Support NeoGAF

[Beer Monkey]

That purple and Green thing sounds similar to this issue people were having with the WiiU (after Nintendo changed something in the firmware), seems they had to update their TV's firmware. I know you're talking about a tv decoder, does the images in that thread seem similar?

That happened to me once. Affected the entire screen, not just edges/transitions. I used to have to run the vWii in 480p to avoid it. Made the system unplayable otherwise.

Somewhere along the line one of the Nintendo updates fixed it because I can now run the vWii in 1080p output. Pioneer hasn't released a firmware update for the 2008 Kuro in many, many years, so I certainly didn't update the TV to fix it.

The Wii graphics chip has the entire Wii GPU in it along with the Wii U GPU. I'm guessing the Wii U GPU buffers frames from the Wii GPU and at some point they weren't converting the colorspace correctly during that process if going from Wii standard def to Wii U high def.

 

[theclaw135]

Thanks for the reply, I'll tinker a bit more tonight. If I can't find a happy medium I think I will just settle at D93 and get used to it. It's not COMPLETELY desaturated, just noticeably less than what I'm used to from YPbPr. As for the Xbox, I'm going to start looking for a 1.6 revision with the Xcalibur video encoder. The RGB out of the 1.0 (Conexant encoder) is mediocre at best.

It makes me wonder, though: is the image desaturation there because RGB carries the exact chroma information output by the console (and I'm just used to an oversaturated image from YPbPr)? Or is it because the CRT is actually aged and incapable of the full RGB quality?

Have you tried YPbPr on Xbox? The vast majority of Xbox games are 480p, and should look clearer than 480i over RGB.
Unless of course it's a PAL system. MS royally ruined that by cutting out the entire progressive option.

 

[Madao]

just to get some perspective, i went and used the 240p test suite with both the Wii U and my regular Wii and got a capture of the test pattern from both.

the streteched images are from the old Wii and the ones with black bars are the Wii U.
(i did this quick and dirty)

i ran the suite in 480p mode because my receiver doesn't pass 240p so i couldn't see it.

 

[missile]

^ What's the signal type and coding, and what's the output device?

 

[missile]

About the Chroma Shift Issue on the Wii

Can someone provide me the following data?

Analog
- chroma shift in mm
- CRT picture width in mm
- line frequency of displayed picture

and/or

Digital
- chroma shift in pixels
- exact screen resolution

Edit:
Using the following game as an example:

 

[Khaz]

Maybe you guys can help me there

My grandmother has an old LCD or Plasma, one of the first generation of TV, with a glass panel. Definitely not HD, it doesn't have HDMI though it does have VGA and DVI-D. And Scart.

Aerial TV recently went to full HD here, and we had to change the decoder. Previously she had a non-HD decoder plugged in through RGB-Scart and it was working fine. Now that all the channels are in HD, we bought a new decoder to put on her old TV. The decoder still has RGB-Scart, but the picture shows with terrible colours. As far as I can tell, the darkest area of a picture are shown as bright purple, or bright green, instead of black or very dark colour.

I strongly suspect that the decoder is faulty and doesn't send the proper RGB signal when the digital info is below a certain threshold (Limited/Full?). But can it be a fault in the TV? Like, maybe the TV is fine with Limited RGB but goes crazy with Full RGB, and the old decoder was Limited only?

Problem solved!

After I managed to go through the user menu of the TV, which was for no reason at all in German (Ich könnste nichts Deutsch sprechen!), I fiddled a bit with it, and it turned out that lowering the contrast from the 100% it was at, to slightly below, like 95%, removed the pink pixels entirely! I just restored the default colour settings because hers was completely out there.

I have no idea why it happens, I tried a couple other TVs, one HD LCD and a couple of both HD and SD CRTs, and none had a problem.

 

[televator]

Thanks for the reply, I'll tinker a bit more tonight. If I can't find a happy medium I think I will just settle at D93 and get used to it. It's not COMPLETELY desaturated, just noticeably less than what I'm used to from YPbPr. As for the Xbox, I'm going to start looking for a 1.6 revision with the Xcalibur video encoder. The RGB out of the 1.0 (Conexant encoder) is mediocre at best.

It makes me wonder, though: is the image desaturation there because RGB carries the exact chroma information output by the console (and I'm just used to an oversaturated image from YPbPr)? Or is it because the CRT is actually aged and incapable of the full RGB quality?

Xbox has a butt ugly picture from RGB Scart and composite because it passes through a deflicker filter. It's compromised. YPbPr and RGB VGA bypass it.

Edit: I have not found a data sheet of the Xcalibur processor to know if this is the case with latter Xbox models.

just to get some perspective, i went and used the 240p test suite with both the Wii U and my regular Wii and got a capture of the test pattern from both.



the streteched images are from the old Wii and the ones with black bars are the Wii U.
(i did this quick and dirty)

i ran the suite in 480p mode because my receiver doesn't pass 240p so i couldn't see it.

This is interesting. The Wii's softer picture sort of blends the color bleed and makes it duller. Meanwhile, the WiiU's sharper picture makes the color bleed brighter.

 

[televator]

So I joined Shmups forum for the sole purpose of presenting fudoh with the spec sheet that appears to say that the FM is a 4:4:4 processor. However, I can't seem to PM him there because I'm a noob.

 

[NormalFish]

So I joined Shmups forum for the sole purpose of presenting fudoh with the spec sheet that appears to say that the FM is a 4:4:4 processor. However, I can't seem to PM him there because I'm a noob.

post in the XRGB mini thread. if he doesn't respond, someone else will. and he'll see it regardless.

 

[PumpkinSpice]

just to get some perspective, i went and used the 240p test suite with both the Wii U and my regular Wii and got a capture of the test pattern from both.



the streteched images are from the old Wii and the ones with black bars are the Wii U.
(i did this quick and dirty)

i ran the suite in 480p mode because my receiver doesn't pass 240p so i couldn't see it.

I bet the artifacts are due to chroma subsampling.

 

[televator]

post in the XRGB mini thread. if he doesn't respond, someone else will. and he'll see it regardless.

Yeah, I just did that. Seems my post hasn't been approved yet though...

 

[Madao]

^ What's the signal type and coding, and what's the output device?

as far as signal goes, that'd be 1080p HDMI for Wii U shots and 480p Component for Wii using regular cables (no adapters and no processing)
what'd be output device? the capture card? i use PEXHDCAP60L.

beyond that i dunno what else to look for.

 

[televator]

as far as signal goes, that'd be 1080p HDMI for Wii U shots and 480p Component for Wii using regular cables (no adapters and no processing)
what'd be output device? the capture card? i use PEXHDCAP60L.

beyond that i dunno what else to look for.

I have some requests, if you'd please entertain them, that'd be great. Could you capture 480p HDMI WiiU? Do you have GameCube component cables and an SD Media Launcher? If so Comparisons to GameCube would be highly interesting.

 

[Madao]

I have some requests, if you'd please entertain them, that'd be great. Could you capture 480p HDMI WiiU? Do you have GameCube component cables and an SD Media Launcher? If so Comparisons to GameCube would be highly interesting.

i do have both so it's possible.

 

[FiveMinuteHero]

Few pics from Framemeister setup at 1080p (PC Engine) with perfect scanlines.

 

[missile]

as far as signal goes, that'd be 1080p HDMI for Wii U shots and 480p Component for Wii using regular cables (no adapters and no processing)
what'd be output device? the capture card? i use PEXHDCAP60L.

beyond that i dunno what else to look for.

Thx a lot.

just to get some perspective, i went and used the 240p test suite with both the Wii U and my regular Wii and got a capture of the test pattern from both. ...

Had a deeper look. There is ringing. Looks like a (windowed) sinc filter.

... i ran the suite in 480p mode because my receiver doesn't pass 240p so i couldn't see it.

Can you make the same test with 100% (instead of 75%) color saturation?

 

[missile]

Few pics from Framemeister setup at 1080p (PC Engine) with perfect scanlines. ...

Nice. But what does perfect means in here?

 

[Madao]

Thx a lot.


Had a deeper look. There is ringing. Looks like a (windowed) sinc filter.


Can you make the same test with 100% (instead of 75%) color saturation?

where do i change the color saturation? in the capture card?

oh, i think the ringing is because the image was rescaled when i captured it.

 

[Madao]

since i realized the different resolutions could be interfering, i went and redid the past shots with their native resolution. also got the captures from GC.

Wii U HDMI 480p

GC 480p (official unmodded Component cable)

Wii 480p (official Component cable)

Wii U HDMI 1080p

for all of them, the capture card settings were identical. also, the test suite was outputting scaled assets to 480p for all.

 

[PumpkinSpice]

K, now I am 100% confident it's the chroma subsampling native to GC/Wii framebuffers that is causing those high contrast artifacts.

Edit: both the wrong color lines and the halos can be attributed to 4:2:2 YCbCr.

 

[televator]

Wow, so even GC has the problem.

K, now I am 100% confident it's the chroma subsampling native to GC/Wii framebuffers that is causing those high contrast artifacts.

Edit: both the wrong color lines and the halos can be attributed to 4:2:2 YCbCr.

I thought I remembered reading somewhere that it was the case with 4:2:2 color that caused this sort of thing. I just can't remember where and I thought I must have made it up in my head somehow. lol

 

[Beer Monkey]

Here's a challenge. Using any settings, try to get the checkerboard pattern from the 240p test suite to display uniformly (no scaling artifacts) in 480i (component) or 480p (component or HDMI) on a Wii U. The Wii can do it.

 

[televator]

Here's a challenge. Using any settings, try to get the checkerboard pattern from the 240p test suite to display uniformly (no scaling artifacts) in 480i (component) or 480p (component or HDMI) on a Wii U. The Wii can do it.

This distance looks way better IRL:
http://imgur.com/bfMyC65

Up close:
http://imgur.com/jj5l879

Looks fine to me.

 

[missile]

where do i change the color saturation? in the capture card?

oh, i think the ringing is because the image was rescaled when i captured it.

Ah, ok. Test image is artificially desaturated (75%, old broadcasting standard).

 

[Madao]

Here's a challenge. Using any settings, try to get the checkerboard pattern from the 240p test suite to display uniformly (no scaling artifacts) in 480i (component) or 480p (component or HDMI) on a Wii U. The Wii can do it.

lol. i tried this and it's some funky stuff.

Wii U HDMI 480p output
all the scaling was done with the capture card

720x480 (native capture)

1280x960 scaled

1280x720 scaled

1920x1080 scaled

640x480 scaled

on my TV the pixels look much better. they look like actual squares.

 

[Beer Monkey]

This distance looks way better IRL:
http://imgur.com/bfMyC65

Up close:
http://imgur.com/jj5l879

Looks fine to me.

I see scaling artifacts in both. Look at the second one from far away if you can't see it.

Run them through a scanline generator and it will be even more apparent.

 

[televator]

I see scaling artifacts in both. Look at the second one from far away if you can't see it.

Run them through a scanline generator and it will be even more apparent.

All my squares look more uniform than Madeo's 480p capture...

I'll hook it up the FM and see.

 

[missile]

Ha! My simulator is at least as bad as the Wii U. That's a good thing! xD

Edit:
Fully based on analog video.

 

[televator]

Ha! My simulator is at least as bad as the Wii U. That's a good thing! xD

Edit:
Fully based on analog video.

You're the only one in this thread that actively wants to create artifacts. lol

Anyway, for all intents and purposes the...errm... blank line generator from my FM looks invisible on the checker pattern from WiiU. Beer, I think what you see is my inablility to hold a camera straight. Haha

Edit: I set my FM to scale up to 720p and I was able to see artifacts on the checker pattern. Yet I can't say what's really at fault - the FM, the TV, the WiiU or some combination there of.

Edit: Fudoh replied to me.

@televator: I don't doubt that the Marvell isn't capable of full internal 4:4:4 processing, but if that's the case, then Micomsoft didn't utilize it. Maybe they chose to save memory and bandwith by reducing the whole A/D section to 4:2:2 YCbCr.

I asked him if there was a way to tell from the data line on the board.

 

[missile]

You're the only one in this thread that actively wants to create artifacts. lol ...

Yeah lol. But I like artifacts from an artistical and technical point of view.

 

[Peagles]

Nice. But what does perfect means in here?

My guess is something closer to normal than the regular untweaked 1080p scanlines (which are awful and don't scale properly).

 

[Khaz]

Anyway, for all intents and purposes the...errm... blank line generator from my FM looks invisible on the checker pattern from WiiU. Beer, I think what you see is my inablility to hold a camera straight. Haha

It's still a scan line generator, it creates artificial scan lines out of the RGBS data and makes them visually distinct by interpolating blank lines in between

 

[Peltz]

"Blankline screenshot thread. Because 240p is all the p's I need."

Doesn't quite sound right.

 

[televator]

Yeah lol. But I like artifacts from an artistical and technical point of view.

Missle, could you explain exactly how this particular artifact occurs? I mean you're good at provoking it to occur. What are are you doing to get that result?

It's still a scan line generator, it creates artificial scan lines out of the RGBS data and makes them visually distinct by interpolating blank lines in between

I can't win.

Here's hoping Fudoh will get back to me tomorrow. I asked him if he could point me toward some literature about the FM's internal color space. Hope I didn't annoy him with too many questions. He also said that there's no way to tell by looking at the data lines on the FM board.

 

[BocoDragon]

I never really got into the whole XRGB mini FM profiles thing. Who out there has done some of the best ones?

 

[televator]

God damn it! I'm staying up late, pulling my hair out, looking for some kind of technical spec sheet on the Framemeister, and finding nothing.

I did find a somewhat illuminating old post of Fudoh's though:

right on both accounts. Every video processor converts to YCbCr and (nearly) every single one to 4:2:2. That's because they're made for video material and video material is stored in 4:2:0 these days.

If you output in YCbCr you're usually just moving the second conversion from the Mini to the display. Eventually a display will have to convert to RGB. Displays who can display full 4:4:4 colors without down sampling usually offer a direct path for RGB signals.

Worst case though is something like this:

RGB signal 444 > YCbCr 422 in VP > Back to RGB 444 on the output > into YCbCr 422 in display > back to RGB 444 for the panel.

Sounds like he came to the conclusion based on presumption, not even aware of the data sheet I found that pretty clearly states CSC happens in 4:4:4.

 

[PumpkinSpice]

Have you considered just emailing micomsoft instead?

It might not even require Japanese if you use simple enough English.

Edit:

Here's the datasheet for the A->D chip: http://www.analog.com/media/en/technical-documentation/data-sheets/ADV7441A.pdf

It explicitly says it can handle doing NTSC/PAL/SECAM composite/s-video/ypbpr/rgb to raw ycbcr 4:4:4 or RGB 4:4:4

STANDARD DEFINITION PROCESSOR PIXEL DATA
OUTPUT MODES
The ADV7441A features the following SDP output modes:

• 8-/10-bit ITU-R BT.656 4:2:2 YCrCb with embedded time codes and/or HS, VS, and FIELD
• 16-/20-bit YCrCb 4:2:2 with embedded time codes and/or HS, VS, and FIELD
• 24-/30-bit YCrCb 4:4:4 with embedded time codes and/or HS, VS, and FIELD

COMPONENT PROCESSOR PIXEL DATA OUTPUTMODES
The ADV7441A features single data rate outputs as follows:

• 8-/10-bit 4:2:2 YCrCb for 525i and 625i
• 16-/20-bit 4:2:2 YCrCb for all standards
• 24-/30-bit 4:4:4 YCrCb/RGB for all standards

re: reading the board: typically you see one data line per byte of raw digital video (unencoded). I took another look at the photo you posted earlier and it's really hard to tell from just that since it's clearly a two sided board, but I've been able to follow the inputs and outputs of asics along pcbs using datasheets and just some eyeballing to figure out how stuff works before including some video-related boards that very clearly used ycbcb 4:4:4 just based on the amount of lines on the board going between chips (like you can count 4 luma lines and 8 chroma lines going where the going exactly where the datasheet says it should).

Edit2: or you know, given what we've seen with those test patterns above, you could just run that test pattern on something like the Genesis on RGB through the framemeister and see what you see.

 

[missile]

My guess is something closer to normal than the regular untweaked 1080p scanlines (which are awful and don't scale properly).

Ah ok, I see.

Missle, could you explain exactly how this particular artifact occurs? I mean you're good at provoking it to occur. What are are you doing to get that result?

I can in the most minute detail, but I need to write other things down first.


Some more artifacts;

In this image you can see the NTSC's color carrier frequency (dots).

Got an idea from where the jailbars could come from. Needs further testing...


I'm thinking about doing a TV Repair app explaining faults/effects and stuff.

 

[televator]

Have you considered just emailing micomsoft instead?

It might not even require Japanese if you use simple enough English.

Edit:

Here's the datasheet for the A->D chip: http://www.analog.com/media/en/technical-documentation/data-sheets/ADV7441A.pdf

It explicitly says it can handle doing NTSC/PAL/SECAM composite/s-video/ypbpr/rgb to raw ycbcr 4:4:4 or RGB 4:4:4



re: reading the board: typically you see one data line per byte of raw digital video (unencoded). I took another look at the photo you posted earlier and it's really hard to tell from just that since it's clearly a two sided board, but I've been able to follow the inputs and outputs of asics along pcbs using datasheets and just some eyeballing to figure out how stuff works before including some video-related boards that very clearly used ycbcb 4:4:4 just based on the amount of lines on the board going between chips (like you can count 4 luma lines and 8 chroma lines going where the going exactly where the datasheet says it should).

Edit2: or you know, given what we've seen with those test patterns above, you could just run that test pattern on something like the Genesis on RGB through the framemeister and see what you see.

Haha! I had the ADC chip on the brain before I went to bed. That's a very interesting find. I just sent an email to Micomsoft support. I tried keeping it in very simple wording just in case. I'm gonna try the 240p test suite for Sega CD. Thanks for the suggestions.

IT'S MORPHIN' TIME. Segazord assemble!

Ah ok, I see.



I can in the most minute detail, but I need to write other things down first.


Some more artifacts;

In this image you can see the NTSC's color carrier frequency (dots).

Got an idea from where the jailbars could come from. Needs further testing...


I'm thinking about doing a TV Repair app explaining faults/effects and stuff.

An app... That would be an interesting app. Unconventional, but interesting.

 

[missile]

... An app... That would be an interesting app. Unconventional, but interesting.

Would be about CRT, video, tuner (component) failures and their effects.

 

[Peltz]

Man, missle, your project looks so interesting. What is the end goal of it?

And how will you know when you're "done"? Do you ever feel like you're aiming at a moving target by trying to simulate all the various CRT quirks out there?

 

[televator]

I think I got a better idea. I'm gonna run a color pattern on a laptop set to RGB via HDMI to my TV. My TV has a 4:4:4 native mode in 1080p. Then I'll lower the resolution on the laptop and feed it to the Framemeister to upscale to 1080 and see what happens...

Edit: So here's what I saw. Turning HDMI Direct On increases artifacts. That's my TV treating the signal as 4:2:2 (it's actually RGB). Strangely though, it means the FM reduces artifacts when processing the image. What the hell? Okay...

Combined with the data sheets, I'm feeling this is not a 4:2:2 device.

 

[Madao]

I think I got a better idea. I'm gonna run a color pattern on a laptop set to RGB via HDMI to my TV. My TV has a 4:4:4 native mode in 1080p. Then I'll lower the resolution on the laptop and feed it to the Framemeister to upscale to 1080 and see what happens...

Edit: So here's what I saw. Turning HDMI Direct On increases artifacts. That's my TV treating the signal as 4:2:2 (it's actually RGB). Strangely though, it means the FM reduces artifacts when processing the image. What the hell? Okay...

Combined with the data sheets, I'm feeling this is not a 4:2:2 device.

aren't the artifacts there because, with Direct On and a lower res, the image is being scaled and loses some details?
wouldn't that be basically FM's scaling vs TV scaling?

 

[missile]

Man, missle, your project looks so interesting. What is the end goal of it? ...

My goals basically surrounds the following two categories:
(a) video games (for building specific atmospheres, looks, effects, etc.)
(b) video art (exposing the dynamics of how analog systems break down)
((c) documentation of video systems, transmission, effects and failures)

I have a couple of ideas for designers/animators for "drawing" video effects.

... And how will you know when you're "done"? ...

When time and money runs out! xD

I need some core video effects for a game of mine (becoming gameplay elements).

... Do you ever feel like you're aiming at a moving target by trying to simulate all the various CRT quirks out there?

Man, I can't even see the target! It's so complex, it draws from everywhere!

 

[televator]

aren't the artifacts there because, with Direct On and a lower res, the image is being scaled and loses some details?
wouldn't that be basically FM's scaling vs TV scaling?

Scaling resolution isn't all that's happening. (BTW, my TV mops the floor with Framemeister in raw scaling.) Color is being "scaled" presumably from 4:2:2 to 4:4:4/RGB. 720p RGB in my TV gets down converted to 4:2:2 and back up to RGB. If the Mini was a 4:2:2 device, I expected it to look the same or worse than my TV with color. Yet it looked slightly better with colors by eliminating/reducing ringing... Aside from edges looking fuzzy compared to my TV.

 

[Madao]

Scaling resolution isn't all that's happening. (BTW, my TV mops the floor with Framemeister in raw scaling.) Color is being "scaled" presumably from 4:2:2 to 4:4:4/RGB. 720p RGB in my TV gets down converted to 4:2:2 and back up to RGB. If the Mini was a 4:2:2 device, I expected it to look the same or worse than my TV with color. Yet it looked slightly better with colors by eliminating/reducing ringing... Aside from edges looking fuzzy compared to my TV.

ah i see. it's an interesting behavior.

i think my TV had a setting for RGB 4:4:4 but i forgot what it was since i only checked when there was a thread about 4:4:4 some time ago. i think i didn't use it because most of my devices don't support it.

 

[Khaz]

I need some core video effects for a game of mine (becoming gameplay elements).

While I usually frown upon adding video artifacts to a game to make it more "retro" looking (Cuphead, although having a promising gameplay, is terrible in that regard), I think this could be an interesting approach in terms of game design. Something like picking up an item that temporarily make your screen look like ugly Composite, or with a sudden lack of convergence. I remember a demo/game for the CPC which had the screen seemingly go out of sync for a couple of frames every time you got hit. That was pretty scary at first as the effect was as real as it could be (the CRT didn't really went out of sync, but the image was delayed enough to make it roll or something). I need to remember which and show you guys, I was quite impressed with it.

 

[televator]

ah i see. it's an interesting behavior.

i think my TV had a setting for RGB 4:4:4 but i forgot what it was since i only checked when there was a thread about 4:4:4 some time ago. i think i didn't use it because most of my devices don't support it.

PS3/4 support RGB HDMI so depending on what resolution your TV supports 4:4:4 in, you can take advantage of that. The OSSC can also potentially make use of it.

 

[Peagles]

Found an old pre-PVM shot of mine, hehe...

 

[missile]

While I usually frown upon adding video artifacts to a game to make it more "retro" looking (Cuphead, although having a promising gameplay, is terrible in that regard), I think this could be an interesting approach in terms of game design. Something like picking up an item that temporarily make your screen look like ugly Composite, or with a sudden lack of convergence. I remember a demo/game for the CPC which had the screen seemingly go out of sync for a couple of frames every time you got hit. That was pretty scary at first as the effect was as real as it could be (the CRT didn't really went out of sync, but the image was delayed enough to make it roll or something). I need to remember which and show you guys, I was quite impressed with it.

Done badly it's just annoying. My stuff should be a lil more advanced.

...


Just testing some filter methods. Don't mind the quality for now.

hsync and vsync distortions (will be very smooth in the end)

hsync coupled with color-burst distortions leading to degraded colors

(work in progress)