AVS Forum banner

Status
Not open for further replies.
1 - 20 of 24 Posts

·
Registered
Joined
·
309 Posts
Discussion Starter · #1 ·
I know this topic has been beaten like a dead horse as it is something that is noticeable on the new samsung RPTV..


However I never noticed it when watching the old Panasonic DLP, or any of the newer HD2 front projectors... namely the infocus 7200 or the new Seleco RPTV


It would seem that it is an issue outside of the DMD ? if it didnt affect the other products using this DMD chip? why the samsung?
 

·
Registered
Joined
·
2,250 Posts
d4lions,


Have you noticed it on the Samsung? If so when?


I've only seen it during fast horizontal pans, usually when watching a football game and a pass occurs. I've not noticed it in movies, where because of the slow framerate fast pans usually don't happen.


What material have you seen on the InFocus or Seleco?


-phil
 

·
Registered
Joined
·
1,688 Posts
Quote:
Have you noticed it on the Samsung? If so when?
PhilB,


Are you serious? If so, is this your definition of Temporal Dithering? If we are using the same definition, myself and many others see it on our sets. If we are not using the right term for this, can you provide us with one?

Quote:
Temporal Dithering - Again not as bad as my old set but, still there. On objects like walls is where it is most apparent. I think there are supposed to be different shades on the walls and the TV cannot make up it mind on what should be displayed so, it keeps changing the color in certain areas. It kind of looks like swarming of the picture. It is constantly changing.
Thanks,


-ftlee
 

·
Registered
Joined
·
309 Posts
Discussion Starter · #4 ·
FT-


This also was why I brought it up again... I have heard some say "A shimering of green on a football game" others like you talk about a color issue...


I myself am not exactly sure what the best description for what I see on these sets are..


I saw the "swarming" both at cedia, and the other days at Tweeter who was displaying an HD NET feed of some basketball game.....


I worked with a few of the older panasonic DLP's, and never saw this... I have seen several DVD's on the infocus 7200, and the seleco RPTV, only the feeds that were shown at CEDIA.. but again I never noticed this swarming affect there..


The reason I ask is simple...


I have been awaiting my 50" sammy for awhile, Now I am told I can get a 61 easier... (I am a dealer... so they are trying to pacify me)....


But the Panny 50" will be released this week or within this month..


I am wondering... if I should jump ship... as I didnt see this on there previous release....


the 61 pre-production Sammy model I saw at CEDIA... was nice... but had several PQ issues... (I stopped paying attention when they said it was pre-production, and the korean engineer couldn't speak very good english)


Further, it doesn't make sense that some people are saying that the dithering is inherent in the HD2 chip, if so... it seems this problem would be across the board... FP, RPTV the whole lot?


MR Wiggles... any insight?


-ADS-
 

·
Registered
Joined
·
2,250 Posts
Quote:
Originally posted by ftlee
PhilB,


Are you serious? If so, is this your definition of Temporal Dithering? If we are using the same definition, myself and many others see it on our sets. If we are not using the right term for this, can you provide us with one?




Thanks,


-ftlee
Frank,


I was simply asking d4lions if he had seen the Sammy. He stated that he's seen the Panny, Infocus and Seleco, but didn't say if he's ever seen the Sammy. It's not a scientific test if he's never seen the subject.


No, that is not temporal dithering. You will not see temporal dithering on images that are static. Temporal dithering occurs on images that are changing with TIME. That's why it's called TEMPORAL dithering. You'll see it on moving images.


Lose the attitude, dude.


-phil
 

·
Registered
Joined
·
1,688 Posts
PhilB,


Sorry for the attitude. Kenland had told me before that the problem with the pixels not being able to make up their mind on what color they should be was Temporal Dithering. I even called it that in my review because that is what I have been led to believe the definition was for this problem. Using this for a definition and reading your message, it sounded like you were saying the problem does not exist on the Samsung DLP and that is not the case.


What is the definition for what I and others are seeing related to above? This way I can use the right terminology for it.


Thanks,


-ftlee
 

·
Registered
Joined
·
309 Posts
Discussion Starter · #8 ·
Phil-


Yes I have seen the Samsung... both at Cedia and locally... and from my last post as you can see... I am awaiting its arrival..


MMM... so you say you have seen this on the old panny...


I definately didnt see it on the seleco at the show...


and it wasnt on any of the FP I saw the Infocus 7200 or the HT1000 (4:3 HD2)
 

·
Registered
Joined
·
2,250 Posts
Frank,


The definition you gave sounds more like generic dithering than specifically temporal dithering. The DMD chip does dither in low IRE areas. From my ColorFacts experience, the amount of dithering can vary according to how the set is tweaked. Sometimes I'd see dithering in 40 or 50 IRE fields, and then with further adjustment I wouldn't.


-phil
 

·
Registered
Joined
·
56 Posts
I did a lot more viewing on this "cannot make up my mind in colors" problem. When looking for the issue on DVD playback from my HTPC, I did see it come forth in other display devices including CRT and LCD's (although fixed pixel displays always looked the worse). I was going to play Episode One from my VHS to the 507 to see if it also exhibited the issue. Alas, my video cable pin broke off (I have no idea) in my video out on the VCR and I haven't bothered to fix it yet. Anyways, why test this? One of the best explanations I heard for this problem was actually inherent to MPEG compression artifacting. The theory is the VHS, not being commpressed, should look like crap compared to the DVD, but would NOT exhibit the shimmering effect.


I won't get around to testing it soon, but I'd look forward to anyone else who can.
 

·
Registered
Joined
·
89 Posts
Quote:
Originally posted by PhilB
The definition you gave sounds more like generic dithering than specifically temporal dithering.
There seems to be a terminology clash here. On old computer displays that could only handle 8 or 16-bit color, trying to display a static 24-bit color image resulted in the 24-bit colors being "approximated" by patches of various 8 or 16-bit color. Slightly different colors at different pixels, statically displaying a static image, but in such a way that if you stood back far enough, the displayed colors "averaged out" to the unavailable 24-bit colors. This was known as just plain "dithering" - although "spatial dithering" would be more accurate.


In the DLP case, I'd call it "spatiotemporal dithering", since it varies in both space (due to the limited color depth of single-chip DLP) and time.


I would also point out that the temporal variations occur even on static images. I was in a meeting today with a DLP front projector, and if you got up real close to the screen, temporal dithering on a statically displayed slide was plain to see. But the magnitude of the temporal fluctuations might be implementation-dependent... depending on how the engineers choose to deal with the limited color depth.
 

·
Registered
Joined
·
3,679 Posts
"Dithering" means making more colors from fewer colors. Usually its all spatial like on your inkjet printer. Those beautiful photo prints are made with tiny dots of only 3-6 colors arranged such that your eye sees the intended color.


"Temporal" Dithering would mean that the color at a single pixel site was changing rapidly in an attempt to approximate a finer color. In a sense this is the very heart of DLP. Rapid application of the 3 primaries at the same pixel site to make you see a precise color.


The "Temporal Dithering" that we are talking about here would mean that the color at the single pixel site was changing so slowly that our eyes could detect the intermediate colors and see them changing.


The TD that I've seen and that Mr. Wiggles has pointed out to me appears more as an oscilation between to colors as though the DLP can't make up its mind. Kind of like the DLP could produce 39.9 IRE and 40.1 IRE and switched back and forth when trying to display 40.0 IRE.


I was told this was do to the mirrors' switching speed not being able to keep up. So somehow certain colors or color changes (on moving video) create a circumstance that limits/quantizes a pixels potential colors to fewer than normal. The compensation algoritm is to oscilate between two colors that it *can* make. (temporal dithering)



Ken
 

·
Registered
Joined
·
114 Posts
Wouldn't you expect that the "experts" at TI have an interest in suggesting a "fix" if there is one available? I suppose, like Samsong, while they'll claim to monitor the web closely, however, strangely, they know nothing about this PB. In the early days of these models, I thought that we were told that Samsung, not having worked very long on this technology, was relying heavily on TI's reference design.


If I was Martha Stewart, I might just now start selling some TI stock short.


ed g
 

·
Registered
Joined
·
2,250 Posts
Ok Ken,


Looks like we're going to have to agree to disagree on this one. Based on this thread:

http://www.avsforum.com/avs-vb/showt...hreadid=158564


It looks like there are two predominant definitions of temporal dithering yours (staitic image chaning in time) and mine (moving image loss of detail). It sounds to me like temporal dithering is an AVS term and thus has no hard and fast definition.


-phil
 

·
Registered
Joined
·
3,679 Posts
Ed,


Don't mistake a technical discussion with a condemnation of the technology.


I've enjoyed many a movie on my older DLP projector whose mirrors switch at least 30% slower than the new HD-2's.


I think the artifacts that many of us (including me until recently) were mistaking for temporal dithering are actually mpeg2 artifacts. Phil, Mr. Wiggles, and I tested this by playing a video game (Crazy Taxi) that had faster pans than any movie you're likely to see. The level of artifacting was on the edge of perception - meaning that they were very slight and noticable only if looking for them.


I've heard from good sources that the Samsung is applying less anti-alias filtering than other sets and so the image and artifacts will be a bit more pronounced. Some will prefer this, others not.


It does mean that the better the source the more pleasing the picture.


Ken
 

·
Registered
Joined
·
1,710 Posts
I want to ask a basic "how does it work" question, How does MPEG2 work? Does MPEG2 compress by storing relative pixel changes instead of actual pixel content? That is to say a static picture will carry little data in the MPEG2 data stream. A rapidly changing picture with ALL pixels changing in time demands a large bandwidth in the data stream. Is it possible that a choke-point exist in the MPEG2 decoder? Does this demonstrate the need for a new compression algorithm? MPEG4?


Just a non-computer geek thinking computers here. :)
 

·
Registered
Joined
·
3,679 Posts
Phil,


I think the difference is only in the *reason* the DLP is applying (or suffering from) TD.


When the video is changing rapidly (fast motion) it puts more pressure on the the DMD to produce more shades of color more rapidly. The result is temporal dithering - putting out the colors that it *can* put out in hopes that your eyes will be tricked into seeing the correct color.


The DMD mirrors can switch 15,000 times per second. (up from 10,000) That's not fast enough to provide 24bit color at 30 fps when colors are changing every frame. So something has to give. That give is motion induced TD.


Ken
 

·
Registered
Joined
·
2,250 Posts
Quote:
Originally posted by gpflepsen
I want to ask a basic "how does it work" question, How does MPEG2 work? Does MPEG2 compress by storing relative pixel changes instead of actual pixel content? That is to say a static picture will carry little data in the MPEG2 data stream. A rapidly changing picture with ALL pixels changing in time demands a large bandwidth in the data stream.
That's a pretty good description of one of the main compressions methods of MPEG2. Every once in a while (every 8 frames, I believe) MPEG2 does redraw the whole frame, but mostly it only sends information about what has changed from the previous frame.


-phil
 

·
Registered
Joined
·
114 Posts
Quote:
Originally posted by KenLand



The DMD mirrors can switch 15,000 times per second. (up from 10,000) That's not fast enough to provide 24bit color at 30 fps when colors are changing every frame. So something has to give. That give is motion induced TD.

It would seem that a test using a fast computer driven display switching between a variably intense white frame and a much longer constant black frame should show the point at which the DMD, of whatever, can't keep up with rate of change and can no longer produce a bright and "pure" white.


Since the human eye perceives each primary color differently, it might be better to test with only one prime (green). I don't see how the 24bit issue would enter into it except in a test of the "bit widths" of the decoding and driving digital paths. If there is enough time for the "slowest" primary, there is enough time for any color.


I won't be surprised to hear that going to 16:9 aspect had a lot to do with timing problems that hadn't showed up in older standard FPs.



ed g
 

·
Registered
Joined
·
3,679 Posts
Phil,


Your TD link says exactly what I'm saying. The only relevent posts are by FlatusM, Mr. Wiggles, and Bob Williams.


The original poster is seeing image tearing not TD. (Bob)


TD is due to the mirrors switching too slowly to fully render the color to 24 bit precision when the colors are changing every frame.


To do this the single chip DMD would require a switching speed of just over 23,000 per sec.


Ken
 
1 - 20 of 24 Posts
Status
Not open for further replies.
Top