It is currently Thu Sep 20, 2018 7:30 pm

All times are UTC - 7 hours





Post new topic Reply to topic  [ 12 posts ] 
Author Message
PostPosted: Mon Aug 20, 2018 6:30 pm 
Offline

Joined: Wed May 19, 2010 6:12 pm
Posts: 2749
How can an adaptive comb filter tell the difference between black and white vertical stripes at the chroma carrier frequency, and complementary orange and blue horizontal stripes? I want to know how slightly more modern TVs decode an NTSC signal, but this has been boggling my mind for a long time.


Top
 Profile  
 
PostPosted: Mon Aug 20, 2018 8:21 pm 
Offline
User avatar

Joined: Wed Feb 13, 2008 9:10 am
Posts: 659
Location: Estonia, Rapla city (50 and 60Hz compatible :P)
Adaptive things I have seen are all done in digital domain. ADC, digital processing, DAC.

_________________
http://www.tmeeco.eu


Top
 Profile  
 
PostPosted: Mon Aug 20, 2018 8:36 pm 
Offline

Joined: Wed May 19, 2010 6:12 pm
Posts: 2749
That doesn't answer my question.


Top
 Profile  
 
PostPosted: Mon Aug 20, 2018 8:46 pm 
Offline

Joined: Sun Apr 13, 2008 11:12 am
Posts: 7532
Location: Seattle
It kinda does.

"Adaptive comb filter" means it's a digital filter that uses the theoretical decoded YUV values┬╣ from the pixel, the pixels above and below (i.e. "comb filter") as well as (most likely) the pixels from some number of fields in the future/past.

So how do you tell if something is color or high-frequency luma? Are its temporal and spatial neighbors consistent with color content? If so, it's probably color. If it's a big shimmery rainbow smear? Definitely luma.

Maybe you throw the motion compensating deinterlacer into the same pile of DSP too. If the vibrantly colored thing is moving, you predict that the place where it's moving to should be chroma instead of luma.

┬╣ maybe they actually can operate just on the un-decoded signal. Not clear, haven't thought about it.


Top
 Profile  
 
PostPosted: Tue Aug 21, 2018 4:47 am 
Offline

Joined: Sun Sep 19, 2004 11:12 pm
Posts: 20562
Location: NE Indiana, USA (NTSC)
Even modern TVs can have trouble with fine diagonal stripes on neckties, misdecoding them as rainbows if their angle coincides with the phase of the chroma subcarrier.

In any case, modern NTSC decoding is probably a digital, noncausal, nonlinear elaboration on the 2-line comb filter in full PAL decoding. It involves comparing chroma of this line to the average chroma of previous and next lines and determining based on that how much of this line's chroma to "trust".


Top
 Profile  
 
PostPosted: Tue Aug 21, 2018 8:53 am 
Offline

Joined: Wed May 19, 2010 6:12 pm
Posts: 2749
My guess is that it combs if either the filtered luma and chroma match, or if the filtered luma matches but the chroma is inversed.


Something that I don't understand is when they invented NTSC color in the 1950s why did they design the color carrier with such a big magnitude in the first place?


Top
 Profile  
 
PostPosted: Wed Aug 22, 2018 8:50 am 
Offline

Joined: Wed May 19, 2010 6:12 pm
Posts: 2749
Another possibility is that if it's a 5 line adaptive filter, it can check correlation of each line combination, quantize the results to one bit (lines correlate or not) then send it through some binary logic that will decide which filtering mode.

Vertical lines of luma can be detect if it's alternating between 2 colors at the same luma, for 4 or more lines in a row.


Top
 Profile  
 
PostPosted: Wed Aug 22, 2018 3:17 pm 
Offline
User avatar

Joined: Wed Feb 13, 2008 9:10 am
Posts: 659
Location: Estonia, Rapla city (50 and 60Hz compatible :P)
A lot of it has to do with requirements for off the air transfers, those signals must travel 100km+ and still be recoverable. If it was meant for cables things would be quite different, many of the limits disappear entirely.

As far these filters go, they're nothing you can represent in a simple formula, they're not linear things, there will be plenty of IFs to alter processing flow to reach output the creator desired. These stages are also usually containing unrelated image enhancement things too. Anything you can come up with is plausible.

_________________
http://www.tmeeco.eu


Top
 Profile  
 
PostPosted: Thu Aug 23, 2018 8:20 am 
Offline

Joined: Wed May 19, 2010 6:12 pm
Posts: 2749
I wish I still had my 2004 CRT so I can study the picture quality to determine what kind of filter it uses. I remember it looking amazing with retro games. I remember rainbow atifacts, but not much dotcrawl other than the pixels having a slight wiggle to them.


Top
 Profile  
 
PostPosted: Thu Aug 23, 2018 8:29 am 
Offline

Joined: Sun Sep 19, 2004 11:12 pm
Posts: 20562
Location: NE Indiana, USA (NTSC)
The NTSC Super NES S-PPU in hi-res mode outputs 227.333 chroma subcarrier cycles per scanline and 3 pixels per cycle. This being above the Nyquist rate, it should be possible to make test patterns for the comb filter by using specially constructed hi-res pictures. But I admit this doesn't help tell the difference between behavior for a standard 227.5-cycle scanline and the nonstandard 227.333-cycle scanline of the NES and Super NES.


Top
 Profile  
 
PostPosted: Thu Aug 23, 2018 6:42 pm 
Offline

Joined: Wed May 19, 2010 6:12 pm
Posts: 2749
It broke so I can't test it.


Top
 Profile  
 
PostPosted: Mon Sep 10, 2018 7:07 pm 
Offline

Joined: Wed May 19, 2010 6:12 pm
Posts: 2749
TmEE wrote:
A lot of it has to do with requirements for off the air transfers, those signals must travel 100km+ and still be recoverable. If it was meant for cables things would be quite different, many of the limits disappear entirely.

As far these filters go, they're nothing you can represent in a simple formula, they're not linear things, there will be plenty of IFs to alter processing flow to reach output the creator desired. These stages are also usually containing unrelated image enhancement things too. Anything you can come up with is plausible.


I've also wondered if it would've worked if they placed the chroma carrier in the vestigial sideband of the luma carrier.


Top
 Profile  
 
Display posts from previous:  Sort by  
Post new topic Reply to topic  [ 12 posts ] 

All times are UTC - 7 hours


Who is online

Users browsing this forum: getafixx and 2 guests


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Search for:
Jump to:  
Powered by phpBB® Forum Software © phpBB Group