AVS Forum banner
Status
Not open for further replies.
1 - 16 of 16 Posts

·
Registered
Joined
·
100 Posts
Discussion Starter · #1 ·
As you all know from my posts I use DV or D8 to record my sporting events.


Essentially it looks about 92% of the original broadcasts -- most would say its pretty darn close to the original.


With D8 I record SP for 1 hour on a D8 tape

DVD HQ looks VERY close to the D8 quality (1 hour on 4.7GB DVD+RW) on my Philips DVDR 985 recorder.

-However with sports SP on DVD does not look good


Below I will list recording mode bitrates

D8 -- 25.0Mb/sec
DVD HQ -- 9.72Mb/sec
DVD SP -- 5.08 Mb/sec


So it figures that somewhere above 9.72Mb/sec is the true threshold of recording Standard Defenition Broadcast Sporting Events to the point where it looks VERY close to the original -- say 92-93%, where SVHS is 75% and VHS SP is like 50%.


I doubt the D8 really needs ALL of the 25.0mb/sec -- but then again it could be a faulty compression scheme compared to DVD-R and D-VHS because they only require 9.72Mb/sec or above to look as good as the D8 recording (of course in my subjective opinion).


However when I record on LP mode on Digital8 which is 1:30hours as oppose to 1 -- it looks probably like 83% of the original -- which is probably a 12.5Mbps if I were to calculate the recording bit rate. However I noticed that DVD HQ @ 9.72Mbps actually might look better than my D8 LP recording.


Which tells you the compression schemes and ratios between D8 and DVD recording and possibly D-VHS recording could be different.


Perhaps I am using faulty logic to compare the two recording formats.


If I were to purchase a D-VHS I would use the STD mode - 5 hours on a 300F tape @ 14.1Mbps and I am hoping it looks as good as my D8 recordings.


Does anyone think it will?


On the JVC 3K deck website it actually says STD mode is better than MiniDV. So does that mean STD @ 14.1Mbps can actually look better than DV @ 25.0Mbps when recording from a standard definition broadcast?


What do you guys think about this and the comparison of the compression schemes?
 

·
Registered
Joined
·
1,270 Posts
DV uses a motion JPEG compression format that allows for frame accurate edits. DVD and pretty much all MPEG-2 based video compression formats sacrifice frame accurate editing for higher compression ratios. As an example MPEG-2 has three frame types I, P, and B. Only the I frame contains all the information needed to reproduce an accurate image for edits. The P and B frames are estimates of true frames that are based on a motion analysis of groups of pixels using an I frame as the point of reference for the estimates. P and B frames use many fewer bits than I frames since they only describe the differences between I frames.


MPEG-2 gets it's compression advantage from the use of multiple P and B frames between I frames. DV is the equivalent of using all I frames.


When it comes time to describe picture quality of a given format using a given bit rate, well the compression hardware makes a big difference, but once you have enough bits there is no big advantage to throwing more bits at the problem. The threshold for how many bits are needed to encode standard definition video varies with hardware, but MPEG-2 can produce a decent image using a lot fewer bits than DV.
 

·
Premium Member
Joined
·
998 Posts
There was a thread in this forum not long ago from a man at a TV station who uses a bank of JVC D-VHS decks for re-broadcasting purposes. He swears by them as being better and cheaper than any tape-based standard def. recorder available in the pro market. I believe he was using STD mode for the recordings. In my own experience I have found the JVC to be very faithful to the original s-video input in STD mode, with no MPEG artifacts.


-Dylan
 

·
Registered
Joined
·
100 Posts
Discussion Starter · #4 ·
Wow -- so STD might even be better than my D8 recordings.


Certainly not noticable worse right?


If so there would probably no use of D-VHS for me -- but if STD is good then I can get 5 hours on one tape!


DAHESTER is there anyway you can find that thread for me?


thanks!
 

·
Banned
Joined
·
3,772 Posts
Digital8 or DV compression (basically motion JPEG) was not just chosen for edit capability... It is because the encoders are much easier and cheaper to make.


Realtime MPEG2 encoders at a cheap/consumer level are only a fairly recent thing.


The new JVC HD camcorder does do MPEG2 encoding in the camera, and so it can record 720p HDTV onto the same DV tapes that are used for standard def (with DV compression).


Anyways, newer compression algorithms like VP6 or H.26L are even more effective than DV or MPEG2... So the future will be able to squeeze more resolution onto the the same storage datarates.
 

·
Registered
Joined
·
100 Posts
Discussion Starter · #6 ·
So what are you saying PVR -- would you guess that my D8 recordings are about the same as the STD recordings on the D-VHS when recording a SD Cable Broadcast?


My interest is to equal my D8 quality and have greater than DVD SP quality when recording sports -- cause anything less and the motion, players, and pictures become blurry.


I just have not found anyone who could tell me how STD looks and has already tried D8 SP mode @ 25Mb/sec.


I am not really sure how to compare the compression schemes considering I haven't used D-VHS STD mode.


It seems that no one has experience in this area :(


Some say LS3 mode looks indisinguishable from the original cable broadcast -- but thats at 4.7Mbps which is about the same as DVD SP mode @ 5.0Mbps recording which I personally tried and it did not look that great -- I could easily see blockiness and pixelization with moving objects/artifacts.
 

·
Registered
Joined
·
1,142 Posts
Quote:
Originally posted by Nikos20

Some say LS3 mode looks indisinguishable from the original cable broadcast -- but thats at 4.7Mbps which is about the same as DVD SP mode @ 5.0Mbps recording which I personally tried and it did not look that great -- I could easily see blockiness and pixelization with moving objects/artifacts. [/b]
I'm not so sure, but, it may depend on the quality of signal. I started taping 'Taken' on SciFi using the LS3 mode on my JVC 30kU, and noticed lots of pixilization and compression artifacts, but, that may have been the poor signal that I get from SciFi by way of D*. I haven't tried with better signals, so it could very well be good with a strong analog cable signal.
 

·
Registered
Joined
·
2,898 Posts
I record on my JVC 40k daily, and for SD material STD mode is pretty much perfect. Recording SD material in HS mode does not result in appreciably better quality. LS3 is far from perfect, probably a notch below SP mode on a S-VHS tape. STD mode is considerably superior to S-VHS SP mode recording. For comparison purposes analog recording on the 40k is superior to any previous analog VHS I have tried, including the JVC 9911 S-VHS VCR.
 

·
Registered
LG 55" C9 OLED, Yamaha RX-A660, Monoprice 5.1.2 Speakers, WMC HTPC, TiVo Bolt, X1
Joined
·
45,656 Posts
The older standard definition RCA/Hughes/Hitachi (DirecTV) and JVC (Dish) D-VHS decks were reputed to make virtually identical copies of digital programming. I would suspect the STD mode of current D-VHS decks would be capable of the same thing.
 

·
Registered
Joined
·
19,586 Posts
Quote:
When it comes time to describe picture quality of a given format using a given bit rate, well the compression hardware makes a big difference, but once you have enough bits there is no big advantage to throwing more bits at the problem.
Well there can be a big advantage, it's just that because of the way MPEG2 and other such algorithms work, you'd have to throw enough new bits that it's impractical. But I don't think that anyone would argue that throwing away more color information than you keep isn't a pretty serious degredation of the image right off the bat compared to the original that couldn't be made much better by throwing lots of bits at it. So much information has already been thrown away in color and high frequency detail, and upping the bit rate only get's you a relatively small amount of it back relative to the increase in storage requirements, that it's not worth it usually. But if you want to throw a lot of bits at it, you can most definitely improve things a a lot over what we are getting these days. I think that most of us could easily tell the difference between MPEG/WM9/etc... compressed content and some 5:1 lossless compressed version running side by side.
 

·
Registered
Joined
·
1,270 Posts
Quote:
I think that most of us could easily tell the difference between MPEG/WM9/etc... compressed content and some 5:1 lossless compressed version running side by side.
This is true, but my point is that if an MPEG-2 codec isn't experiencing bit starvation there is little improvement in the overall image if you double the bit rate. I for one am willing to make the necessary trade offs incumbent with MPEG-2 based video at least when it comes to viewing high quality high resolution images in my own home.


On the other hand, if a 5:1 tape system was released that could be enjoyed by the home consumer at a reasonable price that would be a great format. Of course 5:1 means that only one fifth of the original information is retained on the tape, so this is a compressed format. D5 HD is a 5:1 format, but there is a slightly better D5 format that doesn't have any "compression"


Unfortunately, a "lossless" format like D5 (12 min tape is about $45) with uncompressed 10bit samples is only a fraction of the information captured by the CCD/imager since the bit depth of the sample is only an approximation of the real information that can be detected by the CCD/imager. Of course the 4:2:2 sampling has already thrown away some information so let's try to find a 4:4:4 sampling tape format.


Back to the original point, once an MPEG-2 codec isn't experiencing bit starvation the effect of adding twice the bits needed to encode the video does not result in twice the measurable image information. Just as you've said, the damage is already done.
 

·
Registered
Joined
·
19,586 Posts
Quote:
Of course 5:1 means that only one fifth of the original information is retained on the tape, so this is a compressed format. D5 HD is a 5:1 format, but there is a slightly better D5 format that doesn't have any "compression"
Just as an aside, you can do somewhere around 5:1 lossless generally I think, so nothing would be lost. I just didn't want that statement you made, which was a little too absolute, mislead any newbies. Even without getting anywhere near lossless you could get significant improvements, but they would still be impractical for home use because of the storage requirements.
 

·
Registered
Joined
·
1,270 Posts
Quote:
Just as an aside, you can do somewhere around 5:1 lossless generally I think
It's closer to 2:1 for lossless video compression. To achieve 5:1 you need to use MJPEG which as a DCT based compression format throws away lots of information to achieve that 5:1 ratio.


INTRA frame/field compression is something like MJPEG.


INTER frame/field compression is something like MPEG-2/MPEG-4

This link to a Panasonic D5 HD VTR product description does a good job of explaining the technology being used for D5 HD.


D6 supports recording at 1.188 Gbps. 4:2:2 samples 8 bits. So, this AFAIK is the only uncompressed digtal video source for HD in common use. However, there still is the 4:2:2 sampling issue which is form a analog compression. Last but not least, D6 tapes run about $1 per minute for new tape. Used tape can be found for less.


This link provides a good overview of the available VTR formats.
 

·
Registered
Joined
·
19,586 Posts
It probably depends on what you are trying to do. For a non-linear or broadcast format, you are going to be limited by having to spit out full frames fairly often. For something like a digital cinema format, which is always played start to end, I believe you can get more like 5:1 lossless because you can do much more agressive temporal compression, right?
 

·
Registered
Joined
·
4,039 Posts
Quote:
The older standard definition RCA/Hughes/Hitachi (DirecTV) and JVC (Dish) D-VHS decks were reputed to make virtually identical copies of digital programming. I would suspect the STD mode of current D-VHS decks would be capable of the same thing.
These decks recorded transport streams in STD mode. I know that tape made on JVC from Dish plays perfect in STD mode on 30k or 40k. I have not tried tapes from D* because I do not know anybody who has one.
 

·
Registered
Joined
·
1,270 Posts
Given that the majority of the compression achieved in MPEG-2 is from inter frame temporal compression, it's pretty likely that a compression format that didn't use a DCT and Quantization approach, but did use motion estimation could achieve a 5:1 compression ratio.


As an example, let's say we have a motion vector that describes a group of 8 pixels. This vector can describe where those 8 pixels moved from one frame to the next. A second data structure that describes the RGB color difference between each pixel in that group between frames would be needed.


If the motion vector estimation is done correctly and the image being encoded isn't changing rapidly then the statistics of the pixel grouping work out such that the average number of bits need to describe the color difference between pixels in each group is less than the resolution of each RGB sample that makes up each pixel. Differential encoding depends on the ability to describe differences between elements using less information than it takes to fully describe the element. This translates into using fewer bits to encode the difference information that is takes to fully specify the original value. This differential encoding is where the compression comes from.


In practice and without motion compensation, it works out to something like for an 8 bit sample the average number of bits needed to describe the difference between successive samples is 4 bits. With motion compensation the average number of bits needed to describe the difference between successive samples drops, but the images being compressed and the accuracy of the motion compensation controls the actual number of bits required.


This is why there are differences between hardware MPEG-2 compression codecs. The processing that must be done to search between multiple frames looking for groups of pixels that seem similar to previous or future groups of pixels while minimizing the absolute magnitude difference between all pixels in the group of pixels is a hard problem to solve in real-time. Kind of like trying to find a particular needle in a stack of needles.



MPEG-2 does all this, but the DCT and quantization steps encode the video information in a way that reduces the average number of bits per pixel to something like 12 bits (starting with 24 bits pixels) before applying motion estimation and differential encoding techniques.
 
1 - 16 of 16 Posts
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Status
Not open for further replies.
Top