or Connect
AVS › AVS Forum › Video Components › Video Download Services & Hardware › Netflix streaming quality
New Posts  All Forums:Forum Nav:

Netflix streaming quality - Page 76

post #2251 of 5446
Quote:
Originally Posted by jbdawson View Post

How accurate is this now, has it been changed? If ive got 6mbps would that clock at around the highest HD quality?

I am currently using basic internet I think 1.5 and it looks awful as expected

Those numbers have definitely changed for the PC and there is some evidence that at least the PS3 and Roku 2 are using those encodings as well. In this new set there are 9 encodings, 6 of them SD, two 720p and one is 1080p (in Kbps): 375, 560, 750, 1050, 1400, 1750, 2350, 3600 and (probably) 5100. The first 8 of those numbers can be seen in the PC Netflix player if you play something with an HD encoding, left-click the viewing area, type CTRL-SHIFT-ALT M and select "Stream Manager" from the little menu which pops up:


It probably still gives you the highest encoding that the available bandwidth on your connection is at least 140% of, so multiply those by 1.4 to estimate how much bandwidth you'll require to get that quality level. That's bandwidth on your connection to Netflix's servers, distinct from the nominal bandwidth of your network service.

So if your network service speed is 1500 Kbps, the highest bit rate encoding that you'll get is 1050; you'll only get that if your connection to Netflix is maintaining the 1500 Kbps. You'd need 1960 Kbps to get the 1400 Kbps encoding.
post #2252 of 5446
The technology should have improved a bit in the last 3 years. I've played a bit with Expression to see what the different bit rates would look like. Most of the time though at 6 mbps giving me and average of around 5.1 mbps when tested delivers a fairly crisp 720p picture. I have watched some HDX (1080p) on Vudu at that rate though. For me 720p is okay since my TV is an 11 year old Pioneer that has about 960 lines (due to overscan and need to get the tubes shimmed) but only about ~1000 pixels horizontally (according to the manual). I think I read years ago our eyes are more sensitive to vertical resolution than horizontal.
post #2253 of 5446
We were talking in the "Best possible Netflix streaming device" thread about this (in a long, essentially OT rathole that we probably should have moved to this thread) and it may be that the adaptive bit rate tech is buffer-level driven rather than by bandwidth. If the buffer is draining too fast, which would eventually require a pause to re-buffer, a switch is made to a lower bit rate encoding; if the buffer is rapidly filling, which implies that you could handle more, a switch is made to a higher bit rate encoding. It's probably not nearly as simple as that, but something like that would adapt to multiple different problems. Maybe you have plenty of bandwidth on your connection but you can't get the processing power to play the higher bit rate encoding without falling behind (which probably wouldn't happen in an embedded player, but might on an HTPC one)--a buffer-level based solution would work. So, you might not need 40% bandwidth headroom to get to a particular encoding, but you'll need some amount of bandwidth headroom to sustain playback of a given encoding.
post #2254 of 5446
Summarizing some of the results I recently posted in "Best possible Netflix streaming device" thread. I have an 'up to 7.1Mbps DSL' which consistently tests at around 6.2 Mbps (I mainly use the Vudu Speed test - but the speeds I can see on Real-Time Bandwidth on Tomato on my router seem to confirm 6.2 Mbps)

Top Gear (stereo) - on my BD 390 10 minutes average was around 4.9 Mbps - I switched to ROKU 2 XS it was more like 5.4 Mbps

10 minutes of 'The Girl Who Kicked the Hornets Nest' (5.1 audio) on the ROKU 2 XS was 5.3 Mbps average - see attached.

michaeltscott is reporting higher speed streams on his ROKU XS - which suggests I may not be getting the highest quality stream.

That said the 5.3 - 5.4 MBps 10 minute averages seem surprisingly close to the Netflix statement 'Our highest quality files are 4800 kbps (for 1080p HD video) and 384 kbps audio (for 5.1 audio)'

At this point I think I will stop measuring and enjoy watching - it all looks much better than watching streamed video a few years ago.....
LL
post #2255 of 5446
I'm having a strange issue with netflix streaming, and I'm trying to figure out whether it's my oppo bdp-93 or just netflix's crappy streaming service. Basically, what I'm experiencing on pretty much every netflix feed is consistent judder on any content that has motion (e.g. when the camera pans, the motion is jittery). I'm also seeing random frame drops every couple minutes in all my streaming content (it looks like the bottom half of the screen flashes really quickly, and other times the whole frame blinks as if the tv lost signal for a millisecond). I haven't seen these issue on any other content (dvd or blu-ray), so I assume it's just netflix. However, I can't reproduce the issues at the exact same points. That is, if I see the frame drop/flashing at 2:10 in a netflix stream, if I rewind and play it again, the flashing doesn't show up again at 2:10. This is what leads me to think this may not be a content encoding problem, and maybe a problem with my player or the netflix streaming service as a whole. Since I don't see these judder/frame drop issues on any other content, I'm leaning towards this just being a netflix streaming issue. I have no problem with bandwidth (10mb down connecteced via ethernet directly to my modem.

I don't have other streaming devices to test out whether it's the oppo or netflix, but it doesn't seem to be related to specific streaming titles (it affects all of them randomly). It's also not a 24p conversion issue. My tv is showing the netflix streams as coming out at 60hz as they should. Do you all think this is just sucky streaming or that maybe my player has an issue with streaming. I'm connected directly to my tv via hdmi.
post #2256 of 5446
All lot (maybe most) Netflix content is encoded at 24P. On my LG DB390 if I select 24P then Netflix streams are output at 24P if they are a 24P stream. I don't see jitter then on the 24P streams (all other streams are output at P60 and I do see jitter).

However if I use my ROKU 2 XS everything comes out at 60P and I see jitter.
post #2257 of 5446
Poking through some titles on the PC the other day it seemed that about a third of everything was encoded in 30p, a few titles at 25p and the rest at 24p. On the PC you can bring up an "AV Stats" overlay which tells you the framerate--once a title begins to play (w/o blowing it up fullscreen) left-click the picture and type CTRL-SHIFT-ALT-D.

The only time I see any significant judder on the Roku 2 is when viewing Starz Play titles, nearly all of which are 30p encodings of 24p sources.
post #2258 of 5446
Quote:
Originally Posted by michaeltscott View Post

Poking through some titles on the PC the other day it seemed that about a third of everything was encoded in 30p, a few titles at 25p and the rest at 24p. On the PC you can bring up an "AV Stats" overlay which tells you the framerate--once a title begins to play (w/o blowing it up fullscreen) left-click the picture and type CTRL-SHIFT-ALT-D.

The only time I see any significant judder on the Roku 2 is when viewing Starz Play titles, nearly all of which are 30p encodings of 24p sources.

Try 4 minute mark of The Good The Bad The Weird - or 15 minutes into the Fall

Edit : They may (or may not) be 30P encodes of 24P sources - but they shouldn't look like this.
post #2259 of 5446
Quote:
Originally Posted by undecided View Post

Try 4 minute mark of The Good The Bad The Weird - or 15 minutes into the Fall

Okay--I remember The Good, The Bad, The Weird had major judder problems. I don't believe I've watched The Fall.
post #2260 of 5446
(Continuing a discussion from the "Best possible Netflix streaming device" thread, where it was OT).

Quote:
Originally Posted by undecided View Post

I still feel if I average 5.2 - 5.4 MBps on a Netflix stream and Netlix says that is their highest encoding data rate I may be getting 1080P/5.1 - but agree it would be nice if if more devices had an indicator......

Let's forget everything about the Tomato graphs except for the total throughput number. On my graph of minutes 5-14 of Ong Bak 2 at 1080p 5.1 on the PS3 (posted here), the total amount of data received was 449.35 MiB (btw, I now realize that Tomato's report of average kbit/s is actually, correctly, 1000s of bits per second). msgohan says that when he pulls the network plug the PS3 continues playing for 129 seconds, so we'll assume that the amount of data consumed during that 600 seconds is actually 729 seconds worth of video and audio:

449.35 MiB x 2^20 =
471177625.6 bytes / 10^6 =
471.1776256 MBytes x 8 =
3769.4210048 Mbits / 729 seconds =
5.17 Mbps - .384 Mbps {audio} = 4.79 Mbps

That's in line with the 4.8 Mbps we're told to expect. For 720p 5.1, it was:

341.65 MiB x 2^20 =
358245990.4 bytes / 10^6 =
358.2459904 MBytes x 8 =
2865.9679232 Mbits / 729 seconds =
3.93 Mbps - .384 Mbps {audio} = 3.55 Mbps

msgohan expects that to be 3.6 Mbps. How much more perfectly can we expect this to work out?

Your graph for the same period reports 317.82 MiB consumed (333.26 MB or 2666.06 Mbits). This comes out to 4.443 Mbps (as reported by Tomato), flat, with no buffering. That's a little more than the 3.6+.384 Mbps we'd for 720p 5.1 but not nearly the 4.8+.384 Mbps we're expecting for 1080p, so I think that during that period you were getting 1080p some of the time but not continuously.

These are (highly) variable bit rate encodings. As I said before, that particular 10 minute period was chosen specifically because it was non-stop encoder-challenging video that I'd expect to come out at the highest possible bit rate. 10 minutes of fairly static scenes with little motion (a couple of people having a conversation over a meal in a private setting, for example) should produce a much lower average. During quieter periods of the film I think that you would get the 1080p encoding; it would just drop down to 720p during high action stuff.

Try this for me--go into the display settings on your PS3 and manually set resolution to eliminate 1080p and 1080i (i.e., limiting to 720p) then take the minutes 5-thru-14 Ong Bak 2 measurement. I'm betting that your curve will be very much like mine.
post #2261 of 5446
Quote:
Originally Posted by undecided View Post

Try 4 minute mark of The Good The Bad The Weird - or 15 minutes into the Fall

Edit : They may (or may not) be 30P encodes of 24P sources - but they shouldn't look like this.

Just checked The Good, The Bad, The Weird and I can confirm there is juddering 4 minutes into it. It happens even if I let it buffer beforehand so it gets up to HD quality.
post #2262 of 5446
Quote:
Originally Posted by theslug View Post

Just checked The Good, The Bad, The Weird and I can confirm there is juddering 4 minutes into it. It happens even if I let it buffer beforehand so it gets up to HD quality.

I haven't watched this for awhile, but remember pretty bad juddering about at that point (where the eagle is flying around). I didn't notice as much the rest of the movie...at least to the point where I turned it off (on PS3).
post #2263 of 5446
Quote:
Originally Posted by michaeltscott View Post

(Continuing a discussion from the "Best possible Netflix streaming device" thread, where it was OT).



Your graph for the same period reports 317.82 MiB consumed (333.26 MB or 2666.06 Mbits). This comes out to 4.443 Mbps (as reported by Tomato), flat, with no buffering. That's a little more than the 3.6+.384 Mbps we'd for 720p 5.1 but not nearly the 4.8+.384 Mbps we're expecting for 1080p, so I think that during that period you were getting 1080p some of the time but not continuously.

These are (highly) variable bit rate encodings. As I said before, that particular 10 minute period was chosen specifically because it was non-stop encoder-challenging video that I'd expect to come out at the highest possible bit rate. 10 minutes of fairly static scenes with little motion (a couple of people having a conversation over a meal in a private setting, for example) should produce a much lower average. During quieter periods of the film I think that you would get the 1080p encoding; it would just drop down to 720p during high action stuff.

Try this for me--go into the display settings on your PS3 and manually set resolution to eliminate 1080p and 1080i (i.e., limiting to 720p) then take the minutes 5-thru-14 Ong Bak 2 measurement. I'm betting that your curve will be very much like mine.

I agree but that Ong Bak 2 number seems to be off compared to my other 10 minute averages (I also thought we had agreed that Ong Bat 2 on the ROKU 2 wasn't getting the 5.1 audio). As I reported earlier when I ran Ong Bat 2 again Tomato reported an average 4.9 Mbps

If you look at my other minute averages shown in Tomato

Top Gear average 5.4 Mbps.

Cape the Pilot 373MB average 5.2 Mbps

The Girl Who Kicked the Hornets Nest 381MB average 5.3 Mbps

All would seem to point to the approx 5.2 Mbps we should expect......

As noted in the other thread I don't actually have a PS3 - I will try at my sons place next time I am there on his 6 Mbps DSL.

EDIT - One other thought. Maybe the demanding 5 - 14 minute section on Ong Bak 2 caused me to drop back to 720P as you propose on the first run - on the second I was lucky and held 1080P for longer. The less demanding Top Gear, The Girl..., The Cape etc may have all been able to hold 1080P
post #2264 of 5446
Quote:
Originally Posted by undecided View Post

EDIT - One other thought. Maybe the demanding 5 - 14 minute section on Ong Bak 2 caused me to drop back to 720P as you propose on the first run - on the second I was lucky and held 1080P for longer. The less demanding Top Gear, The Girl..., The Cape etc may have all been able to hold 1080P

That's the point that I've been trying to make. Even in Ong Bak 2 the next 10 minutes after my normal test passage drops considerably (3 minutes of conversation in a dark room followed by one-on-one martial arts training, no rain):


For that segment at 1080p I get:

((326.28 MiB x 2^20) / 10^6) x 8) / 729 = 3.75 Mbps {- .384 = 3.37 Mbps}

That's 1.43 Mbps under the maximum and should be easily sustainable in your bandwidth.

Except in the most difficult to encode passages (typically high action sequences), I think that you'll probably see the 1080p encoding. Most non-action films will have little or none of that.
post #2265 of 5446
Quote:
Originally Posted by michaeltscott View Post

That's the point that I've been trying to make. Even in Ong Bak 2 the next 10 minutes after my normal test passage drops considerably (3 minutes of conversation in a dark room followed by one-on-one martial arts training, no rain):


For that segment at 1080p I get:

((326.28 MiB x 2^20) / 10^6) x 8) / 729 = 3.75 Mbps {- .384 = 3.37 Mbps}

That's 1.43 Mbps under the maximum and should be easily sustainable in your bandwidth.

Except in the most difficult to encode passages (typically high action sequences), I think that you'll probably see the 1080p encoding. Most non-action films will have little or none of that.

So my sense is with adaptive streaming the 40 % headroom no longer applies.

Ong Bak 2 from tonight 5 - 14 minutes attached on my 6.1 Mbps DSL (it is advertised as 3.1 - 7.1 and measures in my location as 6.1) and measures on Tomaro an average of 5.8 mbps

I agree it drops to 5.2 Mbps later - but from around 22 - 31 minutes I saw an over 6 mbps average on Tomato.

All with only stereo (I still can't get 5.1 on this title)
LL
post #2266 of 5446
I'm not sure I understand what you're saying. Are you saying that headroom isn't required because the encoding is a constant rather than a variable bitrate? It seems that headroom above the average bitrate would always be required if avariable bitrate encoding is used.
post #2267 of 5446
Quote:
Originally Posted by undecided View Post

So my sense is with adaptive streaming the 40 % headroom no longer applies.

Yes and no. You still need bandwidth that's higher than the current average bit rate of the stream to maintain that encoding but I think that if that bit rate stays lower than the "maximum average" for extended periods of time then the amount of headroom needed to stay with that encoding will be less. I think that if the player is easily keeping its buffer filled to the top it will try the next higher encoding even if it's available bandwidth is not enough to sustain the "maximum average" for that encoding; until the encoding reaches that "maximum average" and stays there for a while, the player will be able to stay with that encoding.

In the 5-14 graph that you posted I'd say that you weren't getting enough bandwidth to sustain 1080p. (I'd really like to see a graph from you for 15-24--I'd expect the curve to look a lot like mine for that period, with no long plateaus). If you got 6.1 Mbps average for 21-31, it was probably a high-action period for which your bandwidth stayed high enough to stick at 1080p.
post #2268 of 5446
I'd love to see them implement a backup buffer which doesn't appear to be there at the moment so their app can work a bit like a DVR where I might want to back up about 15 seconds you can do so. It appears now when you back up it needs to go fetch that part of the stream. It's for those "what did they just say" moments.
post #2269 of 5446
Quote:
Originally Posted by dfiler View Post

I'm not sure I understand what you're saying. Are you saying that headroom isn't required because the encoding is a constant rather than a variable bitrate? It seems that headroom above the average bitrate would always be required if avariable bitrate encoding is used.

I am sure some headroom is required - but maybe not the 40% from before.

When I see a 5.8Mbps average over 10 minutes on my DSL which maxes out at 6.1 Mbps it would seem to me I am probably getting the hi-res stream for most of this time. After all I pulled 420MBs of something during that 10 minutes - and that seems too much data for the 720P stereo stream.
post #2270 of 5446
One of the Korean films "The Pot" is being streamed minus the audio and with the wrong aspect ratio. I clicked on the "Report Problem" link but also noticed a number of people in the review section were complaining about no audio (apparently didn't realize people and things were "stretched" a bit). Apparently the OAR is 1:85:1 not 2:35:1. I run into this occasionally with some films and check back weeks later only to find they've never fixed it.

Are the encoding jobs going to high schoolers as an after school job these days?
post #2271 of 5446
Geez. Well, at least you can read the dialogue even though you can't hear anything. I left it playing for a few minutes and then reported the problem too. Dunno if it will help.
post #2272 of 5446
Speaking of headroom, this Roku Playback debugging message...

"Cannot play 1.5 mbps stream on your 1.8 mbps network because starting at 1:07:13, the stream needs 4.2 MB in the next 16 seconds (2.2 mbps)."

post #2273 of 5446
Quote:
Originally Posted by msgohan View Post

Speaking of headroom, this Roku Playback debugging message...

"Cannot play 1.5 mbps stream on your 1.8 mbps network because starting at 1:07:13, the stream needs 4.2 MB in the next 16 seconds (2.2 mbps)."


Huh. Never seen that before. I wonder how it knows the bandwidth required for 16 seconds at 1:07:13? To me, this would indicate that the Roku is not using adaptive bit rate tech.
post #2274 of 5446
Quote:
Originally Posted by michaeltscott View Post

Huh. Never seen that before. I wonder how it knows the bandwidth required for 16 seconds at 1:07:13? To me, this would indicate that the Roku is not using adaptive bit rate tech.

This type of debugging is not available for the Netflix client on the Roku 2; it's been disabled. It's only for earlier boxes like the XDS (this debugging does work for other apps like Amazon on the Roku 2). When the Roku 2 was announced, it was implied that only that the Roku 2 could handle adaptive streaming; it was also implied that this was the reason why the new Netflix channel could not be back ported to the earlier Rokus.

Msoghan, what Roku box did you pull this message from? Unless they made a change in the new firmware released yesterday.
post #2275 of 5446
Sorry, Roku 1 XD. It doesn't use adaptive (it pops up a loading screen when it runs out of buffer and has to switch to a lower stream). It's a shame they aren't able to provide that info on the Roku 2.
post #2276 of 5446
My PS3 shows low image quality on a Netflix HD title (Weeds), especially annoying in low light scenes. However, when I connect a VAIO laptop to the same ethernet cable and output to the same receiver/plasma tv through the same HDMI cable, the quality improves a lot. I got Charter 18Mbps/4Mbps service that gives constant speed readings above 7 Mbps on the PS3 test (and 10-16Mbps on any of the 6 computers in the house, wired and wireless). Netflix HD is playing crisp and gorgeous on any of them.

Now I use the PS3 exclusively for Netflix streaming to the plasma tv and I would like to use the laptop for other things too, but image quality is most important for me so for now I keep the laptop on top of the PS3 for netflix streaming purposes only.

Can someone please help me attain the same quality on the dang PS3 heavy/hot expensive box and free my laptop?

Many thanks!
post #2277 of 5446
What does the PS3 say the quality is (Low-HD, Medium-HD, High-HD or X-High HD)?

It doesn't make much sense to me. I don't think the PC does 1080p yet, does it? It seems to me the PS3 should be giving you the best quality. Might want to check some video settings on your PS3.
post #2278 of 5446
Quote:
Originally Posted by mproper View Post

What does the PS3 say the quality is (Low-HD, Medium-HD, High-HD or X-High HD)?

There's no Low/HD, just Medium/HD (2.35 Mbps), High/HD (3.6 Mbps) and X-High/HD (4.8 Mbps)--add .192 Mbps for stereo (.064 on the PC) and .384 for DD+. Medium and High are two flavors of 720p and X-High is 1080p. Just picking nits .

BTW, gcornilescu, to see those encoding values hit SEL on a pad or DISPLAY on the PS3 remote; a little info display will be overlayed in the upper left corner.
Quote:


It doesn't make much sense to me. I don't think the PC does 1080p yet, does it? It seems to me the PS3 should be giving you the best quality. Might want to check some video settings on your PS3.

No, it doesn't make any sense (and no, the PC doesn't do 1080p as yet and might never given IP holder paranoia ). How is the display connected to the PS3 and what are the display settings?
post #2279 of 5446
When you play a title on the ps3, make sure you give it time to work its way up to HD levels. It takes about 15 seconds or so for me to get to X-High/HD. You'll gradually see the picture improve.
post #2280 of 5446
Quote:
Originally Posted by gcornilescu View Post

My PS3 shows low image quality on a Netflix HD title (Weeds), especially annoying in low light scenes. However, when I connect a VAIO laptop to the same ethernet cable and output to the same receiver/plasma tv through the same HDMI cable, the quality improves a lot. I got Charter 18Mbps/4Mbps service that gives constant speed readings above 7 Mbps on the PS3 test (and 10-16Mbps on any of the 6 computers in the house, wired and wireless). Netflix HD is playing crisp and gorgeous on any of them.

Now I use the PS3 exclusively for Netflix streaming to the plasma tv and I would like to use the laptop for other things too, but image quality is most important for me so for now I keep the laptop on top of the PS3 for netflix streaming purposes only.

Can someone please help me attain the same quality on the dang PS3 heavy/hot expensive box and free my laptop?

Many thanks!

The PS3 definitely has problems with the Netflix application. It was pretty much flawless until a little less than a year ago but now I can rarely even get any HD streams, it's mostly low/sd and med/sd which are unwatchable. On rare occasions it will ramp all the way to x-high/hd but only very rarely. I have a Roku box twelve inches from the PS3 that plays the same titles in HD without fail. I have also streamed the same movies on Vudu through the same PS3 and they stream in HDX (1080p) without a hitch. I've spoken to PS3 customer service twice and Netflix customer service three times but nothing has improved the situation. I don't even get better results when I run it with a wired connection. I don't even bother trying Netflix on the PS3 anymore.
New Posts  All Forums:Forum Nav:
  Return Home
AVS › AVS Forum › Video Components › Video Download Services & Hardware › Netflix streaming quality