10bit vs 12bit dvd players

[omac]

ive been using my pioneer dvd717 for "some" time now and i was thinking of buying a new one.

Does anyone know whats de difference (if any!) between a 10 bit d/a converter and a 12 or 14 bit that come with new players? I know my player is old but Return of the king looks stunning despite that...

I have a panasonic 32" pl10 connected via QED rgb only scart.


PS: my tv cant take progressive (grrrrr..) so will 12 bit make a big diference in picture quality?

Thanks!

 

[mtrycrafts]

ive been using my pioneer dvd717 for "some" time now and i was thinking of buying a new one.

Does anyone know whats de difference (if any!) between a 10 bit d/a converter and a 12 or 14 bit that come with new players? I know my player is old but Return of the king looks stunning despite that...

I have a panasonic 32" pl10 connected via QED rgb only scart.


PS: my tv cant take progressive (grrrrr..) so will 12 bit make a big diference in picture quality?

Thanks!

No, not to worry. I believe it is the number of grey scale differences it can produce, the number of colors? It exceeds your capability most likely.
Keep yours until you get a progressive TV.

 

[Karp]

ive been using my pioneer dvd717 for "some" time now and i was thinking of buying a new one.

Does anyone know whats de difference (if any!) between a 10 bit d/a converter and a 12 or 14 bit that come with new players? I know my player is old but Return of the king looks stunning despite that...

I have a panasonic 32" pl10 connected via QED rgb only scart.


PS: my tv cant take progressive (grrrrr..) so will 12 bit make a big diference in picture quality?

Thanks!

I don't know if I fully understand the differences either, but this is my understanding:

The video DAC's convert the digital data from the DVD disc into analog data for your television. The 10 bit vs 12 or 14 bit refers to the sampling (word)size that is used to convert the data. You will also usually see a reference to the speed of the DAC (ie 54MHz or 96MHz). The faster DAC will have a higher sampling rate.

Think of the digital data as a set of stairs. Imagine that the various colors are the depth of the tread (with red having a 1" tread, green having a 2" tread, blue having a 3" tread ect.). Imagine that the larger the "bit" size, the more accurate the measurement is. Consider the speed of the DAC as the height of the steps, with the faster DAC having shorter steps.

If you now draw a line across the edges of the step, and consider that line to be the analog output data, you would get a smoother and more accurate line with the shorter/more exact length (faster/larger bit) steps. Your television then converts that line into the various colors on the screen.

The smoother line should result in fewer color and shadow errors on the picture that you see on the screen.

Since a progressive scan signal has more data than an interlaced signal, the word-length is more critical. 10 bit is usually accurate enough for an interlaced (regular) signal. It will result in very few errors with a progressive scan as well. A 12 bit or 14 bit DAC will have fewer chances of error, but like anything else, you reach a point of diminishing returns.

Try THIS link to get a more detailed view of what I am trying to say, although it is quite a bit over my head (I'm not an engineer).

If you have a digital connection (DVI/HDMI) between the DVD player and the television, the video DAC's aren't even being used. The player (should be) sending an exact representation of the picture.

 

[docferdie]

Again I believe this 10 bit vs 12 bit vs 14 bit is simply a manufacturer's gimmick. DVDs are encoded at 8 bit x 3 (one for red, green, and blue) which comes out to a 24 bit pallete or 16 million colors which is the limit of color discrimination of most humans. Most computers operate with a 32 bit (8 bit x 4--RGB and greyscale I think) color pallete which from a numerical point of view is like billions of colors but from a practical view is useless since the eye can't really distinguish it.

 

[Unregistered]

It's not really a gimmick. Its analogous to converting 16 bit audio samples to 24 bit or higher before doing any manipulation of those samples. For example, a gain change requires a multiplication. The result of multiplying two 16 bit numbers may not fit in 16 bits, so some information would be lost.

 

[docferdie]

even if you convert a 16 bit sound sample to 24 bit the detail is still 16 bit. Same with a 1 megapixel photo you can scale it with photoshop to 100 megapixels but the actual picture detail is still 1 megapixel. Same goes with DVD you can scale up to 10 or 12 bit but the color pallete of the recorded image is still 8 bit and processing power is just wasted.

 

[Unregistered]

Yes, doc but that was not the point. Obviously converting a 16 bit signed number to 24 bit involves nothing more than a sign extension. The value will still be the same.

I don't know what kind of operations might need to be done to video samples but I'm sure there are some and the extension from 8 bit to 10 or 12 is done for the same reason as for audio - to preserve precision.

Take normalization of audio as an example. Normalization is a gain change; ie a multiplication. Unless the magnitude of the sample and the factor to multiply by are very small, multiplying a 16 bit value by that constant will not fit in 16 bits. So the values are extended to 24 bit and the multiplication performed. If the result exceeds 16 bits, the result can be truncated, rounded, or dithered (thats an implementation decision and some audio editors let you choose).

I hope you are not saying that nothing beyond 8 bits for video is ever necessary and that 10 or 13 bit video dacs are purely marketing b.s.

 

[scsiraid]

The extra bits are there to enable oversampling not increase the 'precision' or 'detail'. Oversampling allows the player to simulate a higher sample rate which allows the analog post filter (which removes the sampling frequency from the signal) to be set at a higher frequency and have less effect on the frequencies you are interested in. For example... in audio a 44khz sampling frequency with a 20khz max frequency of interest gives you one octave to roll off the 44Khz information into the noise floor. The filter to do that is UGLY and makes a mess of the music frequencies. Oversampling by 8 (requires 3-4 more bits of precision in the dac's for the intermediate values) inceases the apparent ampling frequency to 44k x 8 = 352khz which is a lot easier to filter out without messing up the music. Same principal for video but higher frequencies. The extra bits allow the player to calculate intermediate values in between the samples from the DVD which is what oversampling is all about. Listen to an old CD player that doesnt oversample... they are HORRIBLE.

The smoother stairs analogy is close but not for the reasons given. The smoother stairs changes states may times faster then the coarser stairs and this rate of stairstepping represents the sampling frequency that you must filter out before sending the signal on. The higher frequency/rate (relative to the signal frequencies of interest) make it easier to filter.

 

[Unregistered]

The extra bits are there to enable oversampling not increase the 'precision' or 'detail'.
---
Oversampling by 8 (requires 3-4 more bits of precision in the dac's for the intermediate values)
---
The extra bits allow the player to calculate intermediate values in between the samples from the DVD which is what oversampling is all about.

That is exactly to increase 'precision' - So that intermediate values are accurate.

 

[docferdie]

"allows the player to simulate a higher sample rate "

So do you think that if you got film which runs at 24 fps used a filter to run it at 60 fps it would look different? Of course not since you still only have 24 unique frames of detail. Sample rate is to audio as frame rate is to video. If your source is running at 48000 Hz upconverting it to 96000 Hz and then doing DAC does not increase fidelity. It allows you to inject filters such as treble or equalization or whatever you want but it does not improve the original bitstream which is what you want to hear.That's like saying that DVDs will become high def if you run it through a hi def video processor. Sure you can alter it but it won't have more detail or be uniformly better. Another example would be lets say you had a DVD image that showed 256 shades of red or 8 bit red an 8 bit DAC will do the same job as a 12 bit DAC or a 1000000 bit DAC as what needs to be displayed is the 256 shades of red that were encoded in the first place.

The improvement in the sound of modern CD players has to do with error correction and improved signal to noise not with oversampling. Error-correction allows the player to decide that if it reads 101010 then the data that needs playback is really 101010 and not 101011 or 1111111.

"I hope you are not saying that nothing beyond 8 bits for video is ever necessary and that 10 or 13 bit video dacs are purely marketing b.s."

Like I said before an 8x8x8 RGB pallete represents over 16 million unique colors and unless you have something in your retina that other people don't then yes it is purely marketing BS. Consider this, a single DLP chip HDTV only quotes a color range of 16.7 million which again is 8x8x8 RGB. Also consider that most humans only perceive about 10 million colors and you'll see how pointless it is to operate video DACs at anything higher than 8 bit.

Again if anybody would like to refute my photoshop analogy then I would be glad to hear it. Just to further illustrate this point. If I had a 1 megapixel photo recorded with a 15 bit color pallete or RGB 5x5x5 and decided to scale it to 1000 megapixels and use a 24 bit color pallete it would still look like the original picture and would not look any better and would just use a lot of CPU time and space on my hard drive. If you were to do editing then it would make sense to use a higher color pallete because you would be adding elements that were at a higher resolution or color pallete. When you watch a DVD all you want is for the DAC to give you the image that the bitstream was encoded off in the first place. You are not trying to edit the video as you go along. Upsamplers and upconverters are only really useful to people who do editing, retouching or restoration. It has nothing to do with someone who just wants to ACCURATELY play back pre-recorded content.

 

[scsiraid]

Doc... you are missing the point... What the extra bits in the dacs do is improve the ability of the hardware to reproduce exactly what the original material looks like. They dont directly 'improve' the signal being reproduced... they enable the reproducing circuits to not screw it up on the way out. You have to understand sampling and the sampling artifacts generated and how you get rid of them to really understand this.

I strongly disagree that oversampling doesnt improve the sound quality. Just listen to an early player without oversampling and you will hear the difference. The first oversampling player with 20 bit dacs was the Denon DCD1520. I bought one when it came out and I can assure you that the oversampling and 'better' dacs it had made on a HUGE difference in sound quality. What enabled that improvement was the 8x oversampling, the extra 4 bits in the DACs and most importantly the lower order digital filter that the oversampling and DAC's allowed.

Lets talk about your photo analogy a minute.... If you photo is quite large to the point where the pixels are visable then the pic will look pretty bad... if you oversample that photo to increase the number of pixels you have to increase the precision of each pixel to be able to interpolate the 'new' pixels in between the original pixels. (You want the interpolated pixels to smoothly transition between the two original pixels. if the two original pixels differ by only 1 RGB point then you must increase bit depth to interpolate) This increase in precision allows the picture to now look 'better' than the original. However... it doesnt have any additional resolution in reality than the original... you just reduced the 'artifacts' you got from sampling.

This really has nothing to do with the topic though because pictures are not electical signals that need to be processed in the frequency domain. The post filters are your eyes... not an electronic circuit. In the audio world... if you dont post filter out the 44Khz sampling artifact before you send the signal on to your amp... then your tweeters burn up. And if you try to go down 80db in 1 octave then you screw up the music because of the nonlinearity of a filter that radical. oversampling moves the artifact up in frequency so the filter can be lower in order and thus more linear in the audio region and thus doing less damage to the signals you want while crushing the signal that you dont want (the sampling frequency spike).

 

[docferdie]

I strongly disagree that oversampling doesnt improve the sound quality. Just listen to an early player without oversampling and you will hear the difference. The first oversampling player with 20 bit dacs was the Denon DCD1520. I bought one when it came out and I can assure you that the oversampling and 'better' dacs it had made on a HUGE difference in sound quality. What enabled that improvement was the 8x oversampling, the extra 4 bits in the DACs and most importantly the lower order digital filter that the oversampling and DAC's allowed.

Would you be willing to test that in a blinded fashion?
I also refer you to the flagship denon receiver (5805) which can do DAC even with SACDs. If you look at the feature comparison on the website oversampling is notoriously absent. For $6000 surely they would throw in a feature such as this if it were really important. I agree that their CD players have the feature but wouldn't it make sense to include it, if it really mattered. Simply put 16 bit sound will still be 16 bit sound even if you have a 16, 24, or a 2,0000000000 bit DAC. If it doesn't then the DAC is failing at its job of decoding the bitstream as recorded. Like I said before if the data on the disk recorded 256 shades of red then on playback you should still see 256 shades of red. No more no less. Anything else is an alteration and not necessarily an improvement.
Although I digress the original question is whether 12 bit video DACs matter and I think I have presented enough scientific evidence to point out that it really doesn't matter if your player can do 2 to the power of (12 times 3) colors if your eye can only see in between 2 to the power of 23 and 24.

 

[Rob Babcock]

Would you be willing to test that in a blinded fashion?
I also refer you to the flagship denon receiver (5805) which can do DAC even with SACDs. If you look at the feature comparison on the website oversampling is notoriously absent. For $6000 surely they would throw in a feature such as this if it were really important.

You're confusing oversampling with upsampling- they're two totally unrelated concepts. Nearly all DACs oversample, as do virtually all standalone CD players regardless of price, although there's a fad right now for non-oversampling DACs. A couple of examples are the Chris Own dACK! and the Nixon TubeDAC.

Denon does include upsampling on all their receivers over $1k. Their trade name for it is ALPHA24. They may possibly throw in a dash of interpolation for good measure. I believe most of Onk's upper models will also do so.

At any rate, I doubt that inclusion or the lack thereof of a given feature by any specific equipment manufacturer is a reliable indicator of merit.

 

[docferdie]

we probably should have a separate thread to discuss oversampling and upsampling altogether-- I do agree that the two terms were being thrown around loosely. After all the original question was for video and not for audio.
But Rob just to clarify your point on Alpha processing, here is what I got from the 3805 manual

10.AL24 Processing for All Channels
DENON has further developed its proprietary AL24 Processing, an
analog waveform reproduction technology, to support the 192-kHz
sampling frequency of DVD-Audio,AL24 Processing Plus,
thoroughly suppresses quantization noise associated with D/A
conversion of LPCM signals to reproduce the low-level signals
with optimum clarity that will bring out all the delicate nuances of
the signals with optimum clarity that will bring out all the delicate
nuances of the music.
Equipped for not only the front left and right channels but also for
the surround left and right, center and subwoofer channels.

This doesn't sound like an upsampling technique at all. This sounds like a way of properly doing digital to analog conversion of source material to include that which is natively 192 Khz to begin with. Or are you saying that alpha 24 will convert any stream be it 44 khz or 96 khz to 192 Khz prior to amplification? It sure doesn't sound like it to me.

 

[Rob Babcock]

Their description on the website glosses over the technical points a bit. I've read a bit more detailed of their process in other mags. Wish I could find the links. At any rate, they do upsample as part of the process.

There'd be virtually no material at a native 192 frequency. Keep in mind Alpha24 was used well before there was any way to digitally transfer DVD-A, and that you can apply ALPHA24 to Redbook CD.

 

[mtrycrafts]

Their description on the website glosses over the technical points a bit. I've read a bit more detailed of their process in other mags. Wish I could find the links. At any rate, they do upsample as part of the process.

There'd be virtually no material at a native 192 frequency. Keep in mind Alpha24 was used well before there was any way to digitally transfer DVD-A, and that you can apply ALPHA24 to Redbook CD.

Do a google search for oversampling or up sampling. You will get a number of good explanations