OSSC (DIY video digitizer & scandoubler)
Re: OSSC (DIY video digitizer & scandoubler)
Been having a weird issue with this new firmware.
I spent all of yesterday converting all of my old profiles with the new site, setting by setting. For whatever reason, my SNES is now extremely noisy when capturing with my Vision E1s in line2x 256x240 optim 8:7.
The first level of SMW has a huge amount of noise and jailbars on the green background right at the start. Honestly, it looks awful.
I've double checked sampling phase, video LPF, reverse LPF, and sync LPF, and it just generally seems worse.
I'll try going back to the old firmware and trying it again when I get a chance, but any ideas? Nothing in my setup has changed at all.
I spent all of yesterday converting all of my old profiles with the new site, setting by setting. For whatever reason, my SNES is now extremely noisy when capturing with my Vision E1s in line2x 256x240 optim 8:7.
The first level of SMW has a huge amount of noise and jailbars on the green background right at the start. Honestly, it looks awful.
I've double checked sampling phase, video LPF, reverse LPF, and sync LPF, and it just generally seems worse.
I'll try going back to the old firmware and trying it again when I get a chance, but any ideas? Nothing in my setup has changed at all.
Re: OSSC (DIY video digitizer & scandoubler)
Two things to check:Jademalo wrote:
I'll try going back to the old firmware and trying it again when I get a chance, but any ideas? Nothing in my setup has changed at all.
E1S might be defaulting to 5-5-5 color space. It can make your OSSC source image look incredibly noisy, and you have to set it back to 8-8-8.
If that's not it, check H. samplerate just to make sure you have the right value. Use the 240p Test Suite checkerboard pattern to make sure.
-FBX
Re: OSSC (DIY video digitizer & scandoubler)
I'm afraid generic mode does not allow the same trick, because you don't get to set optimal timing for it (as in H. samplerate). With what I was describing, you can optimally time the 1024 by setting the H. active and H. samplerate to perfectly land on each pixel for the background plane, and then the foreground's slight shift will have 4 point samples per pixel to help smooth out the misalignment.Harrumph wrote:Maybe I’m misunderstanding you, but it seems to me you can just use generic mode, because the OSSC could not do any pixel repetition in such a mode. Think about it (example for 4x mode): 256x240 opt mode (in ”8:7” mode) pixel repeats 4 times per sample, 512x240 pixel repeats 2 times, thus 1024x240 would pixel repeat once, ie same as input, ie same as generic mode.FBX wrote: In other news: I'd like to request another feature in the optimal timing modes of the OSSC: 1024x240.
...
Now if we had access to 1024x240 optimal timing for the LineX menus, I would then have 4 sample points on the horizontal per pixel instead of just two. It would make this trick even better looking than it already is now.
Just use a 4x multiple of the optimized sample rate in generic mode, and you’d achieve the same thing. Afaik (and I might be wrong) sample phase does have effect even in generic mode, just that normally it doesnt make much difference, but in this case you would need to adjust it to achieve perfect sampling.
Keep in mind what I'm doing is an advanced trick that most people won't understand even when I try to explain it. Generic mode does not allow the same trick that I am doing with 512x240 on a 320x240 image (and before people chime in about 512 not being a double of 320, remember the H. active function lets you extend 512 to 640 in the optimal timing menu).
Re: OSSC (DIY video digitizer & scandoubler)
FBX wrote:Two things to check:
E1S might be defaulting to 5-5-5 color space. It can make your OSSC source image look incredibly noisy, and you have to set it back to 8-8-8.
If that's not it, check H. samplerate just to make sure you have the right value. Use the 240p Test Suite checkerboard pattern to make sure.
-FBX
Awesome, totally forgot about 5-6-5 and 8-8-8.
I was using the window to quickly check rather than OBS, and that has 5-6-5 by default. Setting it back to 8-8-8 helped a ton. Phew, I thought I was going crazy =p
It's still not perfect, I've got perfectly vertical bars across the whole image, but I think that's the ghosting capacitor issue. I really need to get around to getting Voultar's amp kit and that cap...
Thanks!
Re: OSSC (DIY video digitizer & scandoubler)
Here is a picture from a bvm (capture from one of phonedork´s videos)Fudoh wrote:looks terrible for me. I have been battling this kind of scanline rendition on quite a few emus in the past. Why would you prefer that ?For the near future I would really like for the OSSC being able to make the 1st and 5th line the scanline.
Here you can see the "color scanline" goes from thin to thick in both directions (up and down) depending on brightness. With the OSSC you don´t get this effect, you can have the hybrid effect above (top alignment) of below (bottom alignment), but not in both directions.
Anyway, even with the "split" mode I mention before, it wouldn't be the same. On the yoshi picture you still get a dark portion between scanlines, even on the parts where the color blooms. On the OSSC you might get this (I'm just speculating here) with an option that allows something like this:
5x:
5th row: dark
4th row: hybrid
3rd row: regular
2nd row: regular
1st row:hybrid
maybe soon? the OSSC wiki says a custom scanline option is on the to do list, but I understand the memory is full and it´s hard (or impossible) to add more options.
Re: OSSC (DIY video digitizer & scandoubler)
Have you tried it ? You can create these kind of scanlines in many emus (Magic Engine for example). Doesn't look good. And I get the idea in regard to the BVM example - believe me.
Re: OSSC (DIY video digitizer & scandoubler)
v0.82 has zero-riscy core running the housekeeping/control tasks. There's now over 8kB free space which enables implementing some features like fully customizable scanlines and OSD, but it's not enough for anything that requires extra line buffers.awe444 wrote:I’m curious about this as well. As of this new update, how much free space is left?maxtherabbit wrote: is this version using the new CPU core?
Re: OSSC (DIY video digitizer & scandoubler)
I disagree, I think the 1+5 line pattern looks great and better approximates the symmetric flux profile of a real scanline pattern.Fudoh wrote:looks terrible for me. I have been battling this kind of scanline rendition on quite a few emus in the past. Why would you prefer that ?For the near future I would really like for the OSSC being able to make the 1st and 5th line the scanline.
-
- Posts: 386
- Joined: Sun Jul 04, 2010 11:14 pm
Re: OSSC (DIY video digitizer & scandoubler)
I have been testing vertical scanlines with an external SLG to get this low line count filter to work on lcd screens and would want to see if you or someone else can code this into the ossc to make it work.marqs wrote:v0.82 has zero-riscy core running the housekeeping/control tasks. There's now over 8kB free space which enables implementing some features like fully customizable scanlines and OSD, but it's not enough for anything that requires extra line buffers.
Here's some image results.
https://ibb.co/fShy5L
https://ibb.co/cTb85L
Vertical scanlines are running in 960p, so all you have to do is run both horizontal and vertical at the same time. Also this means you'll have to figure a way to add scanline strength and hybrid adjustments to them independently, but I have faith this can be done easily enough.
This works better then trying to play some games that are overly pixelated on a some 900 line count monitor as it shows them the way they're supposed to be viewed in the first place. Fun part of this is when you get the right resolution for some games, the line count will do an exact line per pixel matchup (or x2 lines depending on resolution, sampling option), so it's all perfectly even rather then misaligned like most older crts.
The problem is most people are doing this on computer lcd's and that means you have no control over vertical sharpness settings. You need vertical sharpness to get the blend effect right or play it on an hdtv that has that option.Fudoh wrote:Have you tried it ? You can create these kind of scanlines in many emus (Magic Engine for example). Doesn't look good. And I get the idea in regard to the BVM example - believe me.
"Don't HD my SD!!"
-
NoAffinity
- Posts: 1033
- Joined: Mon May 07, 2018 5:27 pm
- Location: Escondido, CA, USA
Re: OSSC (DIY video digitizer & scandoubler)
Hey FBX, can you post the JSON code from this file? I uploaded the settings, and they work great. All consoles that I'm using (most of them) are spot on with the settings you have included (and thank you for providing this file! ). However, I have a couple I'd like to add/delete. For instance, I don't have a Sega MS, and would like to replace that profile with one for CPS1/CPS2.FBX wrote:I made the first set of import profiles for the OSSC with custom labels (sans the last slot as I didn't use it). Please note that I found in my particular case I have to make Win32DiskImager write the file TWICE to the card before my OSSC will recognize the data (at least for firmware updates). So I always do this for profile imports as well. Also please note the 32X profile is tied to HD Retrovision cables, while the others are tied to RGBS SCART input.
As usual, sampling phase, H. backporch, and V. backporch are random variables on a console/cable case-by-case basis, so you may still need to adjust these for your given setup. Here's the import file link:
http://www.firebrandx.com/downloads/oss ... es_fbx.zip
Simply go into the settings option and look for the import function.
Cheers!
-FBX
Granted, I can simply change the settings through the OSSC's menu and then save them; but I don't believe there is a way to change the name of the profile via the OSSC menu, and I'm just trying to keep things tidy.
Maybe there's another way to do this, but I'm not seeing it.
If possible to post the JSON code, I think I can solve my problem with that. Much appreciated!
Re: OSSC (DIY video digitizer & scandoubler)
Sorry about that. I posted the download on Twitter, but forgot to do so here:NoAffinity wrote:
If possible to post the JSON code, I think I can solve my problem with that. Much appreciated!
https://t.co/ErshCF946T
Just copy and paste, then edit labels and attributes as you need.
-
NoAffinity
- Posts: 1033
- Joined: Mon May 07, 2018 5:27 pm
- Location: Escondido, CA, USA
Re: OSSC (DIY video digitizer & scandoubler)
Perfect, exactly what I needed. Thank you again for putting these together!FBX wrote:Sorry about that. I posted the download on Twitter, but forgot to do so here:NoAffinity wrote:
If possible to post the JSON code, I think I can solve my problem with that. Much appreciated!
https://t.co/ErshCF946T
Just copy and paste, then edit labels and attributes as you need.
Re: OSSC (DIY video digitizer & scandoubler)
thanks for these.FBX wrote:Sorry about that. I posted the download on Twitter, but forgot to do so here:NoAffinity wrote:
If possible to post the JSON code, I think I can solve my problem with that. Much appreciated!
https://t.co/ErshCF946T
Just copy and paste, then edit labels and attributes as you need.
do you have any profiles for 5x 1920x1080 as well? Can't seem to dial in the H. samplerate perfectly. Maybe a multiple of 427 that works with 4x? Will any other settings need to be reconfigured if using 5x?
I notice in your profiles you keep the gains at 26 but in your youtube video you up it to 40. Any pro/con to doing this?
Re: OSSC (DIY video digitizer & scandoubler)
Correct, I do not understand what you are going for in this case lol.FBX wrote: people won't understand even when I try to explain it.
Re: OSSC (DIY video digitizer & scandoubler)
Why would you not be able to set optimal timing for generic mode?FBX wrote: I'm afraid generic mode does not allow the same trick, because you don't get to set optimal timing for it (as in H. samplerate).
A 1024x240 mode would not be different to generic mode with equal timing settings.
Re: OSSC (DIY video digitizer & scandoubler)
All I know is when you go to advanced timing while in Generic 4:3 mode, it's not hooked to a specific resolution in that list and so nothing lets you adjust the settings for it. I mean, I could be wrong, but nothing I tried seemed to be hooked into Generic mode.paulb_nl wrote:Why would you not be able to set optimal timing for generic mode?FBX wrote: I'm afraid generic mode does not allow the same trick, because you don't get to set optimal timing for it (as in H. samplerate).
A 1024x240 mode would not be different to generic mode with equal timing settings.
Yeah I got that the moment I read your first reply. It's MENSA-level OSSC settings work I'm doing, and I know exactly why I'd need a 1024x240 optimal timing mode in order to make it more effective.Harrumph wrote:
Correct, I do not understand what you are going for in this case lol.
Re: OSSC (DIY video digitizer & scandoubler)
...it’s not hooked to a specific resolution...
EDIT: reading paulb_nls answer below I finally understand what you meant by "hooked to a specific resolution". So this was all based on a misunderstanding from the beginning. Of course you can adjust h.samplerate and all the rest in generic mode. Put my old reply in spoilers.
Spoiler
Neither is optimized mode really, it’s just pixel repetition which is different between 256x240 and 320x240 mode. Still very intrigued how you envision pixel repetition would work in the 1024x240 mode you propose. Or you mean some kind of interpolation of the 3 ”in between pixels” into one pixel?
Just curious.
And I would maintain, if there is no pixel repetition then, like paulb_nl seems to agree, there is no practical difference from generic mode.
Just curious.
And I would maintain, if there is no pixel repetition then, like paulb_nl seems to agree, there is no practical difference from generic mode.
Last edited by Harrumph on Tue Nov 20, 2018 8:08 am, edited 3 times in total.
Re: OSSC (DIY video digitizer & scandoubler)
I have changed the timings for Generic modes all the time. The OSSC automatically chooses the right resolution in the list but for NTSC 240p the resolutions are:
Passtrough/Linex2: 240p
Linex3: 4:3 960x240, 16:9 1280x240
Linex4: 1280x240
Linex5: 1600x240
You do need to turn off Allow TVP HPLL 2x to keep a stable phase when setting optimal timings in Generic modes.
Passtrough/Linex2: 240p
Linex3: 4:3 960x240, 16:9 1280x240
Linex4: 1280x240
Linex5: 1600x240
You do need to turn off Allow TVP HPLL 2x to keep a stable phase when setting optimal timings in Generic modes.
-
- Posts: 2187
- Joined: Mon Aug 14, 2017 8:34 pm
Re: OSSC (DIY video digitizer & scandoubler)
The varying thickness of scanlines that you see there can be produced by the OSSC even without hybrid scanlines, but I think it's really up to how your TV/monitor handles or upscales the signal it gets from the OSSC. On my LG OLED the scanlines vary in thickness depending on the brightness of the color displayed, surprisingly like on a CRT (though less dramatic for darker colors and more rounded/less pointy at the edges), BUT only when using x3 i.e. when upscaling 720p to 4k (generic, I didn't try any optimized settings) and on non-PC mode. Any other combination resulted in the fake-looking uniform scanlines that seem to be the norm.naz wrote:Here is a picture from a bvm (capture from one of phonedork´s videos)Fudoh wrote:looks terrible for me. I have been battling this kind of scanline rendition on quite a few emus in the past. Why would you prefer that ?For the near future I would really like for the OSSC being able to make the 1st and 5th line the scanline.
Here you can see the "color scanline" goes from thin to thick in both directions (up and down) depending on brightness. With the OSSC you don´t get this effect, you can have the hybrid effect above (top alignment) of below (bottom alignment), but not in both directions.
Anyway, even with the "split" mode I mention before, it wouldn't be the same. On the yoshi picture you still get a dark portion between scanlines, even on the parts where the color blooms. On the OSSC you might get this (I'm just speculating here) with an option that allows something like this:
5x:
5th row: dark
4th row: hybrid
3rd row: regular
2nd row: regular
1st row:hybrid
maybe soon? the OSSC wiki says a custom scanline option is on the to do list, but I understand the memory is full and it´s hard (or impossible) to add more options.
I found this using 100% scanlines on the OSSC. Adding any preferred percentage of hybrid scanlines will add to the effect. At the time I thought that I had finally found a replacement for CRT, but of course that didn't last and now my OSSC is just sitting there looking pretty. Still, it's nice to have as a backup.
-
bobrocks95
- Posts: 3472
- Joined: Mon Apr 30, 2012 2:27 am
- Location: Kentucky
Re: OSSC (DIY video digitizer & scandoubler)
What was the nail in the coffin for you?fernan1234 wrote:I found this using 100% scanlines on the OSSC. Adding any preferred percentage of hybrid scanlines will add to the effect. At the time I thought that I had finally found a replacement for CRT, but of course that didn't last and now my OSSC is just sitting there looking pretty. Still, it's nice to have as a backup.
PS1 Disc-Based Game ID BIOS patch for MemCard Pro and SD2PSX automatic VMC switching.
-
- Posts: 386
- Joined: Sun Jul 04, 2010 11:14 pm
Re: OSSC (DIY video digitizer & scandoubler)
Maybe my previous post can change his mind?
I do get differing levels of scanline thickness, but you really have to rely on your tv's scaler and settings to make that effect happen.
I do get differing levels of scanline thickness, but you really have to rely on your tv's scaler and settings to make that effect happen.
"Don't HD my SD!!"
Re: OSSC (DIY video digitizer & scandoubler)
Indeed, depending on the next scaler in the chain (built-in or external) and your display's properties you can get scanlines varying in thickness.
In x5 mode over a 1080 or 1200 display though (1:1 pixel mapping) I can only get that with the help of an external scaler in the chain (VP50 Pro in particular for me since it features workable fine control over both the scaling and interpolation quality even in 1:1 situations)
Guess with OSSC hybrid scanlines + fitting scaler (built-in or external) + OLED's natural per-pixel brightness properties, you get the best non-emulation combination for simulating that effect.
Over an LCD it's only half-convincing because you get the shape but not really the live brightness variation only self-emitting displays can actually produce.
In x5 mode over a 1080 or 1200 display though (1:1 pixel mapping) I can only get that with the help of an external scaler in the chain (VP50 Pro in particular for me since it features workable fine control over both the scaling and interpolation quality even in 1:1 situations)
Guess with OSSC hybrid scanlines + fitting scaler (built-in or external) + OLED's natural per-pixel brightness properties, you get the best non-emulation combination for simulating that effect.
Over an LCD it's only half-convincing because you get the shape but not really the live brightness variation only self-emitting displays can actually produce.
Strikers1945guy wrote:"Do we....eat chicken balls?!"
Re: OSSC (DIY video digitizer & scandoubler)
Didn't have the chance to try the new firmware yet, so I'm asking out of curiosity: is it possible to name the profiles on the OSSC as well, or is that only doable through the web app?
Re: OSSC (DIY video digitizer & scandoubler)
Okay I will look into this and confirm if it works on my end. If 1280x240 for 4x is in fact hooked to Generic mode, this means I can make a splendid 32x profile. I'll post my results after looking into this.paulb_nl wrote:I have changed the timings for Generic modes all the time. The OSSC automatically chooses the right resolution in the list but for NTSC 240p the resolutions are:
Passtrough/Linex2: 240p
Linex3: 4:3 960x240, 16:9 1280x240
Linex4: 1280x240
Linex5: 1600x240
You do need to turn off Allow TVP HPLL 2x to keep a stable phase when setting optimal timings in Generic modes.
-
- Posts: 2187
- Joined: Mon Aug 14, 2017 8:34 pm
Re: OSSC (DIY video digitizer & scandoubler)
Motion persistence inherent to flat panel display technology was the main issue.bobrocks95 wrote:What was the nail in the coffin for you?fernan1234 wrote:I found this using 100% scanlines on the OSSC. Adding any preferred percentage of hybrid scanlines will add to the effect. At the time I thought that I had finally found a replacement for CRT, but of course that didn't last and now my OSSC is just sitting there looking pretty. Still, it's nice to have as a backup.
I really really wanted to make the transition to a flat panel+OSSC. I think we'll get there but it's probably gonna take a few more years.
-
bobrocks95
- Posts: 3472
- Joined: Mon Apr 30, 2012 2:27 am
- Location: Kentucky
Re: OSSC (DIY video digitizer & scandoubler)
I'm kind of unsure why OLED is still using sample-and-hold when its pixel response time could allow it to be impulse-driven (at least I think that's what was holding back LCDs?). I guess when LG is the only company making the panels and they have no competition, there's no reason to drastically improve anything. Maybe there are longevity concerns too if being impulse-driven would require more brightness to counteract the lesser amount of time the pixels are on?fernan1234 wrote:Motion persistence inherent to flat panel display technology was the main issue.bobrocks95 wrote:What was the nail in the coffin for you?fernan1234 wrote:I found this using 100% scanlines on the OSSC. Adding any preferred percentage of hybrid scanlines will add to the effect. At the time I thought that I had finally found a replacement for CRT, but of course that didn't last and now my OSSC is just sitting there looking pretty. Still, it's nice to have as a backup.
I really really wanted to make the transition to a flat panel+OSSC. I think we'll get there but it's probably gonna take a few more years.
PS1 Disc-Based Game ID BIOS patch for MemCard Pro and SD2PSX automatic VMC switching.
Re: OSSC (DIY video digitizer & scandoubler)
Currently it's only possible to name them using the web app.Galdelico wrote:Didn't have the chance to try the new firmware yet, so I'm asking out of curiosity: is it possible to name the profiles on the OSSC as well, or is that only doable through the web app?
Some OLEDs already support BFI which is a step towards that direction. Their motion clarity is closer to CRT and plasma, but come with the price of some flicker. Speaking of motion clarity, refresh rate tests in the display test suite fw can also be used to evaluate motion clearness of a panel.bobrocks95 wrote:I'm kind of unsure why OLED is still using sample-and-hold when its pixel response time could allow it to be impulse-driven (at least I think that's what was holding back LCDs?). I guess when LG is the only company making the panels and they have no competition, there's no reason to drastically improve anything. Maybe there are longevity concerns too if being impulse-driven would require more brightness to counteract the lesser amount of time the pixels are on?
Re: OSSC (DIY video digitizer & scandoubler)
Sony has rolling scan on OLED professional monitors.
I think Fudoh mentioned a couple specific models, but I don't recall the details.
I think Fudoh mentioned a couple specific models, but I don't recall the details.
We apologise for the inconvenience
-
- Posts: 2187
- Joined: Mon Aug 14, 2017 8:34 pm
Re: OSSC (DIY video digitizer & scandoubler)
The biggest drawback of BFI is not flicker, at least at 60Hz. The big drawback is a huge loss of brightness as a result, and current OLEDs are already limited in their brightness. And it's still far from the motion resolution of a CRT, but definitely a nice step forward.
The rolling scan of pro OLEDs has drawback of its own too, apparently. There really won't be something good enough for motion in a while, sadly.
The rolling scan of pro OLEDs has drawback of its own too, apparently. There really won't be something good enough for motion in a while, sadly.
-
- Posts: 208
- Joined: Thu Sep 27, 2018 1:04 am
Re: OSSC (DIY video digitizer & scandoubler)
The latest LG OLEDs can do 900+ cd/m² peaks and ~600 cd/m² nominal on real-world video when fed a HDR signal and ~400 cd/m² all day long on a all-white SDR test pattern. Even with the brightness cut in half from BFI, that's way brighter than the display calibration standard of 120 cd/m² or a bright office lighting standard of ~150 cd/m². Arguably lacking for good HDR, but for legacy SD/HD stuff with SDR brightness, that should be enough.fernan1234 wrote:The biggest drawback of BFI is not flicker, at least at 60Hz. The big drawback is a huge loss of brightness as a result, and current OLEDs are already limited in their brightness. And it's still far from the motion resolution of a CRT, but definitely a nice step forward.