• Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.

Upscalers, CRTs, PVMs & RGB: Retro gaming done right!

Status
Not open for further replies.

televator

Member
Addendum: I lost track of the conversation. In regards to the Wii specifically, as I said, it's hardware limited to YUY2. In other words it's already sub samples color down to 4:2:2. You wanna guess what component is? It's 4:2:2. So the Wii renders RGB from already compressed signal. If you think Wii RGB looks better than component... Well... I can't help but chuckle at that.
 
Addendum: I lost track of the conversation. In regards to the Wii specifically, as I said, it's hardware limited to YUY2. In other words it's already sub samples color down to 4:2:2. You wanna guess what component is? It's 4:2:2. So the Wii renders RGB from already compressed signal. If you think Wii RGB looks better than component... Well... I can't help but chuckle at that.

Well it's not like the actually color space is the only factor here. RGB on the PS2 looks quite a bit nicer regardless of colors due to having a cleaner output for whatever reason. Wii component is soft as a baby, so when does that get introduced?
 

televator

Member
Well it's not like the actually color space is the only factor here. RGB on the PS2 looks quite a bit nicer regardless of colors due to having a cleaner output for whatever reason. Wii component is soft as a baby, so when does that get introduced?

The PS2 renders actual RGB though. The only processing that occurs is analog conversion. Its completely independent from the component processing. So this is not analogous to the Wii. The best case scenario for the Wii is that RGB looks exactly the same as component.
 

televator

Member
well, all the more reason to hope that the GCvideo developments bring 480p RGsB wii as well.

You misunderstood. The Wii and GC in, their frame buffers, convert RGB565 into YUY2. Any RGB extraction, short of very highly complex reprogramming, is rendered from 4:2:2 component standard.
 
You misunderstood. The Wii and GC in, their frame buffers, convert RGB565 into YUY2. Any RGB extraction, short of very highly complex reprogramming, is rendered from 4:2:2 component standard.
so you can't pull full RGB out of the GPU? At all?
No, the CRT's user-entered settings are causing two signals from a device that should look near identical to look more different than they should. Hence the conclusion that phonedork needs to calibrate his CRT setup.
So it's not that monitors have totally separate settings for Component and RGB, it's that the two look different, and thus require different settings. Right. I misunderstood your post.
 

Mega

Banned
hm. so the CRTs are differentiating based on the incoming signal itself, not on the input as a consumer television might?

No, the CRT's user-entered settings are causing two signals from a device that should look near identical to look more different than they should. Hence the conclusion that phonedork needs to calibrate his CRT setup.
 
Exactly... more or less unless you find some way to interrupt the process in the frame buffer.

so an HDMI mod for Wii would still be confined to the component color space. This seems to run counter to some of the things I've read but I can believe nintendo would do something stupid like this. Is the gamecube similarly confined? A lot of people seem to be arguing that the Wii video processing is pretty analogous.
 
Nintendo is baffling.

Thankfully I'm not dealing with those consoles right now lol. Put away on the shelf where I don't have to worry about them.
 
When it's 2001 and you only have so many $$$ to work with for components and you can only afford 3MB of fast enough RAM for a framebuffer you do whatever tricks you need to get things out the door.

You shouldn't really notice 4:2:2 chroma subsampling in most cases anyway, other than the artifacts.

Edit: I mean this kind of trickery was pretty common. Why do you think the Xbox 360's gamma curve was so broken? Cause they used a piecewise linear approximation of the SRGB curve to save some processing.
 

televator

Member
so an HDMI mod for Wii would still be confined to the component color space. This seems to run counter to some of the things I've read but I can believe nintendo would do something stupid like this. Is the gamecube similarly confined? A lot of people seem to be arguing that the Wii video processing is pretty analogous.

It would still have that bottleneck in the video chain, yes. And lots of people are stupid. :p
 
When it's 2001 and you only have so many $$$ to work with for components and you can only afford 3MB of fast enough RAM for a framebuffer you do whatever tricks you need to get things out the door.

You shouldn't really notice 4:2:2 chroma subsampling in most cases anyway, other than the artifacts.

Edit: I mean this kind of trickery was pretty common. Why do you think the Xbox 360's gamma curve was so broken? Cause they used a piecewise linear approximation of the SRGB curve to save some processing.
Yeah but the gamecube quickly dropped in price to, what, 100$? It seems to odd to save a few bucks on things like this. And I don't think the X360's issues are any less stupid.
It would still have that bottleneck in the video chain, yes. And lots of people are stupid. :p
fair enough lol.
 
Yeah but the gamecube quickly dropped in price to, what, 100$? It seems to odd to save a few bucks on things like this. And I don't think the X360's issues are any less stupid.
I doubt it would have been only a few bucks. RAM used to be super expensive, especially super low latency on-die RAM.

And considering the lead time to develop this stuff it was probably set in stone way a year or two out, when RAM was even more expensive (stuff dropped like a stone from 2000 to 2010).

They didn't really have the luxury that Microsoft did with OGXbox of slapping the thing together at the last minute and selling above cost.

And especially when you consider most people would be sending it through composite anyway...
 

Timu

Member
I know this isn't exactly the place to ask but it's where I've been having this discussion.

With the Startech USB3HDCAP ( micosoft xcapture-1 clone) I've gotten everything working fine except dreamcast over vga. It's being recognized correctly as a 720x480 source but every software says no signal. Am I missing something?
That's weird, works with my Startech card with AmaRecTV. Also is video input set to VGA?
 
That's weird, works with my Startech card with AmaRecTV. Also is video input set to VGA?

Thanks for the prompt response Timu !

I figured out my problem though. I'm going to go smash my head off a wall. a really hard wall.

I was running the capture card through a usb 3 hub ( hooked to usb 3). Everything worked perfect except VGA capture.

I know you dont run things like a capture card through a hub but I tried it anyways. Hours on a friday night wasted. =(
 

Madao

Member
it seems every Nintendo console had some sort of limitation on their video output ever since the N64 days

N64: too blurry
GC: lower than RGB sampled
Wii: same as GC
Wii U: RGB limited

and the NES also had the limitation of going compoiste max. only the SNES seems to have had good output since it had native RGB output in the original model.

i wonder what will they mess up with the NX console...


regarding earlier Wii discussion, the best way to play depends on what type of gear do you own and what your priority is.
it's a lot of possible configurations but the best would be to test with what you have and if it works out, nice. if not, then work up your way from easiest to get to hardest to get.

atm, the best way to play for me is plugging it directly to my LCD TV since the TV has low input lag and does an okay job with 480p. with FM i'd get better IQ at the expense of more lag due to the FM's additional 20ms of lag and i do speed run realetd stuff often so the lower lag is better.

Thanks for the prompt response Timu !

I figured out my problem though. I'm going to go smash my head off a wall. a really hard wall.

I was running the capture card through a usb 3 hub ( hooked to usb 3). Everything worked perfect except VGA capture.

I know you dont run things like a capture card through a hub but I tried it anyways. Hours on a friday night wasted. =(

avoiding that kind of problem is why i went with an internal card :p
 

televator

Member
it seems every Nintendo console had some sort of limitation on their video output ever since the N64 days

N64: too blurry
GC: lower than RGB sampled
Wii: same as GC
Wii U: RGB limited

WiiU being RGB limited is no bad thing at all. It's RGB on a TV standard luminance scale. That's it.

Also, the Wii does have blurry vide compared to GC. The sub sampling isn't too bad on its own though. What really makes a bigger difference is bit depth. Many GC games seem to be limited to 16 bit and that's the source of the bigger eye sores for the system. The banding and dithering... PS2 also seems to be rather limited, but I'm not sure if many games are 16 bit on that system too. Both topped out at 24bit.

Xbox was 32 bit, and 31KHz component on it looks pristine compared to those systems.
 
it seems every Nintendo console had some sort of limitation on their video output ever since the N64 days

N64: too blurry
GC: lower than RGB sampled
Wii: same as GC
Wii U: RGB limited

and the NES also had the limitation of going compoiste max. only the SNES seems to have had good output since it had native RGB output in the original model.

Nintendo deleted A/V from the NES toploader, they deleted s-video and RGB from the SNES mini, deleted component from the Gamecube and Wii, and so on.

RGB limited isn't really that big of a deal in the grand scheme of things.
 

Madao

Member
WiiU being RGB limited is no bad thing at all. It's RGB on a TV standard luminance scale. That's it.

Also, the Wii does have blurry vide compared to GC. The sub sampling isn't too bad on its own though. What really makes a bigger difference is bit depth. Many GC games seem to be limited to 16 bit and that's the source of the bigger eye sores for the system. The banding and dithering... PS2 also seems to be rather limited, but I'm not sure if many games are 16 bit on that system too. Both topped out at 24bit.

Xbox was 32 bit, and 31KHz component on it looks pristine compared to those systems.

it's funny that i was able to notice dithering way back in the late GC days/early Wii days on games like TWW and TP but i didn't really know what it was until later when i read about it.
 

missile

Member
... Many GC games seem to be limited to 16 bit and that's the source of the bigger eye sores for the system. The banding and dithering... PS2 also seems to be rather limited, but I'm not sure if many games are 16 bit on that system too. Both topped out at 24bit. ...
We even see banding for some 24-bit graphics, if not addressed properly.
 

televator

Member
We even see banding for some 24-bit graphics, if not addressed properly.

That's probably what is happening on some PS2 games.

Back on the topic of Wii/GC: I've been trying to read what little I can find on the topic down/up sampling methods. It seems to me, that the method that causes the ringing artifact that we've seen in screen captures in this thread is one called "multi-tap FIR".

However, according to my reading, that method is a more intensive process than other methods. So it's kinda strange that Nintendo would go for that. Then again, it is the BT.601 standard...
 

missile

Member
That's probably what is happening on some PS2 games.

Back on the topic of Wii/GC: I've been trying to read what little I can find on the topic down/up sampling methods. It seems to me, that the method that causes the ringing artifact that we've seen in screen captures in this thread is one called "multi-tap FIR".

However, according to my reading, that method is a more intensive process than other methods. So it's kinda strange that Nintendo would go for that. Then again, it is the BT.601 standard...
You're getting closer, but man, read about filters and you are set! ;)

classic ringing:
- a filter's way to over- or under shoot from ideal gain

classic filter delay: (rainbow artifacts (in parts) etc.)
- a filter's way to delay certain frequency different than others

About FIR tabs
More tabs:
- better roll-off of the filter, but perhaps more delay and ringing etc.
(depends on the FIR filter type, some filters can never ring)
- more computation needed (implemented in DSP, they work on FIR with many tabs)

Alternative
IIR filter:
- better roll-off characteristics
- more ringing, more (non-linear phase) delay, usually
- much faster than FIR for the same filter characteristics, usually
- may become unstable esp. when doing a fixed-point approximation
(filter coefficients need to be re-adjusted for fixed-point calculation)
- not so easy to control than FIRs, FIRs are a safe bet for DSPs

Each filter type of each class (FIR, IIR) has its strengths and weaknesses.

There is no ideal filter.
ideal: steepest roll-off, linear phase, and no overshoot over all frequencies
 

Rich!

Member
Right, it's confirmed. I'm off to get another PVM tomorrow.

Luckily I now know exactly what to look for, and this time I'm taking the 240p suite. If what he's saying is right, I may have one that's near new.

What's great about these things is that they're so small and easily storable. Having two or even three of them takes about as much usable space as my old 28" Sony CRT I used to have!

Also, Ill be adding phono jacks to the back of my SFC later seeing as I have a couple spare. Fun times.
 

Rich!

Member
sorry to double post but

I have a SCART RGB cable for an NTSC SNES wired for c-sync.

I don't need it. Does anyone want it? I'll take a tenner.
 
Right, it's confirmed. I'm off to get another PVM tomorrow.

Luckily I now know exactly what to look for, and this time I'm taking the 240p suite. If what he's saying is right, I may have one that's near new.

What's great about these things is that they're so small and easily storable. Having two or even three of them takes about as much usable space as my old 28" Sony CRT I used to have!

Also, Ill be adding phono jacks to the back of my SFC later seeing as I have a couple spare. Fun times.

I've got 3 PVMs, 1 BVM, 20" and a 27" Trinitrons and 3 CRT computer monitors. My wife is on the Verge of killing me.
 

Galdelico

Member
Ok, so... I'm about to receive my first PVM and I'm mildly excited.
I'm keeping my expectations low, partly because it's not a super hot model - I've got a PVM-1440QM, which is able to display 300 TV lines only, but it's literally the first not mentally overpriced monitor I came by in ages, here in Italy - partly because I couldn't really find any solid information about it, online.

Anyone here on CRT-GAF ever owned/messed around with one of these models?
My main concern is that it may lack any kind of H/V center option, let alone a proper service menu. I downloaded the manual, which doesn't mention any hidden option to tweak the geometry.
 
PM me and I can send it over on Monday



I don't have a wife anymore, my daughter loves Mario and "sega" (sonic), and my current girlfriend thinks my setup is awesome

I'm good for at least another seven CRTs!

I was always told to find a women you hate and buy her a house and get it over with because the second one is so much better. Can you confirm?

Sweet story about your daughter though. Keep her away from mobile gaming please. She our last hope.
 

televator

Member
You're getting closer, but man, read about filters and you are set! ;)

classic ringing:
- a filter's way to over- or under shoot from ideal gain

classic filter delay: (rainbow artifacts (in parts) etc.)
- a filter's way to delay certain frequency different than others

About FIR tabs
More tabs:
- better roll-off of the filter, but perhaps more delay and ringing etc.
(depends on the FIR filter type, some filters can never ring)
- more computation needed (implemented in DSP, they work on FIR with many tabs)

Alternative
IIR filter:
- better roll-off characteristics
- more ringing, more (non-linear phase) delay, usually
- much faster than FIR for the same filter characteristics, usually
- may become unstable esp. when doing a fixed-point approximation
(filter coefficients need to be re-adjusted for fixed-point calculation)
- not so easy to control than FIRs, FIRs are a safe bet for DSPs

Each filter type of each class (FIR, IIR) has its strengths and weaknesses.

There is no ideal filter.
ideal: steepest roll-off, linear phase, and no overshoot over all frequencies

Thank you, missle! Yes, there is no perfect method according to what I've found as well. It's just that FIR tends to look sharpest. I think it's better than the other methods and that's probably why it is the 601 standard. It's been interesting reading about this after being told that component is virtually lossles or indistinguishable. Yet when one turns on the Wii, the artifacts are imediately apparent. The menu with Wiimote settings shows a heavy chroma shift on the red power button displayed on screen.

After reading just a bit on this subject, it's now easier to attribute certain picture anomalies via component back to sub sampling, and by the look of it, there will always be anomalies no matter what methods are applied.

So anyway, this is a filtration method? I suppose that makes sense since you told me that upscaling from one resolution to another employs filtering. And sub sampling is down scaling chroma and later it's is up scaled via another filtration pass.
 

TeaJay

Member
I was always told to find a women you hate and buy her a house and get it over with because the second one is so much better. Can you confirm?

Sweet story about your daughter though. Keep her away from mobile gaming please. She our last hope.

Funny enough, the same anecdote works for me too - I dated a girl for a while who was adamant in trying to make me stop wasting time with my gaming. Needless to say it didn't last. Current gf is not really a "gamer" as such, but she's way more supportive and into it than I even hoped.
 

missile

Member
Thank you, missle! Yes, there is no perfect method according to what I've found as well. It's just that FIR tends to look sharpest. I think it's better than the other methods and that's probably why it is the 601 standard. It's been interesting reading about this after being told that component is virtually lossles or indistinguishable. Yet when one turns on the Wii, the artifacts are imediately apparent. The menu with Wiimote settings shows a heavy chroma shift on the red power button displayed on screen.

After reading just a bit on this subject, it's now easier to attribute certain picture anomalies via component back to sub sampling, and by the look of it, there will always be anomalies no matter what methods are applied. ...
And there is another issue no one really thinks about when doing filtering:
Who says that all filtered YCbCr samples also lay in RGB space again? Ha!
 
Status
Not open for further replies.
Top Bottom