OSSC (DIY video digitizer & scandoubler)

[ZellSF]

Have you tried the OSSC on a limited range only display? Really curious about that.

[BuckoA51]

Not under the kind of intense scrutiny that would be required to notice any issues, maybe Marqs can comment on that?

[marqs]

Have you tried the OSSC on a limited range only display? Really curious about that.

Does there really exist a display that is limited-range only, even in DVI mode? Even the most lacking displays (regarding RGB range) I've seen can still be forced to full-range mode by using a Infoframe flag in HDMI signal, which OSSC does.

[bobrocks95]

Have you tried the OSSC on a limited range only display? Really curious about that.

Does there really exist a display that is limited-range only, even in DVI mode? Even the most lacking displays (regarding RGB range) I've seen can still be forced to full-range mode by using a Infoframe flag in HDMI signal, which OSSC does.

Would not be surprised if my 2005 Panasonic plasma locks the RGB range to limited (DVI and HDMI inputs are on separate input boards you plug in), but I have no clue how to test for that without having a baseline brightness setting to go off of.

[marqs]

I've made some updates on the firmware, including the following:

* Improved remote control handling code. Now it's possble to hold down key to traverse menu/ change values faster
* Fixed occasional mode change loop when switching to a non-interlace mode utilizing odd-field sync signal
* Fixed randomly missing blue channel at power-on when using DVI output mode

I'll release the v0.69 image tomorrow after some more testing.

There's one bug, though, that escaped all testers until borti4938 found it: left and right channels in 3.5mm audio output connector are swapped. Fortunately most are probably planning to use 3.5mm-2xRCA cable if using the connector at all so it shouldn't be a big issue. However, if it's a deal breaker for someone who has reservation on the first batch, please contact me. Future pre-assembled boards and DIY kits will have naturally that fixed.

[ZellSF]

Hmm... I need to test my Panasonic plasma again. I didn't get full range working over HDMI, but maybe if I start out with a DVI signal it will work. But then I'll have no excuse to replace it

I don't care about audio, so the (hilarious) channel mistake doesn't bother me. I just want an OSSC. Now

[lettuce]

I've made some updates on the firmware, including the following:

* Improved remote control handling code. Now it's possble to hold down key to traverse menu/ change values faster
* Fixed occasional mode change loop when switching to a non-interlace mode utilizing odd-field sync signal
* Fixed randomly missing blue channel at power-on when using DVI output mode

I'll release the v0.69 image tomorrow after some more testing.

There's one bug, though, that escaped all testers until borti4938 found it: left and right channels in 3.5mm audio output connector are swapped. Fortunately most are probably planning to use 3.5mm-2xRCA cable if using the connector at all so it shouldn't be a big issue. However, if it's a deal breaker for someone who has reservation on the first batch, please contact me. Future pre-assembled boards and DIY kits will have naturally that fixed.

What date did you have to register your interest to be in the first batch?, im not sure if im first batch or not!

[Shuco13]

Has anybody done a direct lag comparison between the OSSC and the Framemeister yet using the same equipment? I couldn't find a single review that stated to have performed a real lag test with this device. Thanks in advance!

[Fudoh]

why do you want to see a lag test on a unit with a lag-free design ? On my tests the FM got 24.5ms of delay on a CRT (same tests that ranked the XRGB-3 in B1 mode at less than 2ms). That makes the OSSC 1.5 frames faster than the FM.

[Shuco13]

Just out of couriosity and I was surprised nobody has done it so far. It would be especially neat to see some results from a plasma/LCD screen since that's what most people here use. Would be nice to see the average difference between Framemeister outputting on 1080p vs OSSC outputting on 480p/720p.

[blizzz]

It would be especially neat to see some results from a plasma/LCD screen since that's what most people here use.

You always have to add the lag from your own TV/monitor to the lag of the device, but that extra delay wouldn't be part of a review.

[Xyga]

Yes.
Testing the OSSC itself for lag is not pertinent.
As Fudoh says by design the OSSC will produce barely any measurable lag, like the old XRGB series (hell even less in cases from what I've read, which means timings under 1ms lol).
Then at the other end each flat panel display, whether it's lcd, plasma or oled, indeed produces its own lag depending on its built-in scaler performance and settings.
The upscaling factor, as well as the panel tech will play a practically insignificant part in the final total lag measurement.
In other words like with the old XRGBs it's only the performance of your display's internal electronics that will make a difference in lag and picture quality.
Other devices you put in-between the OSSC and display though, may or may not influence the signal and therefore how the display will react, possibly changing the results, but that's a different topic.

[Shuco13]

It would be especially neat to see some results from a plasma/LCD screen since that's what most people here use.

You always have to add the lag from your own TV/monitor to the lag of the device, but that extra delay wouldn't be part of a review.

That's why I said AVERAGE but I understand that nobody is willing to test this with numerous sets. Hooking a Framemeister to a CRT to test input lag doesn't make sense to me either.

[Xyga]

No need to bother there are many display lag tests available, today pc monitors measure between 0ms and 25ms depending on their category and equipment.
Full-Hd very often around 1/4 or 1/2 of a frame, WQHD close to zero if they're scaler-less or purposedly aimed at gamers commercially speaking, or around 20ms for more general purpose ones where they didn't care. 4K are a mixed bag in the same area.
Decent TVs are typically somewhere between 20ms and 35ms, and the many not really fit for gaming often around 50ms and more up to 100ms in the worst cases.
Since the OSSC is not involved in the production of lag, you've got your answer.

[Shuco13]

No need to bother there are many display lag tests available, today pc monitors measure between 0ms and 25ms depending on their category and equipment.
Full-Hd very often around 1/4 or 1/2 of a frame, WQHD close to zero if they're scaler-less or purposedly aimed at gamers commercially speaking, or around 20ms for more general purpose ones where they didn't care. 4K are a mixed bag in the same area.
Decent TVs are typically somewhere between 20ms and 35ms, and the many not really fit for gaming often around 50ms and more up to 100ms in the worst cases.
Since the OSSC is not involved in the production of lag, you've got your answer.

Ok so considering the average lag/scaling of a 1080p LCD/Plasma the mini and OSSC should be on par. My results for the Framemeister on 1080p in the 240p suite are within the 17-35 ms range which matches exactly the specified lag of my TV.

[Xyga]

No, that's essentially the Mini's lag, with the OSSC you'd get minus this.

[blizzz]

My results for the Framemeister on 1080p in the 240p suite are within the 17-35 ms range which matches exactly the specified lag of my TV.

The manual lag test in the 240p suite will not give you any reliable measurements. Your brain adapts to the lag and makes the result useless.

Lag for the OSSC is 1ms + the lag of your TV, lag for the Framemeister is 25ms + the lag of your TV.

[Xyga]

And I'll add if it's a plasma be careful many models didn't receive a proper camera test.
You didn't tell which is your display, but be careful about 'specified' lag since that basically doesn't exist, only tests done by people with different devices/techniques.

[bobrocks95]

And I'll add if it's a plasma be careful many models didn't receive a proper camera test.
You didn't tell which is your display, but be careful about 'specified' lag since that basically doesn't exist, only tests done by people with different devices/techniques.

What's the preferred method for this? I used a manual shutter speed on my phone's camera and set it as low as it could go while still showing a picture (brightness goes way down with such a short exposure) and saw 5 frames, which certainly isn't what the display feels like. Do plasmas just kind of feel faster than they are because of how they display an image, or did I do the test wrong, or am I really bad at noticing lag?

[Guspaz]

To test lag, I put the 240p test suite in my SNES (using the frame counter test) and connected it to both a lag-free monitor (PVM) and the monitor to be tested, and then take a photo of both screens at the same horizontal level (to account for rolling shutter) at the fastest shutter speed possible. Since the CRT has no lag, the number of frames of lag can be determined simply by looking at the difference in the numbers on the two screens.

This method is only accurate to maybe one frame, maybe half a frame, but anything more accurate than that is really irrelevant to human use anyhow.

Measuring lag more accurately requires a high-speed video recording so that you can observe how much actual time passes between one screen getting a certain update and the other screen. The accuracy would be dependent on how high a framerate you could record at. 240 fps (which I believe is as fast as the iPhone goes) would let you measure lag with an accuracy of around 4ms.

[Shuco13]

Framemeister is 25ms + the lag of your TV.

Which should be arround 50 ms and is simply ridiculous to assume. As for 240p suite, I also think the manual lag test is highly inaccurate but there is no way one could completely compensate the display's lag without wanting to. After numerous tests with a CRT and Plasma that showed constant results and differences there is no way that assumption is correct. I wish I had the means to test this with a camera but unfortunately I don't have dual outputs.

[Guspaz]

You could wire an LED to a game controller and use a game that is known to have no input lag on a CRT (for example, if the character begins jumping the instant the jump button is depressed). Film with the controller at the same horizontal level as the on-screen character (to account for rolling shutter) and press the button. Measure how much time elapses between LED turning on and the character reacting on-screen.

[Xyga]

What's the preferred method for this? I used a manual shutter speed on my phone's camera and set it as low as it could go while still showing a picture (brightness goes way down with such a short exposure) and saw 5 frames, which certainly isn't what the display feels like. Do plasmas just kind of feel faster than they are because of how they display an image, or did I do the test wrong, or am I really bad at noticing lag?

Still SMTT 2.0 afaik, since the LB tester isn't ideal for plasmas. Check the article and interview on tftcentral if you haven't yet.
There were lots of random tests back in the day, just stopwatch pics posted without specifying some essential test parameters and results, often even against lcds lol. Results could vary enough to make the reliability of those pics questionable, like with all flat panels plasma readings suffered from this trend and not many people went back to redo proper tests later when things were more clear. Or they did but using the LB, so in the end actual trustworthy-enough lag figures of many models were never known.

[marqs]

Still SMTT 2.0 afaik, since the LB tester isn't ideal for plasmas. Check the article and interview on tftcentral if you haven't yet.
There were lots of random tests back in the day, just stopwatch pics posted without specifying some essential test parameters and results, often even against lcds lol. Results could vary enough to make the reliability of those pics questionable, like with all flat panels plasma readings suffered from this trend and not many people went back to redo proper tests later when things were more clear. Or they did but using the LB, so in the end actual trustworthy-enough lag figures of many models were never known.

Could you give link to that article? I don't see why it wouldn't be good for measuring plasmas - I used a similar DIY-built device years ago, and was able to measure my plasma without issues. Speaking of lag testing, it'd easy to turn OSSC into a lag tester with a photodiode circuit and a tailored fw

I managed to loan a MD today for testing Sonic 2 interlace in more detail. Found a setting that allowed also my pickier monitor to display the signal without issues. I'll postpone releasing the new FW a few days to get that option accessible via menu.

[Xyga]

The tftcentral article is about the testing methods, not specifically about plasmas, http://www.tftcentral.co.uk/articles/input_lag.htm
SMTT author interview included to explain why it's the preferred camera method.

But the LB vs. plasmas issue has been bugging people in its time, showing readings above what people could feel, hdtvtest then ended up doing both test when reviewing new models to show the discrepancy: http://www.hdtvtest.co.uk/news/input-lag
Blurbusters also expressed their doubts although didn't publish comparative measurements.

Dunno about the way you did it for your plasma, maybe it's better than the LB, anyway I'd be interested in using my OSSC for lag testing yup! ^^

[marqs]

Fw 0.69 is now released with update instructions on the wiki. In addition to the earlier points, the following features are added:

* Finer tuning range for scanline strength and mask
* H-PLL coast options (pre=3, post=3 should be used for Megadrive)

[BuckoA51]

Sweet I'll give this new firmware a test tomorrow.

* H-PLL coast options (pre=3, post=3 should be used for Megadrive)

Is that for Megadrive 480i or just generally?

Think that could help with the SNES on some sets too?

[marqs]

Sweet I'll give this new firmware a test tomorrow.

* H-PLL coast options (pre=3, post=3 should be used for Megadrive)

Is that for Megadrive 480i or just generally?

Think that could help with the SNES on some sets too?

Those values can be used generally (are actually recommended values in datasheet), but so far MD interlace has been the first source I've seen that really needed them for clean output. It's probably good idea to use them always with MD, especially if csync is not used as sync source. However, SNES requires the default values (pre=1, post=0) or otherwise OSSC can't properly sync with it.

[BuckoA51]

Megadrive 480i works with those settings

Frustratingly, so does the SNES, but as you suspected, now the OSSC periodically loses sync with the SNES if you use pre=3, post=3.

Atari Jaguar PAL is fixed now too for the 3 people in the world that cared

[Chacranajxy]

The wait for this thing is killing me right now. My XRGB-3 broke at some point, so I haven't been able to play any old-school stuff in a while... and right now, literally all I want to play is old-school stuff. Not a great combination.

Any word on when we can expect this thing to start shipping?