AVS Forum banner
Status
Not open for further replies.
1 - 20 of 32 Posts

·
Registered
Joined
·
1,534 Posts
Discussion Starter · #1 ·
Attached you will find a complete analysis run on the BMW short films recorded from HDNet via AVX1. I haven't anaylyzed any data in this document yet. If anybody has time and can shed some light as to why certain decoders don't like those streams that would be fairly cool.

 

bmw.txt 30.197265625k . file
 

Attachments

·
Registered
Joined
·
3,533 Posts
Same question here. Is it from Manzanita? Anyways,

that stream is pretty busted. Assuming that your

downlink was error free at the time, I think you may

be seeing the quality of the AVX1 DSS Transport to

MPEG-2 Transport conversion.


Ron
 

·
Registered
Joined
·
800 Posts
Ron, do you think you could send [email protected] a few specific observations about the ways in which the stream is broken? Perhaps with only a few simple changes they could improve the processing done by the AVX-1.


I've only just discovered that AVX-1 recordings no longer play back properly on the Panasonic TU-DST50.


I am in no way implying that your time is worth nothing or that you should do work for free. Only that if the fix is a simple one, and you don't mind sharing the information with 169Time, there are a lot of people here who would be grateful.
 

·
Registered
Joined
·
3,533 Posts
rrg,


There's so many errors, I wouldn't know where to start.

However, some errors are worst than others with the

buffer related errors ranking high and the continuity

errors ranking low.


It looks like a pretty big job to understand exactly

what's going on, so I'd say the ball is squarely in

169time's court. They would need to make sure the

test conditions are known (no errors in the downlink),

perform the tests on as many bitstreams as possible

and then attack each problem one by one in an iterative

process (change the software, see if things are better

or worse).


I would suggest that 169time get their hands on the

Manzanita Transport Stream verifier. I'm sure it's

not cheap, but it looks very thorough.


Bitstreams that pass this verifier have a very high

chance of being decoded successfully by most, if

not all, decoders.


Ron
 

·
Registered
Joined
·
1,534 Posts
Discussion Starter · #7 ·
The Dishnetwork's Modulator streams have a lot of errors too, including continuity errors (due to lost packets etc.). However, they do not have buffer over/under run problems nor PCR jitter that's 1/10 of a second.
 

·
Registered
Joined
·
3,686 Posts
FYI from RIchard.


"Bitstreams that pass this verifier have a very high chance of being decoded

successfully by most, if not all, decoders."


The DTC100's built in decoder has been refined several times since it's

introduction more than three years ago. In sharp contrast, although the

decoder in the Panasonic STB has been around longer, it doesn't seem to have

the same refinements that the DTC100's decoder enjoys.


Thomson's (DTC100 / Proscan) engineering didn't waste time going around the

world pointing fingers at all the bitstream weirdnesses they encountered, or

demand that everyone buy their favorite brand of "verifier." Instead, they

put their nose to the grindstone and REFINED THEIR OWN DECODER, making it

better equipped to handle these bitstreams.


There is no magic to the DTC100's decoder. It's time tested refined design

compensates for common bitstream problems. This is the reality of the DTV

world. Some decoders work better. Perhaps a better tool than a verifier

would be a set of test bitstreams that are designed to press the limits and

to expose decoder weaknesses. A decoder designed to just barely handle the

data sheet, without any extra margin, will fall flat on its face so to

speak.


Fortunately, since its introduction, the decoder in the JVC deck has also

been improved by firmware updates. Although JVC hasn't responded as fast as

I would have liked, they have made updates and are continuing to improve the

DVHS deck decoder's robustness.


HDTV from satellite (bitstreams) have as their target the decoder in the

satellite set top box. They are not required to "pass a verifier," or work

with any and all decoders. Instead the one requirement they need to meet is

that the are compatible with the decoder in the set top box.


The 169time AVX1 process was intended to make the signal compatible with the

deck, for playback through the deck and the DTC100 STB. In fact, the

bitstream doesn't originate in the AVX1. The AVX1 modifies it only as

necessary.


It is no surprise that the HDTV signal that is intended to play on the

DTC100 does in fact play back from tape quite well on the DTC100 when it is

played back, and similary well on the JVC deck's outputs.


It is likely that 169time will continue to improve the AVX1 process to ease

the requirements on other decoder where possible, utilizing future AVX! disk

updates. However, some improvement in some other decoders may be necessary

to allow for the nature of the HDTV signal that originate on satellite.

Those decoders that enjoy these improvements will enjoy wider acceptance in

the marketplace.




Dave
 

·
Registered
Joined
·
326 Posts
I'm wondering what the AVX1 actially does to the DSS stream. The DSS stream is not ATSC compliantso the AVX once must intercept things after decryption and then convert it to an ATSC complian stream which can then be packaged up inside firewire packets and sent to a D-VHS.


Looking at the following thread, one can see a bit of what is involved.
http://dealdatabase.com/forum/showth...&threadid=6890


The AVX1 probably has to demux the DSS data stream to MPEG2 program stream format and then remux it into an ATSC complian transport stream.


If they are doing all this in software, what they have done is created an mpeg2 to transport stream converter (along with a DSS to mpeg2 converter). This can be done in software (since Manzanita has a solution) but is not easy. If the AVX1 buffering implementation has bugs or is not designed right, then the errors that are showing up will happen. If the AVX1 relies on its hardware clock for timing resolution, this may not be enough and bad jitter will result. I've read a few research papers on jittier issues with transport streams. There are some software smoothing algorithms that can help.



-- John
 

·
Premium Member
Joined
·
998 Posts
In the case of the ST Micro MPEG decoders (found in DSS receivers), they use a rate estimator to regenerate the presentation time stamps internally, which gets rid of most of the problems with improperly formatted transport streams.


As a presentation device, it's good engineering practice to make the MPEG decoder be as robust as possible in handling flawed streams. However, as a source device (such as the AVX-1), it's bad engineering practice to output flawed data for exactly the reason that not all MPEG decoders may be able to handle a stream that deviates from the spec (worse yet, there may be IP issues over some of the tricks used by MPEG decoder manufacturers to work around poorly generated transport streams, limiting the scope of their usage). That's why the spec there - for compatibility.


-Dylan
 

·
Registered
Joined
·
3,325 Posts
Which decoders like the BMW clip from HDNET, and which don't?


Why is the bit rate shown as 15.8... Mbps? Is that standard for D-VHS?
 

·
Registered
Joined
·
1,534 Posts
Discussion Starter · #12 ·
Quote:
Originally posted by balazer
Which decoders like the BMW clip from HDNET, and which don't?
like:

DTC100, JVC build-in, Elecard


don't like:

Samsung T165, Sonic, WinDVD

Quote:


Why is the bit rate shown as 15.8... Mbps? Is that standard for D-VHS?
That is the bitrate of the original. DVHS decks will accept any bit rate up to and including 28.2mbps. Bit rates less than 14.1mbps are recorded at STD speed and bit rates between 14.1 and 28.2mbps are recorded at HS speed.
 

·
Registered
Joined
·
4,735 Posts
Quote:
Originally posted by leszek1
like:

DTC100, JVC build-in, Elecard


don't like:

Samsung T165, Sonic, WinDVD
Leszek,


Do you have a HiPix? If so, is it possible to play these back through the HiPix?


I have an AccessDTV, but I'm also buying a used HiPix. I seem to have some trouble (they have lots of stutters) playing OTA AccessDTV recordings through ZoomPlayer after converting to .ts files, but I'm not sure if I've tried Elecard with these.


Thanks,

Darin
 

·
Registered
Joined
·
4,735 Posts
Quote:
Originally posted by balazer
What do you mean "after converting to .ts"? I thought AccessDTV records native .ts?
They still use their old naming scheme and .adtv on the end and I didn't realize that these were really transport streams inside and could be played back through ZoomPlayer. So, it looks like I wasn't really converting them to .ts format, but was changing all their little files to a few much bigger files with .ts (and deleting commercials as I went). I'm still not sure why they need to have such a seemingly random naming scheme at this point. Maybe they'll change that at some point.


Now I guess I just need to figure out how to get smooth transitions between files in a playlist from ZoomPlayer and do some more testing. I'm still seeing some stutters with the Kentucky Derby and Gladiator from yesterday, but maybe there is another reason for that.


The main advantages of the HiPix are the better reception (I've heard), the automatic commercial skipping, more file size control, and a native mode 1360x768 for my D-ILA projector, so I'm sure I'll still find a use for it. I'm hoping that they'll either add (or have already added) a 1360x1024 output mode for those using anamorphic lenses with D-ILA projectors.


I have 2 projectors, so I may build another HTPC that is faster than my Athlon 2600+ with Radeon 9500 at some point. For now I'll probably just swap my AccessDTV card in when I want to pause live HD and hope that their software is opened up so that it can get some improvements.


--Darin
 

·
Banned
Joined
·
3,772 Posts
Quote:
Originally posted by darinp


The main advantages of the HiPix are the better reception (I've heard), the automatic commercial skipping, more file size control, and a native mode 1360x768 for my D-ILA projector, so I'm sure I'll still find a use for it. I'm hoping that they'll either add (or have already added) a 1360x1024 output mode for those using anamorphic lenses with D-ILA projectors.

--Darin
The HiPix devel team lost access to a critical module that was required to add custom resolutions. 1360x1024p may not have been possible given hardware limitations, but we can't even try to do it at this point.


Sorry.
 

·
Registered
Joined
·
1,534 Posts
Discussion Starter · #18 ·
Even with access to the critical module 1360x1024 would not have been possible due to maximum dot clock limitation of the hardware itself. [email protected] was already pushing it and even [email protected] already has a higher dot clock. And don't forget the hardware limit on memory itself. There is 8MB to be shared between video memory and the rest of the card's functions.


I know the person who desperately tried to add higher resolutions than the current one (when access was still available) and failed. :(


Leszek
 

·
Banned
Joined
·
3,772 Posts
Well thankfully we have DVHStool that can do 1080p on a fast PC. I treat the HiPix mainly as a recording device only now...
 

·
Registered
Joined
·
4,735 Posts
Quote:
Originally posted by PVR
Well thankfully we have DVHStool that can do 1080p on a fast PC. I treat the HiPix mainly as a recording device only now...
Last time I checked it didn't support anamorphic lenses, though. That is the main reason that I still try to use ZoomPlayer or my AccessDTV card.


--Darin
 
1 - 20 of 32 Posts
Status
Not open for further replies.
Top