of course)

[amaradona]

Nothing in the mailbox today either, Finland is too far away !

(I remember when I was working in e-commerce that anything shipping from or to most Northern-Europe destinations took ridiculously long paths with many warehouse stops, unless you paid for really expensive third party carrier solutions of course)

ahahha bien fait pour toi

I got it today, cant wait to test it tonight.
I will also partner it with the optoma hd3000.

[BuckoA51]

An order for the next batch has now been placed with the manufacturer. Hoping to have the units late this month/early next.

[Blair]

An order for the next batch has now been placed with the manufacturer. Hoping to have the units late this month/early next.

Great news, the price was around $200-ish, right?

With that value, on my setup with the dell monitor, the total chain average lag should be about 18ms.
Just over a frame, which should be fine.

Yeah, the Edge should be outputting a maximum delay of 6ms.

Besides turning on game mode, don't forget to turn off other extras like. Cadence detection, Noise Reduction. (And make sure deinterlacing is set to video)

[FinalBaton]

Blair, your setup on the LaCie looks friggin sick!

[BuckoA51]

Great news, the price was around $200-ish, right?

Give or take twenty bucks either way I think that should be about right. The pound is all over the place at the moment so it's hard to say.

[Blair]

Blair, your setup on the LaCie looks friggin sick!

Thanks FinalBaton!

It also has a good feel to it as well, thanks to the only 6ms of input lag in the chain (one of the major advantages of the DVDO's over the XRGB's)
and in motion it has a wonderful smooth and detailed scroll (movement is perfect). overall its great (sometimes even better than my 20L5) but I think it can be further enhanced with the OSSC added to the mix.



anther question (probably for marqs) is, could the OSSC ever use filters? I ask because to me this seems like a similar project to the "Hi-Def NES" (another RGB to HDMI FPGA solution with almost zero input delay) Hi Def NES uses a few filtering algorithms that (some) people might like (depending on the game).

here are some examples (keep in mind the filters have been improved with the latest FW)






And if that could be possible, would something like a de-dithering filter also be achievable? (perhaps with a future version, if not this hardware)

here are shots of a few early experiments I did in retroarch with bunny's dithering shader. i've used the profile feature to alter it and add scanlines, my current version looks much better now (especially on a CRT), i'll post some of those later if anyone is interested.

(apology for the direct video-out being so dark)

composite video



de-dithering tests















as you can see, the advantage is a sharp RGB quality image with composite transparency (dithering) effects and no blur. (its actually better because you don't get any of that awful banding) looks great in motion.

could that be done with the OSSC?

[Xyga]

IIRC things like Hi-Def NES and Ultra-HDMI use a 1 frame buffer to display those filters and stuff.
Not the kind of 'feature' I would welcome on the OSSC if it cannot be shut down completely to keep the output lagless...

[Blair]

IIRC things like Hi-Def NES and Ultra-HDMI use a 1 frame buffer to display those filters and stuff.
Not the kind of 'feature' I would welcome on the OSSC if it cannot be shut down completely to keep the output lagless...

according to this data sheet from Retro RGB, it has no frame buffer. so the filters have no added cost to input lag. for me i'm not interested in things like the HQ2x style of filter, but I would love to have something like the OSSC give us a de-dithering option.

[Xyga]

Ah. Dunno how they pulled that but the Ultra HDMI on its side specifies 1 frame delay when the effects are on...oh well.

[Unseen]

Ah. Dunno how they pulled that but the Ultra HDMI on its side specifies 1 frame delay when the effects are on...oh well.

Looking at the algorithms it seems that both hq_x and scale_x can be implemented by using data from the previous and next lines of the picture. Since looking into the future is a bit hard even for an FPGA you'd have to delay the output by at least a single line, but a complete frame buffer shouldn't be necessary for those two.

[parodius]

I never thought about de-dithering. On 16 bit games it looks nice, but on a Saturn it would be a godsend. Even at the expense of 1 frame of delay.

[ZellSF]

I'm conflicted. Dithering looks very bad, but so does the false positives (look at the crosses in the gui on the Mario World shot) and the blurriness dedithering causes. I'm somehow guessing the OSSC won't do it anyway (would be nice though).

[paulb_nl]

Here is some information about HQ2X on an FPGA http://fpganes.blogspot.nl/2013/02/the- ... rilog.html and even Verilog source https://github.com/strigeus/fpganes/blo ... src/hq2x.v

[BuckoA51]

De-Dither is quite clever, but does anyone really want HQ2X Scale 2X Scale 3X type effects? You might as well just go play an emulator.

[citrus3000psi]

Does anyone have a gerber paste file for V1.3?

I have a V1.2 prototype board and want to see if the audio board is a direct fit or will need modifications.

[Xyga]

does anyone really want HQ2X Scale 2X Scale 3X type effects?

Ack. No. ><

[blizzz]

does anyone really want HQ2X Scale 2X Scale 3X type effects?

Judging by the amount of screenshots, videos and streams that use these abominable filters I have to say 'yes', there are a lot of people who use them. But are those people also the same people who would pay 200€+ for video equipment to play on original hardware? Probably not.

[BuckoA51]

To be completely honest I remember when Scale2x etc first came out rather liking the effect in Yoshi's Island, but for most games it just looks like someone made an oil painting fan art out of it.

I suppose it's a cool novelty but I wouldn't want Marqs to waste valuable development time on it.

[bobrocks95]

Gaze into the mouth of madness and weep for your lost pixels



(The funny thing is that this looks pretty good compared to most Scale2x/HQx results)

[pyrotek85]

does anyone really want HQ2X Scale 2X Scale 3X type effects?

Judging by the amount of screenshots, videos and streams that use these abominable filters I have to say 'yes', there are a lot of people who use them. But are those people also the same people who would pay 200€+ for video equipment to play on original hardware? Probably not.

Exactly, nothing really wrong with it in itself, but I'd wager it's mostly a different segment of players who'd want that. Those of us on here want it looking as close to original as possible, generally speaking.

[alphagamer]

Gaze into the mouth of madness and weep for your lost pixels



(The funny thing is that this looks pretty good compared to most Scale2x/HQx results)

The beard looks very italian though

[marqs]

Feature request (hoping you get time to work on firmware soon): discreet IR commands for all 6 input modes (not just the three where you have to toggle to get right sync mode).

There'll be actually 8 input modes after YPbPr support will be added to SCART (or 9 if I add it to VGA also). Easy to map to buttons 0-9 as long as there won't be over 10 modes

Nothing in the mailbox today either, Finland is too far away !

(I remember when I was working in e-commerce that anything shipping from or to most Northern-Europe destinations took ridiculously long paths with many warehouse stops, unless you paid for really expensive third party carrier solutions of course)

I've now shipped all paid orders. Shipping should take no more than 6 workdays according to the estimate, which better be true considering the price.

Looking at the algorithms it seems that both hq_x and scale_x can be implemented by using data from the previous and next lines of the picture. Since looking into the future is a bit hard even for an FPGA you'd have to delay the output by at least a single line, but a complete frame buffer shouldn't be necessary for those two.

Yes, these filters should require no more than a couple of lines in buffer. Currenly OSSC can only store 2 scanlines, but with some memory reorganization that could be expanded to 5-8 lines. However, that would require quite a bit of work and I'm not personally a big fan of there filters so it's not currently on my todo list. Nothing prevents somebody else adding these filters on the firmware, though.

[marqs]

Does anyone have a gerber paste file for V1.3?

I have a V1.2 prototype board and want to see if the audio board is a direct fit or will need modifications.

It should fit directly. The only relevant difference is scart connector being 0.25mm closer to component connector but it should be within tolerance.

[BuckoA51]

Since some folks can't seem to configure their spam filters I just created a Telegram channel for OSSC announcements:- https://telegram.me/OSSC_announce

Telegram works on everything... Android, iOS, Windows Phone, Linux, OSX, Windows Desktop and the web.

I've also added the option to submit your Telegram username so I can use that as a fallback if I contact you via e-mail and don't hear anything. Use the welcome e-mail and edit your subscriber preferences to submit your username.

[pyrotek85]

Feature request (hoping you get time to work on firmware soon): discreet IR commands for all 6 input modes (not just the three where you have to toggle to get right sync mode).

There'll be actually 8 input modes after YPbPr support will be added to SCART (or 9 if I add it to VGA also). Easy to map to buttons 0-9 as long as there won't be over 10 modes

So if I understand correctly, this will soon be able to accept a 480p component video signal through the scart input? I ask because I'm in the process of getting an extron switch set up, and while it can pass signals other than rgb, the cable output will be a scart plug. So this might be just what I need, as I don't think my xrgb mini would accept ypbpr through the scart input, I'd have to convert the signal first.

[NJRoadfan]

Any plans for a US based distributor to sell assembled units?