AVS Forum banner

281 - 300 of 1855 Posts

·
Registered
Joined
·
1,708 Posts
Apple TV fans rejoice! Apparently, according to a thread here on AVS, the ATV2 has just added DD support for NF titles that offer it, via firmware update.


So, for those of you who were on the fence on picking one up...
 

·
Registered
Joined
·
1,194 Posts
Discussion Starter · #282 ·

Quote:
Originally Posted by [Irishman] /forum/post/20127740


Apple TV fans rejoice! Apparently, according to a thread here on AVS, the ATV2 has just added DD support for NF titles that offer it, via firmware update.


So, for those of you who were on the fence on picking one up...

I know, that is great news! Now if they could just make the picture quality look as good as the PS3 at 720p... When they finally do my PS3 goes on ebay ;-)
 

·
Registered
Joined
·
1,708 Posts
Quote:
Originally Posted by rpauls
I know, that is great news! Now if they could just make the picture quality look as good as the PS3 at 720p... When they finally do my PS3 goes on ebay ;-)
I gotta think it's the A4 chip's fault.
 

·
Registered
Joined
·
1,194 Posts
Discussion Starter · #284 ·
Quote:
Originally Posted by [Irishman]
I gotta think it's the A4 chip's fault.
Possibly, but my guess is it's just that the PS3 is receiving a different stream than the ATV (even when both are at 720p), and the one the PS3 gets just plain looks better. Either because it is a higher bitrate or it has a better compression algorithm employed.


It would be interesting to do a bitrate test of these two devices.
 

·
Registered
Joined
·
750 Posts
Quote:
Originally Posted by [Irishman]
I gotta think it's the A4 chip's fault.
This is my theory as well.


With xbmc installed on an ATV2, it is easy to see that post processing results in dropped frames on higher bitrate video. Post processing is what I think gives the PS3 better picture quality. It is able to eliminate blocking, banding, and other artifacts present in aggressively compressed video.


In a rather simplified sense, more processing power equates to better post processing.
 

·
Registered
Joined
·
20,517 Posts
Okay--my DMP-BDT110 came yesterday and I've played around with it some. It seems to run the same Netflix player as the PS3 (though it performs a bit differently) and has access to the 5.1 soundtracks and closed captions available on some titles. AFAICT, the presentation is pixel-for-pixel identical to the PS3's--it's an HTML5 app, much of which is dynamically downloaded at startup (which is why there are still multiple variants of it running on the PS3).


I ran a handful of experiments and the player apparently does not have access to 1080p encodings (or, as some would have it, the higher-bit-rate-720p encodings
). It does seem to use the same "DASH"-like tech the PS3 player does, though it doesn't perform quite as well. (DASH, or "Dynamic Adaptive Streaming over HTTP", is an emerging ISO standard for adaptive bit rate video streaming; Netflix and several major tech concerns are supporting the effort--see this ). The PS3's player seems to use the adaptive bit rate streaming tech to start streams very nearly instantly (typically 3 seconds or less after pushing PLAY); the BDT110's stream start time is very average, usually 15 seconds or more. The PS3 switches between encodings of different bit rates seamlessly--you can see that it's happening, but it's so smooth it's a bit like watching the focus of a lens sharpen and soften. The BDT110's encoding switches often involve a skipped frame or two, the entire screen flashing black for a fraction of a second. During my playing around with it, conditions on my connection were never so turbulent to make it happen more than a couple of times during stream start--if I saw a lot of that, I'd probably stop watching. The algorithm might be sensitive to that possibility and have provisions to deal with it.


Here are some graphs of bandwidth consumption during Netflix play, 10 minutes of a long high-action sequence at the beginning of The Good, the Bad, the Weird starting about 3 minutes in (I chose it because its PQ has received a lot of praise in these forums). 5.1 audio was explicitly selected (strangely it was the default on the PS3 but not the BDT110). My method was to push PLAY, start a stopwatch when the video appeared and snap a picture of the bandwidth usage graph at 13 minutes. I'd run this before on the PS3, but I did it again as a sanity check--the results turn out to be highly repeatable. If you blow this page up to fill the screen, you'll see the graphs from a single machine displayed side-by-side. The scales are not the same--there doesn't seem to be any way to control that in my router's monitor; pay attention to the TX numbers (in these, the router's reporting bandwidth of data transmitted to the attached devices):

Warning: Spoiler! (Click to show) Spoiler  
Warning: Spoiler! (Click to show)






The Roku's buffering algorithm is obviously different; it seems to be loading up and playing until some low-water mark, then loading again, lather, rinse, repeat
:

Warning: Spoiler! (Click to show) Spoiler  
Warning: Spoiler! (Click to show)




Here's some graphs of the HD version of The Dark Knight as streamed from Amazon on the Panny and the Roku. No 5.1 on this one, but some Amazon streams do have it. Pretty damn decent PQ for the bit rate, though I don't know that I'd feel like I got my money's worth for a $5 new release rental. Again, the Roku's buffering pattern is distinctive:

Warning: Spoiler! (Click to show) Spoiler  
Warning: Spoiler! (Click to show)




Note that all of these streaming players start by downloading a large bolus of data--with my Network service it was a 25+ Mbps spike for a couple seconds.


I'm probably going to keep the Panny and return the Roku--I only bought either device to get access to Amazon streaming (a weak excuse for buying a new toy
) and I certainly don't need both. The Roku has more apps (mostly junk), but the Panny does everything I need it to do and has VUDU (VUDU surprisingly gave me another $6 credit for registering a new device on my existing account--sweet
). The Panny has a bunch of other capabilities that I haven't explored yet, like an ability to play content from DLNA servers (not to mention 3D BD playback, though I'm not interested in that). I also got a coupon with the Panny good for a free copy of the 3D Avatar BD.
 

·
Registered
Joined
·
1,194 Posts
Discussion Starter · #288 ·

Quote:
Originally Posted by michaeltscott /forum/post/20137112


Okay--my DMP-BDT110 came yesterday and I've played around with it some. It seems to run the same Netflix player as the PS3 (though it performs a bit differently) and has access to the 5.1 soundtracks and closed captions available on some titles. AFAICT, the presentation is pixel-for-pixel identical to the PS3's--it's an HTML5 app, much of which is dynamically downloaded at startup (which is why there are still multiple variants of it running on the PS3).


I ran a handful of experiments and the player apparently does not have access to 1080p encodings (or, as some would have it, the higher-bit-rate-720p encodings
). It does seem to use the same "DASH"-like tech the PS3 player does, though it doesn't perform quite as well. (DASH, or "Dynamic Adaptive Streaming over HTTP", is an emerging ISO standard for adaptive bit rate video streaming; Netflix and several major tech concerns are supporting the effort--see this ). The PS3's player seems to use the adaptive bit rate streaming tech to start streams very nearly instantly (typically 3 seconds or less after pushing PLAY); the BDT110's stream start time is very average, usually 15 seconds or more. The PS3 switches between encodings of different bit rates seamlessly--you can see that it's happening, but it's so smooth it's a bit like watching the focus of a lens sharpen and soften. The BDT110's encoding switches often involve a skipped frame or two, the entire screen flashing black for a fraction of a second. During my playing around with it, conditions on my connection were never so turbulent to make it happen more than a couple of times during stream start--if I saw a lot of that, I'd probably stop watching. The algorithm might be sensitive to that possibility and have provisions to deal with it.


Here are some graphs of bandwidth consumption during Netflix play, 10 minutes of a long high-action sequence at the beginning of

Wow! That's impressive reporting. Thanks for taking the time to do this.


It appears the panasonic and the PS3 both run about the same average bitrate for 720p, so I'm guessing the PQ is pretty much equal in this case? But what jumps out, is that although the avg bitrates are the same, the instantaneous bitrates are drastically different. The panasonic has a much more bursty stream, and its maximum bitrate is consequentially much higher than the PS3. I would guess this indicates different buffering schemes are being used, with the PS3 taking more of a "constant bitrate" approach. Most likely this means a little more latency on this stream, but of course that has no impact on home viewing.


The Roku is even more bursty. How very interesting. I also bought a Roku recently to take advantage of that Amazon prime offering.


It is interesting to note that all three streamers show about the same average bitrate for 720p, but in my experience, the PS3 definitely looked much better than the Roku. Your graphs would indicate that whatever the PQ difference was, it was not due to raw bitrate. Two reasons I can think of that may have made the PS3 look better would be (A) it could be using a more effective compression algorithm, or (B) it is doing a lot of post processing as dfiler said.
 

·
Registered
Joined
·
1,194 Posts
Discussion Starter · #289 ·
Question: Your data show an average bitrate of 4.36 mbps for the Roku HD of the first movie. I think the highest I recall seeing on the Roku debug window on my TV was either 3.5 or 3.8. Do you have the debugging option activated on you Roku yet? What bitrate does your Roku report when it starts playing that movie?
 

·
Registered
Joined
·
20,517 Posts

Quote:
Originally Posted by DaveFi /forum/post/20138023


Is the DMP-BDT110 bitstreaming DD+ over Netflix or just DD like the PS3?

Unfortunately I have a 4 y/o AVR without DD+ decoding; the Panny is extracting the AC-3 core and bitstreaming that to my AVR (it's capable of processing the lossless formats and rendering them as uncompressed 7.1 PCM over HDMI, like the PS3). Others have reported getting DD+ out of it though, in the DMP-BDT110/210/3xx thread.


It's always perplexed me that the 720p stream (High/HD on the PS3) looks to be 4.4 or 4.5 Mbps when it was advertised as being 3.8 Mbps, but now I think that it could be the difference between the bit rate of the stereo sound component and that of DD+. That would make it 700 Kbps richer than whatever the old rate was. But that wouldn't seem to jibe with the Roku's average bit rate being nearly the same as the Panny's. I'm taking some stereo data on the PS3 as I type this and will follow it up with stereo on the Panny.
 

·
Registered
Joined
·
20,517 Posts

Quote:
Originally Posted by rpauls /forum/post/20137830


Question: Your data show an average bitrate of 4.36 mbps for the Roku HD of the first movie. I think the highest I recall seeing on the Roku debug window on my TV was either 3.5 or 3.8. Do you have the debugging option activated on you Roku yet? What bitrate does your Roku report when it starts playing that movie?

I didn't know about the debug mode of the Roku's Netflix player--I'll try it and get back to you.
 

·
Registered
Joined
·
1,194 Posts
Discussion Starter · #293 ·

Quote:
Originally Posted by michaeltscott /forum/post/20138130


I didn't know about the debug mode of the Roku's Netflix player--I'll try it and get back to you.

To enter Roku debug mode use the following sequence on the remote:


home home home home home rew rew rew FF FF
 

·
Registered
Joined
·
13,095 Posts

Quote:
Originally Posted by DaveFi /forum/post/20138023


Is the DMP-BDT110 bitstreaming DD+ over Netflix or just DD like the PS3?

My receiver reports DD+ (sounds great).
 

·
Registered
Joined
·
20,517 Posts

Quote:
Originally Posted by rpauls /forum/post/20138644


To enter Roku debug mode use the following sequence on the remote:


home home home home home rew rew rew FF FF

Yeah--it was pretty easy to find that out once you mentioned that the mode exists
. I tried it last night and it said that the stream was 3.8 Mpbs. I tried to get some data last night but I couldn't reliably get 5 Mbps out of my supposedly 25 Mbps service (I have got to get around to having a chat with Cox
). I'm doing it now, at 7 AM, when there should be much less contention for service from all concerned parties.
 

·
Registered
Joined
·
570 Posts

Quote:
Originally Posted by michaeltscott /forum/post/20138124


Unfortunately I have a 4 y/o AVR without DD+ decoding; the Panny is extracting the AC-3 core and bitstreaming that to my AVR (it's capable of processing the lossless formats and rendering them as uncompressed 7.1 PCM over HDMI, like the PS3). Others have reported getting DD+ out of it though, in the DMP-BDT110/210/3xx thread.


It's always perplexed me that the 720p stream (High/HD on the PS3) looks to be 4.4 or 4.5 Mbps when it was advertised as being 3.8 Mbps, but now I think that it could be the difference between the bit rate of the stereo sound component and that of DD+. That would make it 700 Kbps richer than whatever the old rate was. But that wouldn't seem to jibe with the Roku's average bit rate being nearly the same as the Panny's. I'm taking some stereo data on the PS3 as I type this and will follow it up with stereo on the Panny.

Nice Work with those graphs. I wonder if the difference between your Panny and Roku is a larger buffer in the Roku? I ask that because my Samsung 6500 really doesn't do all that well compared to Roku and my Apple TV--and I know the ATV has about 8GB to work with so I would bet the graph would be similar to the Roku. I tried an HDX movie on the Samsung but it got downgraded to HD--would be nice to see how it would work on Roku if it were available. So for me, I use the Roku and ATV the most since they never seem to have any picture freeze issues and PQ is quite good. Very interesting for sure.
 

·
Registered
Joined
·
20,517 Posts
I took some more data this morning. To start, I tried the PS3 and Panny in 2.0 sound to see if that made much difference. Suprisingly, it didn't account for much (I was expecting it to add 400-700 Kbps to the average):

Warning: Spoiler! (Click to show) Spoiler  
Warning: Spoiler! (Click to show)





Then I took some data from the other platforms that I have (all without 5.1 or 1080p capability, of course):

Warning: Spoiler! (Click to show) Spoiler  
Warning: Spoiler! (Click to show)





It's also interesting to see the different way that each of these players start:

Warning: Spoiler! (Click to show) Spoiler  
Warning: Spoiler! (Click to show)
[








I thought that the Xbox player was particularly interesting--it sucked down data at a furious pace for about 1 minute, then played from the buffer exclusively for over 2 minutes; interesting tactic. It would be interesting to see how it (and all of the rest of these things) deal with different amounts of available bandwidth at startup.


One problem with interpreting these averages is that we don't know how much video time they represent. The data streamed in 10 minutes doesn't just represent 10 minutes of video, but 10 minutes plus some amount in the buffer, of unknown playback duration. Different buffer sizes and buffering schemes in use by these players prevents us from drawing any firm conclusion about average bit rate.


One thing that I notice about these is that, except for the very beginning (when the Xbox is playing out of that mammoth buffer fill) the curves for TiVo and the Xbox are very nearly identical. The curve for the PC is the least like any of the others (except for the Roku); its average comes closest to the nominal 3.8 Mbps--I think that it may use the smallest buffer of them all.


Unless someone else can think of something else interesting for me to try, I think that will do it for now. I've seen the beginning of The Good, the Bad, the Weird so many times that I believe that I could recite all of the lines, in Mandarin and Korean
 

·
Registered
Joined
·
1,194 Posts
Discussion Starter · #298 ·

Quote:
Originally Posted by michaeltscott /forum/post/20138824


Yeah--it was pretty easy to find that out once you mentioned that the mode exists
. I tried it last night and it said that the stream was 3.8 Mpbs. I tried to get some data last night but I couldn't reliably get 5 Mbps out of my supposedly 25 Mbps service (I have got to get around to having a chat with Cox
). I'm doing it now, at 7 AM, when there should be much less contention for service from all concerned parties.

So the Roku says 3.8 mbps and your graphs say 4.4 mbps. I wonder if the difference can be attributed to header info on the packets? The Roku may be measuring the actual data payload (just the movie) and the router reporting total raw throughput? Just a guess.
 

·
Registered
Joined
·
20,517 Posts

Quote:
Originally Posted by rpauls /forum/post/20141695


So the Roku says 3.8 mbps and your graphs say 4.4 mbps. I wonder if the difference can be attributed to header info on the packets? The Roku may be measuring the actual data payload (just the movie) and the router reporting total raw throughput? Just a guess.

As I said above, I think that the data sent to a streaming player in the 10 minute shown in the graphs does not represent 10 minutes of video. The 10 minutes worth of video played while it was receiving that data is part of it, but at the end of that 10 minutes there's still unplayed video data in the player's buffer. If you stopped sending the stream after that 10 minutes and let the unplayed buffer play to its end, the total duration of the video would be more than 10 minutes; how much more depends upon the size of the player's buffer, the buffering algorithm it uses and the amount of excess bandwidth available.


If you look at the first Roku Netflix graph that I posted (the blue and red one), you'll see the player was sent 319.41 MB of video data in that 10 minutes:

319.41 MB x 8 bits-per-byte = 2555.28 Mbits / 3.8 Mbps = 672.44 seconds
I think that at the end of that 10 minute window, the Roku had approximately 72.44 seconds worth of video left to play in its buffer.
 
281 - 300 of 1855 Posts
Top