OSSC (DIY video digitizer & scandoubler)

[bobrocks95]

Looking at rtings' database (http://www.rtings.com/tv/tests/inputs/input-lag) shows that results are all over the place.

Sets that are slower with 1080p sources (compared to 4K):

• LG UH7700
  Vizio E Series
  Sony Z9D
  Sony X930E

Sets that are faster with 1080p sources:

• Vizio P Series
  Vizio M Series
  LG B6
  Sony X900E/C
  Sony X850D

I don't think manufacturers have any clue what they're doing in regards to lag lol. Yet most all companies nail it with PC monitors right? Do their PC display teams and their TV teams not say a word to each other?

At the most, one millisecond. Usually a fraction.

My OP on the subject said no significant lag difference.

Ok. It's more than a decile of a millisecond on a few of them. Then again, the ones I saw with that much gap had almost two frames. Slow display gonna be slow.

Even a full millisecond isn't going to be significant. I could lock you in a room with two tv's for a week and ask you to find the one millisecond slower display. You would be guessing.

This isn't a thing. It's about image quality, the lag difference is negligible. Slow displays are always slow and the fast ones are always fast.

Did you look at the link??? Some TVs in there have nearly a 20ms lag difference. Sony Z9D 41.3ms vs 24.5ms. If you look at the 4:4:4 category there's even more discrepancies.

[orange808]

Looking at rtings' database (http://www.rtings.com/tv/tests/inputs/input-lag) shows that results are all over the place.

Sets that are slower with 1080p sources (compared to 4K):

• LG UH7700
  Vizio E Series
  Sony Z9D
  Sony X930E

Sets that are faster with 1080p sources:

• Vizio P Series
  Vizio M Series
  LG B6
  Sony X900E/C
  Sony X850D

I don't think manufacturers have any clue what they're doing in regards to lag lol. Yet most all companies nail it with PC monitors right? Do their PC display teams and their TV teams not say a word to each other?

At the most, one millisecond. Usually a fraction.

My OP on the subject said no significant lag difference.

Ok. It's more than a decile of a millisecond on a few of them. Then again, the ones I saw with that much gap had almost two frames. Slow display gonna be slow.

Even a full millisecond isn't going to be significant. I could lock you in a room with two tv's for a week and ask you to find the one millisecond slower display. You would be guessing.

This isn't a thing. It's about image quality, the lag difference is negligible. Slow displays are always slow and the fast ones are always fast.

Did you look at the link??? Some TVs in there have nearly a 20ms lag difference. Sony Z9D 41.3ms vs 24.5ms. If you look at the 4:4:4 category there's even more discrepancies.

Yep. You sorta found one. But, it isn't penalizing low res: 720p/480p. It's anything below native resolution.

Seems to have a fixed processing chain for all upscaling. It's a bad and slow upscaling display.

Pintos explode. Should I insulate the gas tank in every automobile?

The OSSC still performs about the same no matter what. Still no need to panic about 720p or 480p.

[bobrocks95]

You're shifting the goalposts and that's far from the only example in rtings' database.

[orange808]

You're shifting the goalposts and that's far from the only example in rtings' database.

OSSC doesn't do 4k, bro.

It doesn't matter. It rarely will. 480p, 960p, 720p, 1080p... not a common concern.

That expensve Sony is an outlier. Poor bastards that buy a Sony Z9D will need a DVDO vp50pro and an HDFury Linker to get respectable lag with the OSSC. That's not normal.

[bobrocks95]

You still haven't looked through the list, huh?

[orange808]

You still haven't looked through the list, huh?

Given that the basis my argument is that scaling 1080p, 960p, 720p, and 480p takes about the same amount of time, I don't see what that chart has to do with anything. There's no 720p column.

OSSC doesn't do 4k. That column means jack nothing.

Given that I am the one that walked through Best Buy and used my 720p Leo Bodnar device, before I bought my tv last year, I think you're talking out of your behind.

Getting 3x, 4x, or 5x working isn't going to unleash a magic beast in your display and cut the lag. It just won't.

[bobrocks95]

The myth won't die because we've come full circle and now have some 4K TVs that are indeed adding lag when scaling low-res sources.

I would be fascinated to read about that.

All I have seen to date is 4k displays that--actually--benefit from (go faster) at lower resolutions. Given that information is being repeated in a predictable way, it makes sense that a proper algorithm would save time by not performing redundant processing; so, it only shocked me for a moment.

But, I am interested if there are new displays that struggle with scaling.

You're shifting the goalposts and that's far from the only example in rtings' database.

[orange808]

The myth won't die because we've come full circle and now have some 4K TVs that are indeed adding lag when scaling low-res sources.

I would be fascinated to read about that.

All I have seen to date is 4k displays that--actually--benefit from (go faster) at lower resolutions. Given that information is being repeated in a predictable way, it makes sense that a proper algorithm would save time by not performing redundant processing; so, it only shocked me for a moment.

But, I am interested if there are new displays that struggle with scaling.

You're shifting the goalposts and that's far from the only example in rtings' database.

Is that my OP?

No. It isn't.

This is.

On the subject of lag, I have seen some folks in the community (still) claiming that: scaling lower resolutions (480p or 720p, etc.) on new displays will add significant lag. The Leo Bodnar devices have busted that myth. We really shouldn't be spreading it, anymore.

There's no 4k output option on an OSSC or Framemeister. It doesn't matter.

[bobrocks95]

You were interested in new displays that struggled with scaling and I gave you a list of them.

We're off-topic.

[headlesshobbs]

Yes and I'm seeing an argument that's only getting a bit hostile.

[RGB0b]

On the subject of lag, I have seen some folks in the community (still) claiming that: scaling lower resolutions (480p or 720p, etc.) on new displays will add significant lag. The Leo Bodnar devices have busted that myth. We really shouldn't be spreading it, anymore.

I still see questions out here on the internets about it. People ask if using different output modes from the OSSC will reduce lag. In reality, the display is only adding about a decile of a millisecond (at the most) to handle scaling 480p or 720p.

I was watching a video review at MLIG and there was a big warning about scaling 720p adding lag--and it rubbed me the wrong way. That's rubbish.

To tie this back to the OSSC: there's no real lag advantage to the different output modes of the OSSC. It's just image quality. Slow displays are always slow and fast displays are always fast.

Sorry for the rant, but this myth just won't die.

Well, my display was consistently faster on the HDMI port via 480p then component video via 240p. We're talking microseconds here and probably imperceptible when playing, but I have the videos to prove it. The only information I'm "spreading" has proof to back it up.

[RGB0b]

My DC+Kuro combo has a bit of noise (especially at 15kHz), but after setting correct sampling rate and optimal phase, it's barely noticeable.

What's the correct sampling rate and optimal phase to use?

[Harrumph]

On the subject of lag, I have seen some folks in the community (still) claiming that: scaling lower resolutions (480p or 720p, etc.) on new displays will add significant lag. The Leo Bodnar devices have busted that myth. We really shouldn't be spreading it, anymore.

While I agree with you, I don't see how RetroRGB's post has anything to do with "spreading myths". You might have misunderstood his post. The point was to check the lag of the generic SCART to HDMI scaler, not the lag between 480/720/1080p input. In fact, his finding that setting the scaler output to 720p or 1080p has no impact on lag directly supports your point. So I don't know where your accusations are coming from.
The reason he is even doing this in the first place is that the TV (and iirc, his (previous?) capture card) doesn't accept the x3 (and above) modes direct from the OSSC.

[orange808]

For clarity, I was referencing connecting the OSSC directly to a display.

It's still true this morning. On most displays, there's no significant latency benefit or penalty for OSSC output resolutions.

[Gunstar]

Very informative info here, I was one of those people wrongly under impression that the closer (but not native) in res you are to the screen res on a flat screen monitor/tv then the less lag you would have but that doesn't seem to be the case at all, consider me schooled.

^^
Just curious but is the OSSC PSU or the console PSU that was problematic?

I think Xyga was talking about the OSSC PSU.

[Xyga]

I think Xyga was talking about the OSSC PSU.

Yup!

[CobraKing]

@Gunstar @Xyga

Wow surprised to hear that OSSC PSU is problematic. Do you think yours was just a one off?

Would you also mind sharing what you replaced it with? Thanks.

[Xyga]

There might be a little misunderstanding there; it was just one of the many small PSUs I have around, not one I've ordered along with the OSSC.

Tested it and it was too weak (probably shitty old caps), picked another one that was fine: problems gone.

[CobraKing]

@Xgya

Perfect thanks for clarifying. I got the 'Official' OSSC PSU that you can order it with.

Haven't noticed any issues apart from my old Sony not handling X3,X4 & X5. Grrr.

[headlesshobbs]

I was thinking about wanting to bypass the need for vga altogether and going the scart route with my Dreamcast, but I really want to keep my 480p signals. While I do have a spare vga to scart cable, it seems defective and I tried using it with other equipment to no success.

Will the ossc accept 480p if I go this route and I'm wondering where I can get a (working) replacement cable?

[marqs]

My DC+Kuro combo has a bit of noise (especially at 15kHz), but after setting correct sampling rate and optimal phase, it's barely noticeable.

Ok just to keep me on the right path, under which setting is this under? I have a pretty decent VGA cable as it is, so I thought I'd do another check into this when I get time later.

Sampling opt -> 480p in sampler. Set that to DTV-480p if you use DC via VGA. If your display is able to resize output to match active area, set additionally H.active to 640 in Adv. timing.

Will the ossc accept 480p if I go this route and I'm wondering where I can get a (working) replacement cable?

Yeah, you can feed it all the way up to 1080p via scart.

[headlesshobbs]

That sounds awesome! I didn't know it could support that much.

Anyway I think I found out what my cable's problem was. http://tinypic.com/view.php?pic=vxl1g8&s=9

[ZellSF]

Not sure if you saw it since it got covered up in another discussion marqs, so again: sorry for being impatient, but any chance of getting a bugfix release for PS2 480p YUV sometimes being detected as 960i soon? And maybe higher sidemask brightness settings?

Those are the only things missing that would make the OSSC perfect for me. Other than things that can't be done because of hardware limitations of course.

[Xyga]

One thing too bad about x5 mode in 1080p is that it ends up letterboxed, the lines intended to be cut actually work as a mask/black borders on top/bottom.
The visible source area appears therefore squished and the integer is destroyed by the display's scaling.

If I understand correctly it is not possible to actually leave a number of active lines hidden out in the overscan area beyond 1080 ?
Were it possible all the displays that lack a 1:1 mode could then display a real perfectly integer-scaled source, bypassing the display's own scaling.

[paulb_nl]

If your display respects the V.active settting then it should only be showing 1080 lines in linex5 mode. On my monitor its perfect without letterboxing.

[marqs]

Not sure if you saw it since it got covered up in another discussion marqs, so again: sorry for being impatient, but any chance of getting a bugfix release for PS2 480p YUV sometimes being detected as 960i soon? And maybe higher sidemask brightness settings?
Those are the only things missing that would make the OSSC perfect for me. Other than things that can't be done because of hardware limitations of course.

I noticed that:

Yeah, those are on top the todo list. I'll need to tweak the firmware anyway in a couple weeks (when v1.6 prototypes arrive) so they could be fixed simultaneously.

When your PS2 480p gets detected as 960i, is VSM reported as 1 (press info on remote)? I haven't bumped into this issue personally but I can guess the root cause.

One thing too bad about x5 mode in 1080p is that it ends up letterboxed, the lines intended to be cut actually work as a mask/black borders on top/bottom.
The visible source area appears therefore squished and the integer is destroyed by the display's scaling.

If I understand correctly it is not possible to actually leave a number of active lines hidden out in the overscan area beyond 1080 ?
Were it possible all the displays that lack a 1:1 mode could then display a real perfectly integer-scaled source, bypassing the display's own scaling.

It shouldn't get letterboxed but cut out instead if the display correctly processes active area information embedded in the signal.

[Xyga]

^ It doesn't on any of my displays, not even through a DVDO. Am I doing something wrong?

Starting from cleared settings with an MD2@60Hz I set 5x, generic 4:3, 1080p mode...

PS: same whether in DVI or HDMI mode also.

EDIT/up: turning out be much more complicated than I thought it would. So far none of the full-hd displays I've tried (2 monitors and 2 TVs) were able to display either x4 or x5 at all or properly, especially the latter is still bugging me as it's always showing those black bars top and bottom, even though all displays see a 1080p signal.
Same with the VP30.
More surprisingly the VP50 Pro won't show anything beyond x2, I've tried tweaking the sampling options including what's recommended on the wiki for x5, but still nothing.
Spent several hours trying to find working settings for the displays and VPs with no success.

[ZellSF]

Not sure if you saw it since it got covered up in another discussion marqs, so again: sorry for being impatient, but any chance of getting a bugfix release for PS2 480p YUV sometimes being detected as 960i soon? And maybe higher sidemask brightness settings?
Those are the only things missing that would make the OSSC perfect for me. Other than things that can't be done because of hardware limitations of course.

I noticed that:

Yeah, those are on top the todo list. I'll need to tweak the firmware anyway in a couple weeks (when v1.6 prototypes arrive) so they could be fixed simultaneously.

When your PS2 480p gets detected as 960i, is VSM reported as 1 (press info on remote)? I haven't bumped into this issue personally but I can guess the root cause.

VSM=1 when this problem occurs, yes.

EDIT/up: turning out be much more complicated than I thought it would. So far none of the full-hd displays I've tried (2 monitors and 2 TVs) were able to display either x4 or x5 at all or properly, especially the latter is still bugging me as it's always showing those black bars top and bottom, even though all displays see a 1080p signal.
Same with the VP30.
More surprisingly the VP50 Pro won't show anything beyond x2, I've tried tweaking the sampling options including what's recommended on the wiki for x5, but still nothing.
Spent several hours trying to find working settings for the displays and VPs with no success.

Tried the usual? Making sure cadence detection is disabled (remember it's a per resolution setting), HDCP is disabled and that you have a decent quality HDMI cable (5x mode is very picky when it comes to cables)?

There is one more setting that might have an influence, I've no idea what it is though, before I had my setup working correctly I had to factory reset my VP50 twice because it just would not sync on 3x. Try another HDMI input if you don't want to factory reset (since settings are per input on DVDO VPs).

This is what you should get with 240pX5 on a Genesis:
http://i.cubeupload.com/vBhRBP.png
That capture is from my VP50, but I've heard the VP50 Pro works the same. Yes I know the jailbars on my Genesis are terrible .

[Xyga]

^ Yes I did all the usual checks including factory reset and I'm confident about at least the monitor return cable.

The one going from the OSSC to the DVDO though uses a DVI cable + DVI>HDMI adapter

I'll be surprised if I find later that they're to blame for that letterboxed output...

Anyway reading VGP forums earlier today I realize there's many situations where it can go wrong and it is necessary to take enough time to really explore everything.

[bateman82]

Two questions about compatibility (at least X3):

1)Pioneer Kuro Lx5090h

2)Sony KDL-42W653A (model Fudoh reviewed).